WIP - Started working towards MultiDiffusion batching.

docs/contributing/MODEL_MANAGER.md

@@ -1328,7 +1328,7 @@ from invokeai.app.services.model_load import ModelLoadService, ModelLoaderRegist
 
 config = InvokeAIAppConfig.get_config()
 ram_cache = ModelCache(
- max_cache_size=config.ram_cache_size, logger=logger
+ max_cache_size=config.ram_cache_size, max_vram_cache_size=config.vram_cache_size, logger=logger
 )
 convert_cache = ModelConvertCache(
  cache_path=config.models_convert_cache_path, max_size=config.convert_cache_size

invokeai/app/invocations/compel.py

@@ -103,7 +103,6 @@ class CompelInvocation(BaseInvocation):
                 textual_inversion_manager=ti_manager,
                 dtype_for_device_getter=TorchDevice.choose_torch_dtype,
                 truncate_long_prompts=False,
-                device=TorchDevice.choose_torch_device(),
             )
 
             conjunction = Compel.parse_prompt_string(self.prompt)
@@ -118,7 +117,6 @@ class CompelInvocation(BaseInvocation):
         conditioning_data = ConditioningFieldData(conditionings=[BasicConditioningInfo(embeds=c)])
 
         conditioning_name = context.conditioning.save(conditioning_data)
-
         return ConditioningOutput(
             conditioning=ConditioningField(
                 conditioning_name=conditioning_name,
@@ -205,7 +203,6 @@ class SDXLPromptInvocationBase:
                 truncate_long_prompts=False,  # TODO:
                 returned_embeddings_type=ReturnedEmbeddingsType.PENULTIMATE_HIDDEN_STATES_NON_NORMALIZED,  # TODO: clip skip
                 requires_pooled=get_pooled,
-                device=TorchDevice.choose_torch_device(),
             )
 
             conjunction = Compel.parse_prompt_string(prompt)
@@ -316,6 +313,7 @@ class SDXLCompelPromptInvocation(BaseInvocation, SDXLPromptInvocationBase):
                 )
             ]
         )
+
         conditioning_name = context.conditioning.save(conditioning_data)
 
         return ConditioningOutput(

invokeai/app/invocations/denoise_latents.py

@@ -1,5 +1,4 @@
 # Copyright (c) 2023 Kyle Schouviller (https://github.com/kyle0654)
-import copy
 import inspect
 from contextlib import ExitStack
 from typing import Any, Dict, Iterator, List, Optional, Tuple, Union
@@ -56,6 +55,7 @@ from invokeai.backend.stable_diffusion.diffusion.conditioning_data import (
 )
 from invokeai.backend.stable_diffusion.schedulers import SCHEDULER_MAP
 from invokeai.backend.util.devices import TorchDevice
+from invokeai.backend.util.hotfixes import ControlNetModel
 from invokeai.backend.util.mask import to_standard_float_mask
 from invokeai.backend.util.silence_warnings import SilenceWarnings
 
@@ -66,6 +66,9 @@ def get_scheduler(
     scheduler_name: str,
     seed: int,
 ) -> Scheduler:
+    """Load a scheduler and apply some scheduler-specific overrides."""
+    # TODO(ryand): Silently falling back to ddim seems like a bad idea. Look into why this was added and remove if
+    # possible.
     scheduler_class, scheduler_extra_config = SCHEDULER_MAP.get(scheduler_name, SCHEDULER_MAP["ddim"])
     orig_scheduler_info = context.models.load(scheduler_info)
     with orig_scheduler_info as orig_scheduler:
@@ -183,8 +186,8 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 raise ValueError("cfg_scale must be greater than 1")
         return v
 
+    @staticmethod
     def _get_text_embeddings_and_masks(
-        self,
         cond_list: list[ConditioningField],
         context: InvocationContext,
         device: torch.device,
@@ -194,8 +197,9 @@ class DenoiseLatentsInvocation(BaseInvocation):
         text_embeddings: Union[list[BasicConditioningInfo], list[SDXLConditioningInfo]] = []
         text_embeddings_masks: list[Optional[torch.Tensor]] = []
         for cond in cond_list:
-            cond_data = copy.deepcopy(context.conditioning.load(cond.conditioning_name))
+            cond_data = context.conditioning.load(cond.conditioning_name)
             text_embeddings.append(cond_data.conditionings[0].to(device=device, dtype=dtype))
+
             mask = cond.mask
             if mask is not None:
                 mask = context.tensors.load(mask.tensor_name)
@@ -203,8 +207,9 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
         return text_embeddings, text_embeddings_masks
 
+    @staticmethod
     def _preprocess_regional_prompt_mask(
-        self, mask: Optional[torch.Tensor], target_height: int, target_width: int, dtype: torch.dtype
+        mask: Optional[torch.Tensor], target_height: int, target_width: int, dtype: torch.dtype
     ) -> torch.Tensor:
         """Preprocess a regional prompt mask to match the target height and width.
         If mask is None, returns a mask of all ones with the target height and width.
@@ -226,11 +231,10 @@ class DenoiseLatentsInvocation(BaseInvocation):
         # Add a batch dimension to the mask, because torchvision expects shape (batch, channels, h, w).
         mask = mask.unsqueeze(0)  # Shape: (1, h, w) -> (1, 1, h, w)
         resized_mask = tf(mask)
-        assert isinstance(resized_mask, torch.Tensor)
         return resized_mask
 
+    @staticmethod
     def _concat_regional_text_embeddings(
-        self,
         text_conditionings: Union[list[BasicConditioningInfo], list[SDXLConditioningInfo]],
         masks: Optional[list[Optional[torch.Tensor]]],
         latent_height: int,
@@ -280,7 +284,9 @@ class DenoiseLatentsInvocation(BaseInvocation):
                     )
                 )
                 processed_masks.append(
-                    self._preprocess_regional_prompt_mask(mask, latent_height, latent_width, dtype=dtype)
+                    DenoiseLatentsInvocation._preprocess_regional_prompt_mask(
+                        mask, latent_height, latent_width, dtype=dtype
+                    )
                 )
 
             cur_text_embedding_len += text_embedding_info.embeds.shape[1]
@@ -302,36 +308,41 @@ class DenoiseLatentsInvocation(BaseInvocation):
             )
         return BasicConditioningInfo(embeds=text_embedding), regions
 
+    @staticmethod
     def get_conditioning_data(
-        self,
         context: InvocationContext,
+        positive_conditioning_field: Union[ConditioningField, list[ConditioningField]],
+        negative_conditioning_field: Union[ConditioningField, list[ConditioningField]],
         unet: UNet2DConditionModel,
         latent_height: int,
         latent_width: int,
+        cfg_scale: float | list[float],
+        steps: int,
+        cfg_rescale_multiplier: float,
     ) -> TextConditioningData:
-        # Normalize self.positive_conditioning and self.negative_conditioning to lists.
-        cond_list = self.positive_conditioning
+        # Normalize positive_conditioning_field and negative_conditioning_field to lists.
+        cond_list = positive_conditioning_field
         if not isinstance(cond_list, list):
             cond_list = [cond_list]
-        uncond_list = self.negative_conditioning
+        uncond_list = negative_conditioning_field
         if not isinstance(uncond_list, list):
             uncond_list = [uncond_list]
 
-        cond_text_embeddings, cond_text_embedding_masks = self._get_text_embeddings_and_masks(
+        cond_text_embeddings, cond_text_embedding_masks = DenoiseLatentsInvocation._get_text_embeddings_and_masks(
             cond_list, context, unet.device, unet.dtype
         )
-        uncond_text_embeddings, uncond_text_embedding_masks = self._get_text_embeddings_and_masks(
+        uncond_text_embeddings, uncond_text_embedding_masks = DenoiseLatentsInvocation._get_text_embeddings_and_masks(
             uncond_list, context, unet.device, unet.dtype
         )
 
-        cond_text_embedding, cond_regions = self._concat_regional_text_embeddings(
+        cond_text_embedding, cond_regions = DenoiseLatentsInvocation._concat_regional_text_embeddings(
             text_conditionings=cond_text_embeddings,
             masks=cond_text_embedding_masks,
             latent_height=latent_height,
             latent_width=latent_width,
             dtype=unet.dtype,
         )
-        uncond_text_embedding, uncond_regions = self._concat_regional_text_embeddings(
+        uncond_text_embedding, uncond_regions = DenoiseLatentsInvocation._concat_regional_text_embeddings(
             text_conditionings=uncond_text_embeddings,
             masks=uncond_text_embedding_masks,
             latent_height=latent_height,
@@ -339,23 +350,21 @@ class DenoiseLatentsInvocation(BaseInvocation):
             dtype=unet.dtype,
         )
 
-        if isinstance(self.cfg_scale, list):
-            assert (
-                len(self.cfg_scale) == self.steps
-            ), "cfg_scale (list) must have the same length as the number of steps"
+        if isinstance(cfg_scale, list):
+            assert len(cfg_scale) == steps, "cfg_scale (list) must have the same length as the number of steps"
 
         conditioning_data = TextConditioningData(
             uncond_text=uncond_text_embedding,
             cond_text=cond_text_embedding,
             uncond_regions=uncond_regions,
             cond_regions=cond_regions,
-            guidance_scale=self.cfg_scale,
-            guidance_rescale_multiplier=self.cfg_rescale_multiplier,
+            guidance_scale=cfg_scale,
+            guidance_rescale_multiplier=cfg_rescale_multiplier,
         )
         return conditioning_data
 
+    @staticmethod
     def create_pipeline(
-        self,
         unet: UNet2DConditionModel,
         scheduler: Scheduler,
     ) -> StableDiffusionGeneratorPipeline:
@@ -378,38 +387,38 @@ class DenoiseLatentsInvocation(BaseInvocation):
             requires_safety_checker=False,
         )
 
+    @staticmethod
     def prep_control_data(
-        self,
         context: InvocationContext,
-        control_input: Optional[Union[ControlField, List[ControlField]]],
+        control_input: ControlField | list[ControlField] | None,
         latents_shape: List[int],
         exit_stack: ExitStack,
         do_classifier_free_guidance: bool = True,
-    ) -> Optional[List[ControlNetData]]:
-        # Assuming fixed dimensional scaling of LATENT_SCALE_FACTOR.
-        control_height_resize = latents_shape[2] * LATENT_SCALE_FACTOR
-        control_width_resize = latents_shape[3] * LATENT_SCALE_FACTOR
-        if control_input is None:
-            control_list = None
-        elif isinstance(control_input, list) and len(control_input) == 0:
-            control_list = None
-        elif isinstance(control_input, ControlField):
+    ) -> list[ControlNetData] | None:
+        # Normalize control_input to a list.
+        control_list: list[ControlField]
+        if isinstance(control_input, ControlField):
             control_list = [control_input]
-        elif isinstance(control_input, list) and len(control_input) > 0 and isinstance(control_input[0], ControlField):
+        elif isinstance(control_input, list):
             control_list = control_input
+        elif control_input is None:
+            control_list = []
         else:
-            control_list = None
-        if control_list is None:
-            return None
-        # After above handling, any control that is not None should now be of type list[ControlField].
+            raise ValueError(f"Unexpected control_input type: {type(control_input)}")
 
-        # FIXME: add checks to skip entry if model or image is None
-        #        and if weight is None, populate with default 1.0?
-        controlnet_data = []
+        if len(control_list) == 0:
+            return None
+
+        # Assuming fixed dimensional scaling of LATENT_SCALE_FACTOR.
+        _, _, latent_height, latent_width = latents_shape
+        control_height_resize = latent_height * LATENT_SCALE_FACTOR
+        control_width_resize = latent_width * LATENT_SCALE_FACTOR
+
+        controlnet_data: list[ControlNetData] = []
         for control_info in control_list:
             control_model = exit_stack.enter_context(context.models.load(control_info.control_model))
+            assert isinstance(control_model, ControlNetModel)
 
-            # control_models.append(control_model)
             control_image_field = control_info.image
             input_image = context.images.get_pil(control_image_field.image_name)
             # self.image.image_type, self.image.image_name
@@ -430,7 +439,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 resize_mode=control_info.resize_mode,
             )
             control_item = ControlNetData(
-                model=control_model,  # model object
+                model=control_model,
                 image_tensor=control_image,
                 weight=control_info.control_weight,
                 begin_step_percent=control_info.begin_step_percent,
@@ -584,15 +593,15 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
     # original idea by https://github.com/AmericanPresidentJimmyCarter
     # TODO: research more for second order schedulers timesteps
+    @staticmethod
     def init_scheduler(
-        self,
         scheduler: Union[Scheduler, ConfigMixin],
         device: torch.device,
         steps: int,
         denoising_start: float,
         denoising_end: float,
         seed: int,
-    ) -> Tuple[int, List[int], int, Dict[str, Any]]:
+    ) -> Tuple[torch.Tensor, torch.Tensor, Dict[str, Any]]:
         assert isinstance(scheduler, ConfigMixin)
         if scheduler.config.get("cpu_only", False):
             scheduler.set_timesteps(steps, device="cpu")
@@ -618,7 +627,6 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
         init_timestep = timesteps[t_start_idx : t_start_idx + 1]
         timesteps = timesteps[t_start_idx : t_start_idx + t_end_idx]
-        num_inference_steps = len(timesteps) // scheduler.order
 
         scheduler_step_kwargs: Dict[str, Any] = {}
         scheduler_step_signature = inspect.signature(scheduler.step)
@@ -640,7 +648,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
         if isinstance(scheduler, TCDScheduler):
             scheduler_step_kwargs.update({"eta": 1.0})
 
-        return num_inference_steps, timesteps, init_timestep, scheduler_step_kwargs
+        return timesteps, init_timestep, scheduler_step_kwargs
 
     def prep_inpaint_mask(
         self, context: InvocationContext, latents: torch.Tensor
@@ -657,31 +665,52 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
         return 1 - mask, masked_latents, self.denoise_mask.gradient
 
-    @torch.no_grad()
-    @SilenceWarnings()  # This quenches the NSFW nag from diffusers.
-    def invoke(self, context: InvocationContext) -> LatentsOutput:
-        seed = None
+    @staticmethod
+    def prepare_noise_and_latents(
+        context: InvocationContext, noise_field: LatentsField | None, latents_field: LatentsField | None
+    ) -> Tuple[int, torch.Tensor | None, torch.Tensor]:
+        """Depending on the workflow, we expect different combinations of noise and latents to be provided. This
+        function handles preparing these values accordingly.
+
+        Expected workflows:
+        - Text-to-Image Denoising: `noise` is provided, `latents` is not. `latents` is initialized to zeros.
+        - Image-to-Image Denoising: `noise` and `latents` are both provided.
+        - Text-to-Image SDXL Refiner Denoising: `latents` is provided, `noise` is not.
+        - Image-to-Image SDXL Refiner Denoising: `latents` is provided, `noise` is not.
+
+        NOTE(ryand): I wrote this docstring, but I am not the original author of this code. There may be other workflows
+        I haven't considered.
+        """
         noise = None
-        if self.noise is not None:
-            noise = context.tensors.load(self.noise.latents_name)
-            seed = self.noise.seed
-
-        if self.latents is not None:
-            latents = context.tensors.load(self.latents.latents_name)
-            if seed is None:
-                seed = self.latents.seed
-
-            if noise is not None and noise.shape[1:] != latents.shape[1:]:
-                raise Exception(f"Incompatable 'noise' and 'latents' shapes: {latents.shape=} {noise.shape=}")
+        if noise_field is not None:
+            noise = context.tensors.load(noise_field.latents_name)
 
+        if latents_field is not None:
+            latents = context.tensors.load(latents_field.latents_name)
         elif noise is not None:
             latents = torch.zeros_like(noise)
         else:
-            raise Exception("'latents' or 'noise' must be provided!")
+            raise ValueError("'latents' or 'noise' must be provided!")
 
-        if seed is None:
+        if noise is not None and noise.shape[1:] != latents.shape[1:]:
+            raise ValueError(f"Incompatable 'noise' and 'latents' shapes: {latents.shape=} {noise.shape=}")
+
+        # The seed comes from (in order of priority): the noise field, the latents field, or 0.
+        seed = 0
+        if noise_field is not None and noise_field.seed is not None:
+            seed = noise_field.seed
+        elif latents_field is not None and latents_field.seed is not None:
+            seed = latents_field.seed
+        else:
             seed = 0
 
+        return seed, noise, latents
+
+    @torch.no_grad()
+    @SilenceWarnings()  # This quenches the NSFW nag from diffusers.
+    def invoke(self, context: InvocationContext) -> LatentsOutput:
+        seed, noise, latents = self.prepare_noise_and_latents(context, self.noise, self.latents)
+
         mask, masked_latents, gradient_mask = self.prep_inpaint_mask(context, latents)
 
         # TODO(ryand): I have hard-coded `do_classifier_free_guidance=True` to mirror the behaviour of ControlNets,
@@ -755,7 +784,15 @@ class DenoiseLatentsInvocation(BaseInvocation):
 
             _, _, latent_height, latent_width = latents.shape
             conditioning_data = self.get_conditioning_data(
-                context=context, unet=unet, latent_height=latent_height, latent_width=latent_width
+                context=context,
+                positive_conditioning_field=self.positive_conditioning,
+                negative_conditioning_field=self.negative_conditioning,
+                unet=unet,
+                latent_height=latent_height,
+                latent_width=latent_width,
+                cfg_scale=self.cfg_scale,
+                steps=self.steps,
+                cfg_rescale_multiplier=self.cfg_rescale_multiplier,
             )
 
             controlnet_data = self.prep_control_data(
@@ -777,7 +814,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 dtype=unet.dtype,
             )
 
-            num_inference_steps, timesteps, init_timestep, scheduler_step_kwargs = self.init_scheduler(
+            timesteps, init_timestep, scheduler_step_kwargs = self.init_scheduler(
                 scheduler,
                 device=unet.device,
                 steps=self.steps,
@@ -794,8 +831,7 @@ class DenoiseLatentsInvocation(BaseInvocation):
                 seed=seed,
                 mask=mask,
                 masked_latents=masked_latents,
-                gradient_mask=gradient_mask,
-                num_inference_steps=num_inference_steps,
+                is_gradient_mask=gradient_mask,
                 scheduler_step_kwargs=scheduler_step_kwargs,
                 conditioning_data=conditioning_data,
                 control_data=controlnet_data,

invokeai/app/invocations/latents_to_image.py

@@ -8,7 +8,7 @@ from diffusers.models.attention_processor import (
 )
 from diffusers.models.autoencoders.autoencoder_kl import AutoencoderKL
 from diffusers.models.autoencoders.autoencoder_tiny import AutoencoderTiny
-from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
+from PIL import Image
 
 from invokeai.app.invocations.baseinvocation import BaseInvocation, invocation
 from invokeai.app.invocations.constants import DEFAULT_PRECISION
@@ -23,6 +23,7 @@ from invokeai.app.invocations.fields import (
 from invokeai.app.invocations.model import VAEField
 from invokeai.app.invocations.primitives import ImageOutput
 from invokeai.app.services.shared.invocation_context import InvocationContext
+from invokeai.backend.model_manager.load.load_base import LoadedModel
 from invokeai.backend.stable_diffusion import set_seamless
 from invokeai.backend.util.devices import TorchDevice
 
@@ -48,16 +49,20 @@ class LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
     tiled: bool = InputField(default=False, description=FieldDescriptions.tiled)
     fp32: bool = InputField(default=DEFAULT_PRECISION == torch.float32, description=FieldDescriptions.fp32)
 
-    @torch.no_grad()
-    def invoke(self, context: InvocationContext) -> ImageOutput:
-        latents = context.tensors.load(self.latents.latents_name)
-
-        vae_info = context.models.load(self.vae.vae)
-        assert isinstance(vae_info.model, (UNet2DConditionModel, AutoencoderKL, AutoencoderTiny))
-        with set_seamless(vae_info.model, self.vae.seamless_axes), vae_info as vae:
-            assert isinstance(vae, torch.nn.Module)
+    @staticmethod
+    def vae_decode(
+        context: InvocationContext,
+        vae_info: LoadedModel,
+        seamless_axes: list[str],
+        latents: torch.Tensor,
+        use_fp32: bool,
+        use_tiling: bool,
+    ) -> Image.Image:
+        assert isinstance(vae_info.model, (AutoencoderKL, AutoencoderTiny))
+        with set_seamless(vae_info.model, seamless_axes), vae_info as vae:
+            assert isinstance(vae, (AutoencoderKL, AutoencoderTiny))
             latents = latents.to(vae.device)
-            if self.fp32:
+            if use_fp32:
                 vae.to(dtype=torch.float32)
 
                 use_torch_2_0_or_xformers = hasattr(vae.decoder, "mid_block") and isinstance(
@@ -82,7 +87,7 @@ class LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
                 vae.to(dtype=torch.float16)
                 latents = latents.half()
 
-            if self.tiled or context.config.get().force_tiled_decode:
+            if use_tiling or context.config.get().force_tiled_decode:
                 vae.enable_tiling()
             else:
                 vae.disable_tiling()
@@ -102,6 +107,21 @@ class LatentsToImageInvocation(BaseInvocation, WithMetadata, WithBoard):
 
         TorchDevice.empty_cache()
 
+        return image
+
+    @torch.no_grad()
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        latents = context.tensors.load(self.latents.latents_name)
+        vae_info = context.models.load(self.vae.vae)
+
+        image = self.vae_decode(
+            context=context,
+            vae_info=vae_info,
+            seamless_axes=self.vae.seamless_axes,
+            latents=latents,
+            use_fp32=self.fp32,
+            use_tiling=self.tiled,
+        )
         image_dto = context.images.save(image=image)
 
         return ImageOutput.build(image_dto)

invokeai/app/invocations/tiled_multi_diffusion_denoise_latents.py

@@ -0,0 +1,268 @@
+import copy
+from contextlib import ExitStack
+from typing import Iterator, Tuple
+
+import torch
+from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
+from diffusers.schedulers.scheduling_utils import SchedulerMixin
+from pydantic import field_validator
+
+from invokeai.app.invocations.baseinvocation import BaseInvocation, invocation
+from invokeai.app.invocations.constants import LATENT_SCALE_FACTOR, SCHEDULER_NAME_VALUES
+from invokeai.app.invocations.controlnet_image_processors import ControlField
+from invokeai.app.invocations.denoise_latents import DenoiseLatentsInvocation, get_scheduler
+from invokeai.app.invocations.fields import (
+    ConditioningField,
+    FieldDescriptions,
+    Input,
+    InputField,
+    LatentsField,
+    UIType,
+)
+from invokeai.app.invocations.model import UNetField
+from invokeai.app.invocations.primitives import LatentsOutput
+from invokeai.app.services.shared.invocation_context import InvocationContext
+from invokeai.backend.lora import LoRAModelRaw
+from invokeai.backend.model_patcher import ModelPatcher
+from invokeai.backend.stable_diffusion.diffusers_pipeline import ControlNetData
+from invokeai.backend.stable_diffusion.multi_diffusion_pipeline import (
+    MultiDiffusionPipeline,
+    MultiDiffusionRegionConditioning,
+)
+from invokeai.backend.tiles.tiles import (
+    calc_tiles_min_overlap,
+)
+from invokeai.backend.tiles.utils import TBLR
+from invokeai.backend.util.devices import TorchDevice
+
+
+def crop_controlnet_data(control_data: ControlNetData, latent_region: TBLR) -> ControlNetData:
+    """Crop a ControlNetData object to a region."""
+    # Create a shallow copy of the control_data object.
+    control_data_copy = copy.copy(control_data)
+    # The ControlNet reference image is the only attribute that needs to be cropped.
+    control_data_copy.image_tensor = control_data.image_tensor[
+        :,
+        :,
+        latent_region.top * LATENT_SCALE_FACTOR : latent_region.bottom * LATENT_SCALE_FACTOR,
+        latent_region.left * LATENT_SCALE_FACTOR : latent_region.right * LATENT_SCALE_FACTOR,
+    ]
+    return control_data_copy
+
+
+@invocation(
+    "tiled_multi_diffusion_denoise_latents",
+    title="Tiled Multi-Diffusion Denoise Latents",
+    tags=["upscale", "denoise"],
+    category="latents",
+    # TODO(ryand): Reset to 1.0.0 right before release.
+    version="1.0.0",
+)
+class TiledMultiDiffusionDenoiseLatents(BaseInvocation):
+    """Tiled Multi-Diffusion denoising.
+
+    This node handles automatically tiling the input image. Future iterations of
+    this node should allow the user to specify custom regions with different parameters for each region to harness the
+    full power of Multi-Diffusion.
+
+    This node has a similar interface to the `DenoiseLatents` node, but it has a reduced feature set (no IP-Adapter,
+    T2I-Adapter, masking, etc.).
+    """
+
+    positive_conditioning: ConditioningField = InputField(
+        description=FieldDescriptions.positive_cond, input=Input.Connection
+    )
+    negative_conditioning: ConditioningField = InputField(
+        description=FieldDescriptions.negative_cond, input=Input.Connection
+    )
+    noise: LatentsField | None = InputField(
+        default=None,
+        description=FieldDescriptions.noise,
+        input=Input.Connection,
+    )
+    latents: LatentsField | None = InputField(
+        default=None,
+        description=FieldDescriptions.latents,
+        input=Input.Connection,
+    )
+    # TODO(ryand): Add multiple-of validation.
+    # TODO(ryand): Smaller defaults might make more sense.
+    tile_height: int = InputField(default=112, gt=0, description="Height of the tiles in latent space.")
+    tile_width: int = InputField(default=112, gt=0, description="Width of the tiles in latent space.")
+    tile_min_overlap: int = InputField(
+        default=16,
+        gt=0,
+        description="The minimum overlap between adjacent tiles in latent space. The actual overlap may be larger than "
+        "this to evenly cover the entire image.",
+    )
+    steps: int = InputField(default=18, gt=0, description=FieldDescriptions.steps)
+    cfg_scale: float | list[float] = InputField(default=6.0, description=FieldDescriptions.cfg_scale, title="CFG Scale")
+    # TODO(ryand): The default here should probably be 0.0.
+    denoising_start: float = InputField(
+        default=0.65,
+        ge=0,
+        le=1,
+        description=FieldDescriptions.denoising_start,
+    )
+    denoising_end: float = InputField(default=1.0, ge=0, le=1, description=FieldDescriptions.denoising_end)
+    scheduler: SCHEDULER_NAME_VALUES = InputField(
+        default="euler",
+        description=FieldDescriptions.scheduler,
+        ui_type=UIType.Scheduler,
+    )
+    unet: UNetField = InputField(
+        description=FieldDescriptions.unet,
+        input=Input.Connection,
+        title="UNet",
+    )
+    cfg_rescale_multiplier: float = InputField(
+        title="CFG Rescale Multiplier", default=0, ge=0, lt=1, description=FieldDescriptions.cfg_rescale_multiplier
+    )
+    control: ControlField | list[ControlField] | None = InputField(
+        default=None,
+        input=Input.Connection,
+    )
+
+    @field_validator("cfg_scale")
+    def ge_one(cls, v: list[float] | float) -> list[float] | float:
+        """Validate that all cfg_scale values are >= 1"""
+        if isinstance(v, list):
+            for i in v:
+                if i < 1:
+                    raise ValueError("cfg_scale must be greater than 1")
+        else:
+            if v < 1:
+                raise ValueError("cfg_scale must be greater than 1")
+        return v
+
+    @staticmethod
+    def create_pipeline(
+        unet: UNet2DConditionModel,
+        scheduler: SchedulerMixin,
+    ) -> MultiDiffusionPipeline:
+        # TODO(ryand): Get rid of this FakeVae hack.
+        class FakeVae:
+            class FakeVaeConfig:
+                def __init__(self) -> None:
+                    self.block_out_channels = [0]
+
+            def __init__(self) -> None:
+                self.config = FakeVae.FakeVaeConfig()
+
+        return MultiDiffusionPipeline(
+            vae=FakeVae(),  # TODO: oh...
+            text_encoder=None,
+            tokenizer=None,
+            unet=unet,
+            scheduler=scheduler,
+            safety_checker=None,
+            feature_extractor=None,
+            requires_safety_checker=False,
+        )
+
+    @torch.no_grad()
+    def invoke(self, context: InvocationContext) -> LatentsOutput:
+        seed, noise, latents = DenoiseLatentsInvocation.prepare_noise_and_latents(context, self.noise, self.latents)
+        _, _, latent_height, latent_width = latents.shape
+
+        # Calculate the tile locations to cover the latent-space image.
+        # TODO(ryand): Add constraints on the tile params. Is there a multiple-of constraint?
+        tiles = calc_tiles_min_overlap(
+            image_height=latent_height,
+            image_width=latent_width,
+            tile_height=self.tile_height,
+            tile_width=self.tile_width,
+            min_overlap=self.tile_min_overlap,
+        )
+
+        # Prepare an iterator that yields the UNet's LoRA models and their weights.
+        def _lora_loader() -> Iterator[Tuple[LoRAModelRaw, float]]:
+            for lora in self.unet.loras:
+                lora_info = context.models.load(lora.lora)
+                assert isinstance(lora_info.model, LoRAModelRaw)
+                yield (lora_info.model, lora.weight)
+                del lora_info
+
+        # Load the UNet model.
+        unet_info = context.models.load(self.unet.unet)
+
+        with ExitStack() as exit_stack, unet_info as unet, ModelPatcher.apply_lora_unet(unet, _lora_loader()):
+            assert isinstance(unet, UNet2DConditionModel)
+            latents = latents.to(device=unet.device, dtype=unet.dtype)
+            if noise is not None:
+                noise = noise.to(device=unet.device, dtype=unet.dtype)
+            scheduler = get_scheduler(
+                context=context,
+                scheduler_info=self.unet.scheduler,
+                scheduler_name=self.scheduler,
+                seed=seed,
+            )
+            pipeline = self.create_pipeline(unet=unet, scheduler=scheduler)
+
+            # Prepare the prompt conditioning data. The same prompt conditioning is applied to all tiles.
+            conditioning_data = DenoiseLatentsInvocation.get_conditioning_data(
+                context=context,
+                positive_conditioning_field=self.positive_conditioning,
+                negative_conditioning_field=self.negative_conditioning,
+                unet=unet,
+                latent_height=self.tile_height,
+                latent_width=self.tile_width,
+                cfg_scale=self.cfg_scale,
+                steps=self.steps,
+                cfg_rescale_multiplier=self.cfg_rescale_multiplier,
+            )
+
+            controlnet_data = DenoiseLatentsInvocation.prep_control_data(
+                context=context,
+                control_input=self.control,
+                latents_shape=list(latents.shape),
+                # do_classifier_free_guidance=(self.cfg_scale >= 1.0))
+                do_classifier_free_guidance=True,
+                exit_stack=exit_stack,
+            )
+
+            # Split the controlnet_data into tiles.
+            # controlnet_data_tiles[t][c] is the c'th control data for the t'th tile.
+            controlnet_data_tiles: list[list[ControlNetData]] = []
+            for tile in tiles:
+                tile_controlnet_data = [crop_controlnet_data(cn, tile.coords) for cn in controlnet_data or []]
+                controlnet_data_tiles.append(tile_controlnet_data)
+
+            # Prepare the MultiDiffusionRegionConditioning list.
+            multi_diffusion_conditioning: list[MultiDiffusionRegionConditioning] = []
+            for tile, tile_controlnet_data in zip(tiles, controlnet_data_tiles, strict=True):
+                multi_diffusion_conditioning.append(
+                    MultiDiffusionRegionConditioning(
+                        region=tile.coords,
+                        text_conditioning_data=conditioning_data,
+                        control_data=tile_controlnet_data,
+                    )
+                )
+
+            timesteps, init_timestep, scheduler_step_kwargs = DenoiseLatentsInvocation.init_scheduler(
+                scheduler,
+                device=unet.device,
+                steps=self.steps,
+                denoising_start=self.denoising_start,
+                denoising_end=self.denoising_end,
+                seed=seed,
+            )
+
+            # Run Multi-Diffusion denoising.
+            result_latents = pipeline.multi_diffusion_denoise(
+                multi_diffusion_conditioning=multi_diffusion_conditioning,
+                latents=latents,
+                scheduler_step_kwargs=scheduler_step_kwargs,
+                noise=noise,
+                timesteps=timesteps,
+                init_timestep=init_timestep,
+                # TODO(ryand): Add proper callback.
+                callback=lambda x: None,
+            )
+
+        # TODO(ryand): I copied this from DenoiseLatentsInvocation. I'm not sure if it's actually important.
+        result_latents = result_latents.to("cpu")
+        TorchDevice.empty_cache()
+
+        name = context.tensors.save(tensor=result_latents)
+        return LatentsOutput.build(latents_name=name, latents=result_latents, seed=None)

invokeai/app/invocations/tiled_stable_diffusion_refine.py

@@ -0,0 +1,380 @@
+from contextlib import ExitStack
+from typing import Iterator, Tuple
+
+import numpy as np
+import numpy.typing as npt
+import torch
+from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
+from PIL import Image
+from pydantic import field_validator
+
+from invokeai.app.invocations.baseinvocation import BaseInvocation, invocation
+from invokeai.app.invocations.constants import DEFAULT_PRECISION, LATENT_SCALE_FACTOR, SCHEDULER_NAME_VALUES
+from invokeai.app.invocations.denoise_latents import DenoiseLatentsInvocation, get_scheduler
+from invokeai.app.invocations.fields import (
+    ConditioningField,
+    FieldDescriptions,
+    ImageField,
+    Input,
+    InputField,
+    UIType,
+)
+from invokeai.app.invocations.image_to_latents import ImageToLatentsInvocation
+from invokeai.app.invocations.latents_to_image import LatentsToImageInvocation
+from invokeai.app.invocations.model import ModelIdentifierField, UNetField, VAEField
+from invokeai.app.invocations.noise import get_noise
+from invokeai.app.invocations.primitives import ImageOutput
+from invokeai.app.services.shared.invocation_context import InvocationContext
+from invokeai.app.util.controlnet_utils import CONTROLNET_MODE_VALUES, CONTROLNET_RESIZE_VALUES, prepare_control_image
+from invokeai.backend.lora import LoRAModelRaw
+from invokeai.backend.model_patcher import ModelPatcher
+from invokeai.backend.stable_diffusion.diffusers_pipeline import ControlNetData, image_resized_to_grid_as_tensor
+from invokeai.backend.tiles.tiles import calc_tiles_with_overlap, merge_tiles_with_linear_blending
+from invokeai.backend.tiles.utils import Tile
+from invokeai.backend.util.devices import TorchDevice
+from invokeai.backend.util.hotfixes import ControlNetModel
+
+
+@invocation(
+    "tiled_stable_diffusion_refine",
+    title="Tiled Stable Diffusion Refine",
+    tags=["upscale", "denoise"],
+    category="latents",
+    version="1.0.0",
+)
+class TiledStableDiffusionRefineInvocation(BaseInvocation):
+    """A tiled Stable Diffusion pipeline for refining high resolution images. This invocation is intended to be used to
+    refine an image after upscaling i.e. it is the second step in a typical "tiled upscaling" workflow.
+    """
+
+    image: ImageField = InputField(description="Image to be refined.")
+
+    positive_conditioning: ConditioningField = InputField(
+        description=FieldDescriptions.positive_cond, input=Input.Connection
+    )
+    negative_conditioning: ConditioningField = InputField(
+        description=FieldDescriptions.negative_cond, input=Input.Connection
+    )
+    # TODO(ryand): Add multiple-of validation.
+    tile_height: int = InputField(default=512, gt=0, description="Height of the tiles.")
+    tile_width: int = InputField(default=512, gt=0, description="Width of the tiles.")
+    tile_overlap: int = InputField(
+        default=16,
+        gt=0,
+        description="Target overlap between adjacent tiles (the last row/column may overlap more than this).",
+    )
+    steps: int = InputField(default=18, gt=0, description=FieldDescriptions.steps)
+    cfg_scale: float | list[float] = InputField(default=6.0, description=FieldDescriptions.cfg_scale, title="CFG Scale")
+    denoising_start: float = InputField(
+        default=0.65,
+        ge=0,
+        le=1,
+        description=FieldDescriptions.denoising_start,
+    )
+    denoising_end: float = InputField(default=1.0, ge=0, le=1, description=FieldDescriptions.denoising_end)
+    scheduler: SCHEDULER_NAME_VALUES = InputField(
+        default="euler",
+        description=FieldDescriptions.scheduler,
+        ui_type=UIType.Scheduler,
+    )
+    unet: UNetField = InputField(
+        description=FieldDescriptions.unet,
+        input=Input.Connection,
+        title="UNet",
+    )
+    cfg_rescale_multiplier: float = InputField(
+        title="CFG Rescale Multiplier", default=0, ge=0, lt=1, description=FieldDescriptions.cfg_rescale_multiplier
+    )
+    vae: VAEField = InputField(
+        description=FieldDescriptions.vae,
+        input=Input.Connection,
+    )
+    vae_fp32: bool = InputField(
+        default=DEFAULT_PRECISION == torch.float32, description="Whether to use float32 precision when running the VAE."
+    )
+    # HACK(ryand): We probably want to allow the user to control all of the parameters in ControlField. But, we akwardly
+    # don't want to use the image field. Figure out how best to handle this.
+    # TODO(ryand): Currently, there is no ControlNet preprocessor applied to the tile images. In other words, we pretty
+    # much assume that it is a tile ControlNet. We need to decide how we want to handle this. E.g. find a way to support
+    # CN preprocessors, raise a clear warning when a non-tile CN model is selected, hardcode the supported CN models,
+    # etc.
+    control_model: ModelIdentifierField = InputField(
+        description=FieldDescriptions.controlnet_model, ui_type=UIType.ControlNetModel
+    )
+    control_weight: float = InputField(default=0.6)
+
+    @field_validator("cfg_scale")
+    def ge_one(cls, v: list[float] | float) -> list[float] | float:
+        """Validate that all cfg_scale values are >= 1"""
+        if isinstance(v, list):
+            for i in v:
+                if i < 1:
+                    raise ValueError("cfg_scale must be greater than 1")
+        else:
+            if v < 1:
+                raise ValueError("cfg_scale must be greater than 1")
+        return v
+
+    @staticmethod
+    def crop_latents_to_tile(latents: torch.Tensor, image_tile: Tile) -> torch.Tensor:
+        """Crop the latent-space tensor to the area corresponding to the image-space tile.
+        The tile coordinates must be divisible by the LATENT_SCALE_FACTOR.
+        """
+        for coord in [image_tile.coords.top, image_tile.coords.left, image_tile.coords.right, image_tile.coords.bottom]:
+            if coord % LATENT_SCALE_FACTOR != 0:
+                raise ValueError(
+                    f"The tile coordinates must all be divisible by the latent scale factor"
+                    f" ({LATENT_SCALE_FACTOR}). {image_tile.coords=}."
+                )
+        assert latents.dim() == 4  # We expect: (batch_size, channels, height, width).
+
+        top = image_tile.coords.top // LATENT_SCALE_FACTOR
+        left = image_tile.coords.left // LATENT_SCALE_FACTOR
+        bottom = image_tile.coords.bottom // LATENT_SCALE_FACTOR
+        right = image_tile.coords.right // LATENT_SCALE_FACTOR
+        return latents[..., top:bottom, left:right]
+
+    def run_controlnet(
+        self,
+        image: Image.Image,
+        controlnet_model: ControlNetModel,
+        weight: float,
+        do_classifier_free_guidance: bool,
+        width: int,
+        height: int,
+        device: torch.device,
+        dtype: torch.dtype,
+        control_mode: CONTROLNET_MODE_VALUES = "balanced",
+        resize_mode: CONTROLNET_RESIZE_VALUES = "just_resize_simple",
+    ) -> ControlNetData:
+        control_image = prepare_control_image(
+            image=image,
+            do_classifier_free_guidance=do_classifier_free_guidance,
+            width=width,
+            height=height,
+            device=device,
+            dtype=dtype,
+            control_mode=control_mode,
+            resize_mode=resize_mode,
+        )
+        return ControlNetData(
+            model=controlnet_model,
+            image_tensor=control_image,
+            weight=weight,
+            begin_step_percent=0.0,
+            end_step_percent=1.0,
+            control_mode=control_mode,
+            # Any resizing needed should currently be happening in prepare_control_image(), but adding resize_mode to
+            # ControlNetData in case needed in the future.
+            resize_mode=resize_mode,
+        )
+
+    @torch.no_grad()
+    def invoke(self, context: InvocationContext) -> ImageOutput:
+        # TODO(ryand): Expose the seed parameter.
+        seed = 0
+
+        # Load the input image.
+        input_image = context.images.get_pil(self.image.image_name)
+
+        # Calculate the tile locations to cover the image.
+        # We have selected this tiling strategy to make it easy to achieve tile coords that are multiples of 8. This
+        # facilitates conversions between image space and latent space.
+        # TODO(ryand): Expose these tiling parameters. (Keep in mind the multiple-of constraints on these params.)
+        tiles = calc_tiles_with_overlap(
+            image_height=input_image.height,
+            image_width=input_image.width,
+            tile_height=self.tile_height,
+            tile_width=self.tile_width,
+            overlap=self.tile_overlap,
+        )
+
+        # Convert the input image to a torch.Tensor.
+        input_image_torch = image_resized_to_grid_as_tensor(input_image.convert("RGB"), multiple_of=LATENT_SCALE_FACTOR)
+        input_image_torch = input_image_torch.unsqueeze(0)  # Add a batch dimension.
+        # Validate our assumptions about the shape of input_image_torch.
+        assert input_image_torch.dim() == 4  # We expect: (batch_size, channels, height, width).
+        assert input_image_torch.shape[:2] == (1, 3)
+
+        # Split the input image into tiles in torch.Tensor format.
+        image_tiles_torch: list[torch.Tensor] = []
+        for tile in tiles:
+            image_tile = input_image_torch[
+                :,
+                :,
+                tile.coords.top : tile.coords.bottom,
+                tile.coords.left : tile.coords.right,
+            ]
+            image_tiles_torch.append(image_tile)
+
+        # Split the input image into tiles in numpy format.
+        # TODO(ryand): We currently maintain both np.ndarray and torch.Tensor tiles. Ideally, all operations should work
+        # with torch.Tensor tiles.
+        input_image_np = np.array(input_image)
+        image_tiles_np: list[npt.NDArray[np.uint8]] = []
+        for tile in tiles:
+            image_tile_np = input_image_np[
+                tile.coords.top : tile.coords.bottom,
+                tile.coords.left : tile.coords.right,
+                :,
+            ]
+            image_tiles_np.append(image_tile_np)
+
+        # VAE-encode each image tile independently.
+        # TODO(ryand): Is there any advantage to VAE-encoding the entire image before splitting it into tiles? What
+        # about for decoding?
+        vae_info = context.models.load(self.vae.vae)
+        latent_tiles: list[torch.Tensor] = []
+        for image_tile_torch in image_tiles_torch:
+            latent_tiles.append(
+                ImageToLatentsInvocation.vae_encode(
+                    vae_info=vae_info, upcast=self.vae_fp32, tiled=False, image_tensor=image_tile_torch
+                )
+            )
+
+        # Generate noise with dimensions corresponding to the full image in latent space.
+        # It is important that the noise tensor is generated at the full image dimension and then tiled, rather than
+        # generating for each tile independently. This ensures that overlapping regions between tiles use the same
+        # noise.
+        assert input_image_torch.shape[2] % LATENT_SCALE_FACTOR == 0
+        assert input_image_torch.shape[3] % LATENT_SCALE_FACTOR == 0
+        global_noise = get_noise(
+            width=input_image_torch.shape[3],
+            height=input_image_torch.shape[2],
+            device=TorchDevice.choose_torch_device(),
+            seed=seed,
+            downsampling_factor=LATENT_SCALE_FACTOR,
+            use_cpu=True,
+        )
+
+        # Crop the global noise into tiles.
+        noise_tiles = [self.crop_latents_to_tile(latents=global_noise, image_tile=t) for t in tiles]
+
+        # Prepare an iterator that yields the UNet's LoRA models and their weights.
+        def _lora_loader() -> Iterator[Tuple[LoRAModelRaw, float]]:
+            for lora in self.unet.loras:
+                lora_info = context.models.load(lora.lora)
+                assert isinstance(lora_info.model, LoRAModelRaw)
+                yield (lora_info.model, lora.weight)
+                del lora_info
+
+        # Load the UNet model.
+        unet_info = context.models.load(self.unet.unet)
+
+        refined_latent_tiles: list[torch.Tensor] = []
+        with ExitStack() as exit_stack, unet_info as unet, ModelPatcher.apply_lora_unet(unet, _lora_loader()):
+            assert isinstance(unet, UNet2DConditionModel)
+            scheduler = get_scheduler(
+                context=context,
+                scheduler_info=self.unet.scheduler,
+                scheduler_name=self.scheduler,
+                seed=seed,
+            )
+            pipeline = DenoiseLatentsInvocation.create_pipeline(unet=unet, scheduler=scheduler)
+
+            # Prepare the prompt conditioning data. The same prompt conditioning is applied to all tiles.
+            # Assume that all tiles have the same shape.
+            _, _, latent_height, latent_width = latent_tiles[0].shape
+            conditioning_data = DenoiseLatentsInvocation.get_conditioning_data(
+                context=context,
+                positive_conditioning_field=self.positive_conditioning,
+                negative_conditioning_field=self.negative_conditioning,
+                unet=unet,
+                latent_height=latent_height,
+                latent_width=latent_width,
+                cfg_scale=self.cfg_scale,
+                steps=self.steps,
+                cfg_rescale_multiplier=self.cfg_rescale_multiplier,
+            )
+
+            # Load the ControlNet model.
+            # TODO(ryand): Support multiple ControlNet models.
+            controlnet_model = exit_stack.enter_context(context.models.load(self.control_model))
+            assert isinstance(controlnet_model, ControlNetModel)
+
+            # Denoise (i.e. "refine") each tile independently.
+            for image_tile_np, latent_tile, noise_tile in zip(image_tiles_np, latent_tiles, noise_tiles, strict=True):
+                assert latent_tile.shape == noise_tile.shape
+
+                # Prepare a PIL Image for ControlNet processing.
+                # TODO(ryand): This is a bit awkward that we have to prepare both torch.Tensor and PIL.Image versions of
+                # the tiles. Ideally, the ControlNet code should be able to work with Tensors.
+                image_tile_pil = Image.fromarray(image_tile_np)
+
+                # Run the ControlNet on the image tile.
+                height, width, _ = image_tile_np.shape
+                # The height and width must be evenly divisible by LATENT_SCALE_FACTOR. This is enforced earlier, but we
+                # validate this assumption here.
+                assert height % LATENT_SCALE_FACTOR == 0
+                assert width % LATENT_SCALE_FACTOR == 0
+                controlnet_data = self.run_controlnet(
+                    image=image_tile_pil,
+                    controlnet_model=controlnet_model,
+                    weight=self.control_weight,
+                    do_classifier_free_guidance=True,
+                    width=width,
+                    height=height,
+                    device=controlnet_model.device,
+                    dtype=controlnet_model.dtype,
+                    control_mode="balanced",
+                    resize_mode="just_resize_simple",
+                )
+
+                timesteps, init_timestep, scheduler_step_kwargs = DenoiseLatentsInvocation.init_scheduler(
+                    scheduler,
+                    device=unet.device,
+                    steps=self.steps,
+                    denoising_start=self.denoising_start,
+                    denoising_end=self.denoising_end,
+                    seed=seed,
+                )
+
+                # TODO(ryand): Think about when/if latents/noise should be moved off of the device to save VRAM.
+                latent_tile = latent_tile.to(device=unet.device, dtype=unet.dtype)
+                noise_tile = noise_tile.to(device=unet.device, dtype=unet.dtype)
+                refined_latent_tile = pipeline.latents_from_embeddings(
+                    latents=latent_tile,
+                    timesteps=timesteps,
+                    init_timestep=init_timestep,
+                    noise=noise_tile,
+                    seed=seed,
+                    mask=None,
+                    masked_latents=None,
+                    scheduler_step_kwargs=scheduler_step_kwargs,
+                    conditioning_data=conditioning_data,
+                    control_data=[controlnet_data],
+                    ip_adapter_data=None,
+                    t2i_adapter_data=None,
+                    callback=lambda x: None,
+                )
+                refined_latent_tiles.append(refined_latent_tile)
+
+        # VAE-decode each refined latent tile independently.
+        refined_image_tiles: list[Image.Image] = []
+        for refined_latent_tile in refined_latent_tiles:
+            refined_image_tile = LatentsToImageInvocation.vae_decode(
+                context=context,
+                vae_info=vae_info,
+                seamless_axes=self.vae.seamless_axes,
+                latents=refined_latent_tile,
+                use_fp32=self.vae_fp32,
+                use_tiling=False,
+            )
+            refined_image_tiles.append(refined_image_tile)
+
+        # TODO(ryand): I copied this from DenoiseLatentsInvocation. I'm not sure if it's actually important.
+        TorchDevice.empty_cache()
+
+        # Merge the refined image tiles back into a single image.
+        refined_image_tiles_np = [np.array(t) for t in refined_image_tiles]
+        merged_image_np = np.zeros(shape=(input_image.height, input_image.width, 3), dtype=np.uint8)
+        # TODO(ryand): Tune the blend_amount. Should this be exposed as a parameter?
+        merge_tiles_with_linear_blending(
+            dst_image=merged_image_np, tiles=tiles, tile_images=refined_image_tiles_np, blend_amount=self.tile_overlap
+        )
+
+        # Save the refined image and return its reference.
+        merged_image_pil = Image.fromarray(merged_image_np)
+        image_dto = context.images.save(image=merged_image_pil)
+
+        return ImageOutput.build(image_dto)

invokeai/app/services/config/config_default.py

@@ -26,13 +26,13 @@ LEGACY_INIT_FILE = Path("invokeai.init")
 DEFAULT_RAM_CACHE = 10.0
 DEFAULT_VRAM_CACHE = 0.25
 DEFAULT_CONVERT_CACHE = 20.0
-DEVICE = Literal["auto", "cpu", "cuda:0", "cuda:1", "cuda:2", "cuda:3", "cuda:4", "cuda:5", "cuda:6", "cuda:7", "mps"]
-PRECISION = Literal["auto", "float16", "bfloat16", "float32", "autocast"]
+DEVICE = Literal["auto", "cpu", "cuda", "cuda:1", "mps"]
+PRECISION = Literal["auto", "float16", "bfloat16", "float32"]
 ATTENTION_TYPE = Literal["auto", "normal", "xformers", "sliced", "torch-sdp"]
 ATTENTION_SLICE_SIZE = Literal["auto", "balanced", "max", 1, 2, 3, 4, 5, 6, 7, 8]
 LOG_FORMAT = Literal["plain", "color", "syslog", "legacy"]
 LOG_LEVEL = Literal["debug", "info", "warning", "error", "critical"]
-CONFIG_SCHEMA_VERSION = "4.0.2"
+CONFIG_SCHEMA_VERSION = "4.0.1"
 
 
 def get_default_ram_cache_size() -> float:
@@ -105,17 +105,14 @@ class InvokeAIAppConfig(BaseSettings):
         convert_cache: Maximum size of on-disk converted models cache (GB).
         lazy_offload: Keep models in VRAM until their space is needed.
         log_memory_usage: If True, a memory snapshot will be captured before and after every model cache operation, and the result will be logged (at debug level). There is a time cost to capturing the memory snapshots, so it is recommended to only enable this feature if you are actively inspecting the model cache's behaviour.
-        device: Preferred execution device. `auto` will choose the device depending on the hardware platform and the installed torch capabilities.<br>Valid values: `auto`, `cpu`, `cuda:0`, `cuda:1`, `cuda:2`, `cuda:3`, `cuda:4`, `cuda:5`, `cuda:6`, `cuda:7`, `mps`
-        devices: List of execution devices; will override default device selected.
-        precision: Floating point precision. `float16` will consume half the memory of `float32` but produce slightly lower-quality images. The `auto` setting will guess the proper precision based on your video card and operating system.<br>Valid values: `auto`, `float16`, `bfloat16`, `float32`, `autocast`
+        device: Preferred execution device. `auto` will choose the device depending on the hardware platform and the installed torch capabilities.<br>Valid values: `auto`, `cpu`, `cuda`, `cuda:1`, `mps`
+        precision: Floating point precision. `float16` will consume half the memory of `float32` but produce slightly lower-quality images. The `auto` setting will guess the proper precision based on your video card and operating system.<br>Valid values: `auto`, `float16`, `bfloat16`, `float32`
         sequential_guidance: Whether to calculate guidance in serial instead of in parallel, lowering memory requirements.
         attention_type: Attention type.<br>Valid values: `auto`, `normal`, `xformers`, `sliced`, `torch-sdp`
         attention_slice_size: Slice size, valid when attention_type=="sliced".<br>Valid values: `auto`, `balanced`, `max`, `1`, `2`, `3`, `4`, `5`, `6`, `7`, `8`
         force_tiled_decode: Whether to enable tiled VAE decode (reduces memory consumption with some performance penalty).
         pil_compress_level: The compress_level setting of PIL.Image.save(), used for PNG encoding. All settings are lossless. 0 = no compression, 1 = fastest with slightly larger filesize, 9 = slowest with smallest filesize. 1 is typically the best setting.
         max_queue_size: Maximum number of items in the session queue.
-        max_threads: Maximum number of session queue execution threads. Autocalculated from number of GPUs if not set.
-        clear_queue_on_startup: Empties session queue on startup.
         allow_nodes: List of nodes to allow. Omit to allow all.
         deny_nodes: List of nodes to deny. Omit to deny none.
         node_cache_size: How many cached nodes to keep in memory.
@@ -180,7 +177,6 @@ class InvokeAIAppConfig(BaseSettings):
 
     # DEVICE
     device:                      DEVICE = Field(default="auto",             description="Preferred execution device. `auto` will choose the device depending on the hardware platform and the installed torch capabilities.")
-    devices:      Optional[list[DEVICE]] = Field(default=None,              description="List of execution devices; will override default device selected.")
     precision:                PRECISION = Field(default="auto",             description="Floating point precision. `float16` will consume half the memory of `float32` but produce slightly lower-quality images. The `auto` setting will guess the proper precision based on your video card and operating system.")
 
     # GENERATION
@@ -190,8 +186,6 @@ class InvokeAIAppConfig(BaseSettings):
     force_tiled_decode:            bool = Field(default=False,              description="Whether to enable tiled VAE decode (reduces memory consumption with some performance penalty).")
     pil_compress_level:             int = Field(default=1,                  description="The compress_level setting of PIL.Image.save(), used for PNG encoding. All settings are lossless. 0 = no compression, 1 = fastest with slightly larger filesize, 9 = slowest with smallest filesize. 1 is typically the best setting.")
     max_queue_size:                 int = Field(default=10000, gt=0,        description="Maximum number of items in the session queue.")
-    max_threads:          Optional[int] = Field(default=None,               description="Maximum number of session queue execution threads. Autocalculated from number of GPUs if not set.")
-    clear_queue_on_startup:        bool = Field(default=False,              description="Empties session queue on startup.")
 
     # NODES
     allow_nodes:    Optional[list[str]] = Field(default=None,               description="List of nodes to allow. Omit to allow all.")
@@ -380,6 +374,9 @@ def migrate_v3_config_dict(config_dict: dict[str, Any]) -> InvokeAIAppConfig:
             # `max_cache_size` was renamed to `ram` some time in v3, but both names were used
             if k == "max_cache_size" and "ram" not in category_dict:
                 parsed_config_dict["ram"] = v
+            # `max_vram_cache_size` was renamed to `vram` some time in v3, but both names were used
+            if k == "max_vram_cache_size" and "vram" not in category_dict:
+                parsed_config_dict["vram"] = v
             # autocast was removed in v4.0.1
             if k == "precision" and v == "autocast":
                 parsed_config_dict["precision"] = "auto"
@@ -427,27 +424,6 @@ def migrate_v4_0_0_config_dict(config_dict: dict[str, Any]) -> InvokeAIAppConfig
     return config
 
 
-def migrate_v4_0_1_config_dict(config_dict: dict[str, Any]) -> InvokeAIAppConfig:
-    """Migrate v4.0.1 config dictionary to a current config object.
-
-    A few new multi-GPU options were added in 4.0.2, and this simply
-    updates the schema label.
-
-    Args:
-        config_dict: A dictionary of settings from a v4.0.1 config file.
-
-    Returns:
-        An instance of `InvokeAIAppConfig` with the migrated settings.
-    """
-    parsed_config_dict: dict[str, Any] = {}
-    for k, _ in config_dict.items():
-        if k == "schema_version":
-            parsed_config_dict[k] = CONFIG_SCHEMA_VERSION
-    config = DefaultInvokeAIAppConfig.model_validate(parsed_config_dict)
-    return config
-
-
-# TO DO: replace this with a formal registration and migration system
 def load_and_migrate_config(config_path: Path) -> InvokeAIAppConfig:
     """Load and migrate a config file to the latest version.
 
@@ -479,10 +455,6 @@ def load_and_migrate_config(config_path: Path) -> InvokeAIAppConfig:
         loaded_config_dict = migrate_v4_0_0_config_dict(loaded_config_dict)
         loaded_config_dict.write_file(config_path)
 
-    elif loaded_config_dict["schema_version"] == "4.0.1":
-        loaded_config_dict = migrate_v4_0_1_config_dict(loaded_config_dict)
-        loaded_config_dict.write_file(config_path)
-
     # Attempt to load as a v4 config file
     try:
         # Meta is not included in the model fields, so we need to validate it separately

invokeai/app/services/invocation_services.py

@@ -53,11 +53,11 @@ class InvocationServices:
         model_images: "ModelImageFileStorageBase",
         model_manager: "ModelManagerServiceBase",
         download_queue: "DownloadQueueServiceBase",
+        performance_statistics: "InvocationStatsServiceBase",
         session_queue: "SessionQueueBase",
         session_processor: "SessionProcessorBase",
         invocation_cache: "InvocationCacheBase",
         names: "NameServiceBase",
-        performance_statistics: "InvocationStatsServiceBase",
         urls: "UrlServiceBase",
         workflow_records: "WorkflowRecordsStorageBase",
         tensors: "ObjectSerializerBase[torch.Tensor]",
@@ -77,11 +77,11 @@ class InvocationServices:
         self.model_images = model_images
         self.model_manager = model_manager
         self.download_queue = download_queue
+        self.performance_statistics = performance_statistics
         self.session_queue = session_queue
         self.session_processor = session_processor
         self.invocation_cache = invocation_cache
         self.names = names
-        self.performance_statistics = performance_statistics
         self.urls = urls
         self.workflow_records = workflow_records
         self.tensors = tensors

invokeai/app/services/invocation_stats/invocation_stats_default.py

@@ -74,9 +74,9 @@ class InvocationStatsService(InvocationStatsServiceBase):
             )
             self._stats[graph_execution_state_id].add_node_execution_stats(node_stats)
 
-    def reset_stats(self, graph_execution_state_id: str):
-        self._stats.pop(graph_execution_state_id)
-        self._cache_stats.pop(graph_execution_state_id)
+    def reset_stats(self):
+        self._stats = {}
+        self._cache_stats = {}
 
     def get_stats(self, graph_execution_state_id: str) -> InvocationStatsSummary:
         graph_stats_summary = self._get_graph_summary(graph_execution_state_id)

invokeai/app/services/model_install/model_install_default.py

@@ -284,14 +284,9 @@ class ModelInstallService(ModelInstallServiceBase):
         unfinished_jobs = [x for x in self._install_jobs if not x.in_terminal_state]
         self._install_jobs = unfinished_jobs
 
-    def _migrate_yaml(self, rename_yaml: Optional[bool] = True, overwrite_db: Optional[bool] = False) -> None:
+    def _migrate_yaml(self) -> None:
         db_models = self.record_store.all_models()
 
-        if overwrite_db:
-            for model in db_models:
-                self.record_store.del_model(model.key)
-            db_models = self.record_store.all_models()
-
         legacy_models_yaml_path = (
             self._app_config.legacy_models_yaml_path or self._app_config.root_path / "configs" / "models.yaml"
         )
@@ -341,8 +336,7 @@ class ModelInstallService(ModelInstallServiceBase):
                         self._logger.warning(f"Model at {model_path} could not be migrated: {e}")
 
             # Rename `models.yaml` to `models.yaml.bak` to prevent re-migration
-            if rename_yaml:
-                legacy_models_yaml_path.rename(legacy_models_yaml_path.with_suffix(".yaml.bak"))
+            legacy_models_yaml_path.rename(legacy_models_yaml_path.with_suffix(".yaml.bak"))
 
         # Unset the path - we are done with it either way
         self._app_config.legacy_models_yaml_path = None

invokeai/app/services/model_load/model_load_base.py

@@ -33,11 +33,6 @@ class ModelLoadServiceBase(ABC):
     def convert_cache(self) -> ModelConvertCacheBase:
         """Return the checkpoint convert cache used by this loader."""
 
-    @property
-    @abstractmethod
-    def gpu_count(self) -> int:
-        """Return the number of GPUs we are configured to use."""
-
     @abstractmethod
     def load_model_from_path(
         self, model_path: Path, loader: Optional[Callable[[Path], AnyModel]] = None

invokeai/app/services/model_load/model_load_default.py

@@ -46,7 +46,6 @@ class ModelLoadService(ModelLoadServiceBase):
         self._registry = registry
 
     def start(self, invoker: Invoker) -> None:
-        """Start the service."""
         self._invoker = invoker
 
     @property
@@ -54,11 +53,6 @@ class ModelLoadService(ModelLoadServiceBase):
         """Return the RAM cache used by this loader."""
         return self._ram_cache
 
-    @property
-    def gpu_count(self) -> int:
-        """Return the number of GPUs available for our uses."""
-        return len(self._ram_cache.execution_devices)
-
     @property
     def convert_cache(self) -> ModelConvertCacheBase:
         """Return the checkpoint convert cache used by this loader."""

invokeai/app/services/model_manager/model_manager_base.py

@@ -1,7 +1,6 @@
 # Copyright (c) 2023 Lincoln D. Stein and the InvokeAI Team
 
 from abc import ABC, abstractmethod
-from typing import Optional, Set
 
 import torch
 from typing_extensions import Self
@@ -32,7 +31,7 @@ class ModelManagerServiceBase(ABC):
         model_record_service: ModelRecordServiceBase,
         download_queue: DownloadQueueServiceBase,
         events: EventServiceBase,
-        execution_devices: Optional[Set[torch.device]] = None,
+        execution_device: torch.device,
     ) -> Self:
         """
         Construct the model manager service instance.

invokeai/app/services/model_manager/model_manager_default.py

@@ -1,10 +1,14 @@
 # Copyright (c) 2023 Lincoln D. Stein and the InvokeAI Team
 """Implementation of ModelManagerServiceBase."""
 
+from typing import Optional
+
+import torch
 from typing_extensions import Self
 
 from invokeai.app.services.invoker import Invoker
 from invokeai.backend.model_manager.load import ModelCache, ModelConvertCache, ModelLoaderRegistry
+from invokeai.backend.util.devices import TorchDevice
 from invokeai.backend.util.logging import InvokeAILogger
 
 from ..config import InvokeAIAppConfig
@@ -65,6 +69,7 @@ class ModelManagerService(ModelManagerServiceBase):
         model_record_service: ModelRecordServiceBase,
         download_queue: DownloadQueueServiceBase,
         events: EventServiceBase,
+        execution_device: Optional[torch.device] = None,
     ) -> Self:
         """
         Construct the model manager service instance.
@@ -77,7 +82,9 @@ class ModelManagerService(ModelManagerServiceBase):
         ram_cache = ModelCache(
             max_cache_size=app_config.ram,
             max_vram_cache_size=app_config.vram,
+            lazy_offloading=app_config.lazy_offload,
             logger=logger,
+            execution_device=execution_device or TorchDevice.choose_torch_device(),
         )
         convert_cache = ModelConvertCache(cache_path=app_config.convert_cache_path, max_size=app_config.convert_cache)
         loader = ModelLoadService(

invokeai/app/services/object_serializer/object_serializer_disk.py

@@ -1,6 +1,5 @@
 import shutil
 import tempfile
-import threading
 import typing
 from pathlib import Path
 from typing import TYPE_CHECKING, Optional, TypeVar
@@ -10,7 +9,6 @@ import torch
 from invokeai.app.services.object_serializer.object_serializer_base import ObjectSerializerBase
 from invokeai.app.services.object_serializer.object_serializer_common import ObjectNotFoundError
 from invokeai.app.util.misc import uuid_string
-from invokeai.backend.util.devices import TorchDevice
 
 if TYPE_CHECKING:
     from invokeai.app.services.invoker import Invoker
@@ -72,10 +70,7 @@ class ObjectSerializerDisk(ObjectSerializerBase[T]):
         return self._output_dir / name
 
     def _new_name(self) -> str:
-        tid = threading.current_thread().ident
-        # Add tid to the object name because uuid4 not thread-safe on windows
-        # See https://stackoverflow.com/questions/2759644/python-multiprocessing-doesnt-play-nicely-with-uuid-uuid4
-        return f"{self._obj_class_name}_{tid}-{uuid_string()}"
+        return f"{self._obj_class_name}_{uuid_string()}"
 
     def _tempdir_cleanup(self) -> None:
         """Calls `cleanup` on the temporary directory, if it exists."""

invokeai/app/services/session_processor/session_processor_default.py

@@ -1,9 +1,8 @@
 import traceback
 from contextlib import suppress
-from queue import Queue
-from threading import BoundedSemaphore, Lock, Thread
+from threading import BoundedSemaphore, Thread
 from threading import Event as ThreadEvent
-from typing import Optional, Set
+from typing import Optional
 
 from invokeai.app.invocations.baseinvocation import BaseInvocation, BaseInvocationOutput
 from invokeai.app.services.events.events_common import (
@@ -27,7 +26,6 @@ from invokeai.app.services.session_queue.session_queue_common import SessionQueu
 from invokeai.app.services.shared.graph import NodeInputError
 from invokeai.app.services.shared.invocation_context import InvocationContextData, build_invocation_context
 from invokeai.app.util.profiler import Profiler
-from invokeai.backend.util.devices import TorchDevice
 
 from ..invoker import Invoker
 from .session_processor_base import InvocationServices, SessionProcessorBase, SessionRunnerBase
@@ -59,11 +57,8 @@ class DefaultSessionRunner(SessionRunnerBase):
         self._on_after_run_node_callbacks = on_after_run_node_callbacks or []
         self._on_node_error_callbacks = on_node_error_callbacks or []
         self._on_after_run_session_callbacks = on_after_run_session_callbacks or []
-        self._process_lock = Lock()
 
-    def start(
-        self, services: InvocationServices, cancel_event: ThreadEvent, profiler: Optional[Profiler] = None
-    ) -> None:
+    def start(self, services: InvocationServices, cancel_event: ThreadEvent, profiler: Optional[Profiler] = None):
         self._services = services
         self._cancel_event = cancel_event
         self._profiler = profiler
@@ -81,8 +76,7 @@ class DefaultSessionRunner(SessionRunnerBase):
         # Loop over invocations until the session is complete or canceled
         while True:
             try:
-                with self._process_lock:
-                    invocation = queue_item.session.next()
+                invocation = queue_item.session.next()
             # Anything other than a `NodeInputError` is handled as a processor error
             except NodeInputError as e:
                 error_type = e.__class__.__name__
@@ -114,7 +108,7 @@ class DefaultSessionRunner(SessionRunnerBase):
 
         self._on_after_run_session(queue_item=queue_item)
 
-    def run_node(self, invocation: BaseInvocation, queue_item: SessionQueueItem) -> None:
+    def run_node(self, invocation: BaseInvocation, queue_item: SessionQueueItem):
         try:
             # Any unhandled exception in this scope is an invocation error & will fail the graph
             with self._services.performance_statistics.collect_stats(invocation, queue_item.session_id):
@@ -216,7 +210,7 @@ class DefaultSessionRunner(SessionRunnerBase):
             # we don't care about that - suppress the error.
             with suppress(GESStatsNotFoundError):
                 self._services.performance_statistics.log_stats(queue_item.session.id)
-                self._services.performance_statistics.reset_stats(queue_item.session.id)
+                self._services.performance_statistics.reset_stats()
 
             for callback in self._on_after_run_session_callbacks:
                 callback(queue_item=queue_item)
@@ -330,7 +324,7 @@ class DefaultSessionProcessor(SessionProcessorBase):
 
     def start(self, invoker: Invoker) -> None:
         self._invoker: Invoker = invoker
-        self._active_queue_items: Set[SessionQueueItem] = set()
+        self._queue_item: Optional[SessionQueueItem] = None
         self._invocation: Optional[BaseInvocation] = None
 
         self._resume_event = ThreadEvent()
@@ -356,14 +350,7 @@ class DefaultSessionProcessor(SessionProcessorBase):
             else None
         )
 
-        self._worker_thread_count = self._invoker.services.configuration.max_threads or len(
-            TorchDevice.execution_devices()
-        )
-
-        self._session_worker_queue: Queue[SessionQueueItem] = Queue()
-
         self.session_runner.start(services=invoker.services, cancel_event=self._cancel_event, profiler=self._profiler)
-        # Session processor - singlethreaded
         self._thread = Thread(
             name="session_processor",
             target=self._process,
@@ -376,16 +363,6 @@ class DefaultSessionProcessor(SessionProcessorBase):
         )
         self._thread.start()
 
-        # Session processor workers - multithreaded
-        self._invoker.services.logger.debug(f"Starting {self._worker_thread_count} session processing threads.")
-        for _i in range(0, self._worker_thread_count):
-            worker = Thread(
-                name="session_worker",
-                target=self._process_next_session,
-                daemon=True,
-            )
-            worker.start()
-
     def stop(self, *args, **kwargs) -> None:
         self._stop_event.set()
 
@@ -393,7 +370,7 @@ class DefaultSessionProcessor(SessionProcessorBase):
         self._poll_now_event.set()
 
     async def _on_queue_cleared(self, event: FastAPIEvent[QueueClearedEvent]) -> None:
-        if any(item.queue_id == event[1].queue_id for item in self._active_queue_items):
+        if self._queue_item and self._queue_item.queue_id == event[1].queue_id:
             self._cancel_event.set()
             self._poll_now()
 
@@ -401,7 +378,7 @@ class DefaultSessionProcessor(SessionProcessorBase):
         self._poll_now()
 
     async def _on_queue_item_status_changed(self, event: FastAPIEvent[QueueItemStatusChangedEvent]) -> None:
-        if self._active_queue_items and event[1].status in ["completed", "failed", "canceled"]:
+        if self._queue_item and event[1].status in ["completed", "failed", "canceled"]:
             # When the queue item is canceled via HTTP, the queue item status is set to `"canceled"` and this event is
             # emitted. We need to respond to this event and stop graph execution. This is done by setting the cancel
             # event, which the session runner checks between invocations. If set, the session runner loop is broken.
@@ -426,7 +403,7 @@ class DefaultSessionProcessor(SessionProcessorBase):
     def get_status(self) -> SessionProcessorStatus:
         return SessionProcessorStatus(
             is_started=self._resume_event.is_set(),
-            is_processing=len(self._active_queue_items) > 0,
+            is_processing=self._queue_item is not None,
         )
 
     def _process(
@@ -451,22 +428,30 @@ class DefaultSessionProcessor(SessionProcessorBase):
                     resume_event.wait()
 
                     # Get the next session to process
-                    queue_item = self._invoker.services.session_queue.dequeue()
+                    self._queue_item = self._invoker.services.session_queue.dequeue()
 
-                    if queue_item is None:
+                    if self._queue_item is None:
                         # The queue was empty, wait for next polling interval or event to try again
                         self._invoker.services.logger.debug("Waiting for next polling interval or event")
                         poll_now_event.wait(self._polling_interval)
                         continue
 
-                    self._session_worker_queue.put(queue_item)
-                    self._invoker.services.logger.debug(f"Scheduling queue item {queue_item.item_id} to run")
+                    self._invoker.services.logger.debug(f"Executing queue item {self._queue_item.item_id}")
                     cancel_event.clear()
 
                     # Run the graph
-                    # self.session_runner.run(queue_item=self._queue_item)
+                    self.session_runner.run(queue_item=self._queue_item)
 
-                except Exception:
+                except Exception as e:
+                    error_type = e.__class__.__name__
+                    error_message = str(e)
+                    error_traceback = traceback.format_exc()
+                    self._on_non_fatal_processor_error(
+                        queue_item=self._queue_item,
+                        error_type=error_type,
+                        error_message=error_message,
+                        error_traceback=error_traceback,
+                    )
                     # Wait for next polling interval or event to try again
                     poll_now_event.wait(self._polling_interval)
                     continue
@@ -481,25 +466,9 @@ class DefaultSessionProcessor(SessionProcessorBase):
         finally:
             stop_event.clear()
             poll_now_event.clear()
+            self._queue_item = None
             self._thread_semaphore.release()
 
-    def _process_next_session(self) -> None:
-        while True:
-            self._resume_event.wait()
-            queue_item = self._session_worker_queue.get()
-            if queue_item.status == "canceled":
-                continue
-            try:
-                self._active_queue_items.add(queue_item)
-                # reserve a GPU for this session - may block
-                with self._invoker.services.model_manager.load.ram_cache.reserve_execution_device():
-                    # Run the session on the reserved GPU
-                    self.session_runner.run(queue_item=queue_item)
-            except Exception:
-                continue
-            finally:
-                self._active_queue_items.remove(queue_item)
-
     def _on_non_fatal_processor_error(
         self,
         queue_item: Optional[SessionQueueItem],

invokeai/app/services/session_queue/session_queue_common.py

@@ -236,9 +236,6 @@ class SessionQueueItemWithoutGraph(BaseModel):
         }
     )
 
-    def __hash__(self) -> int:
-        return self.item_id
-
 
 class SessionQueueItemDTO(SessionQueueItemWithoutGraph):
     pass

invokeai/app/services/session_queue/session_queue_sqlite.py

@@ -37,14 +37,10 @@ class SqliteSessionQueue(SessionQueueBase):
     def start(self, invoker: Invoker) -> None:
         self.__invoker = invoker
         self._set_in_progress_to_canceled()
-        if self.__invoker.services.configuration.clear_queue_on_startup:
-            clear_result = self.clear(DEFAULT_QUEUE_ID)
-            if clear_result.deleted > 0:
-                self.__invoker.services.logger.info(f"Cleared all {clear_result.deleted} queue items")
-        else:
-            prune_result = self.prune(DEFAULT_QUEUE_ID)
-            if prune_result.deleted > 0:
-                self.__invoker.services.logger.info(f"Pruned {prune_result.deleted} finished queue items")
+        prune_result = self.prune(DEFAULT_QUEUE_ID)
+
+        if prune_result.deleted > 0:
+            self.__invoker.services.logger.info(f"Pruned {prune_result.deleted} finished queue items")
 
     def __init__(self, db: SqliteDatabase) -> None:
         super().__init__()

invokeai/app/services/shared/graph.py

@@ -652,7 +652,7 @@ class Graph(BaseModel):
         output_fields = [get_input_field(self.get_node(e.node_id), e.field) for e in outputs]
 
         # Input type must be a list
-        if get_origin(input_field) is not list:
+        if get_origin(input_field) != list:
             return False
 
         # Validate that all outputs match the input type

invokeai/app/services/shared/invocation_context.py

@@ -2,7 +2,6 @@ from dataclasses import dataclass
 from pathlib import Path
 from typing import TYPE_CHECKING, Callable, Optional, Union
 
-import torch
 from PIL.Image import Image
 from pydantic.networks import AnyHttpUrl
 from torch import Tensor
@@ -27,13 +26,11 @@ from invokeai.backend.model_manager.config import (
 from invokeai.backend.model_manager.load.load_base import LoadedModel, LoadedModelWithoutConfig
 from invokeai.backend.stable_diffusion.diffusers_pipeline import PipelineIntermediateState
 from invokeai.backend.stable_diffusion.diffusion.conditioning_data import ConditioningFieldData
-from invokeai.backend.util.devices import TorchDevice
 
 if TYPE_CHECKING:
     from invokeai.app.invocations.baseinvocation import BaseInvocation
     from invokeai.app.invocations.model import ModelIdentifierField
     from invokeai.app.services.session_queue.session_queue_common import SessionQueueItem
-    from invokeai.backend.model_manager.load.model_cache.model_cache_base import ModelCacheBase
 
 """
 The InvocationContext provides access to various services and data about the current invocation.
@@ -326,6 +323,7 @@ class ConditioningInterface(InvocationContextInterface):
         Returns:
             The loaded conditioning data.
         """
+
         return self._services.conditioning.load(name)
 
 
@@ -559,28 +557,6 @@ class UtilInterface(InvocationContextInterface):
             is_canceled=self.is_canceled,
         )
 
-    def torch_device(self) -> torch.device:
-        """
-        Return a torch device to use in the current invocation.
-
-        Returns:
-            A torch.device not currently in use by the system.
-        """
-        ram_cache: "ModelCacheBase[AnyModel]" = self._services.model_manager.load.ram_cache
-        return ram_cache.get_execution_device()
-
-    def torch_dtype(self, device: Optional[torch.device] = None) -> torch.dtype:
-        """
-        Return a precision type to use with the current invocation and torch device.
-
-        Args:
-            device: Optional device.
-
-        Returns:
-            A torch.dtype suited for the current device.
-        """
-        return TorchDevice.choose_torch_dtype(device)
-
 
 class InvocationContext:
     """Provides access to various services and data for the current invocation.

invokeai/app/util/controlnet_utils.py

@@ -289,7 +289,7 @@ def prepare_control_image(
     width: int,
     height: int,
     num_channels: int = 3,
-    device: str = "cuda",
+    device: str | torch.device = "cuda",
     dtype: torch.dtype = torch.float16,
     control_mode: CONTROLNET_MODE_VALUES = "balanced",
     resize_mode: CONTROLNET_RESIZE_VALUES = "just_resize_simple",
@@ -304,7 +304,7 @@ def prepare_control_image(
         num_channels (int, optional): The target number of image channels. This is achieved by converting the input
             image to RGB, then naively taking the first `num_channels` channels. The primary use case is converting a
             RGB image to a single-channel grayscale image. Raises if `num_channels` cannot be achieved. Defaults to 3.
-        device (str, optional): The target device for the output image. Defaults to "cuda".
+        device (str | torch.Device, optional): The target device for the output image. Defaults to "cuda".
         dtype (_type_, optional): The dtype for the output image. Defaults to torch.float16.
         do_classifier_free_guidance (bool, optional): If True, repeat the output image along the batch dimension.
             Defaults to True.

invokeai/backend/model_manager/config.py

@@ -25,7 +25,6 @@ from enum import Enum
 from typing import Literal, Optional, Type, TypeAlias, Union
 
 import torch
-from diffusers.configuration_utils import ConfigMixin
 from diffusers.models.modeling_utils import ModelMixin
 from pydantic import BaseModel, ConfigDict, Discriminator, Field, Tag, TypeAdapter
 from typing_extensions import Annotated, Any, Dict
@@ -38,7 +37,7 @@ from ..raw_model import RawModel
 
 # ModelMixin is the base class for all diffusers and transformers models
 # RawModel is the InvokeAI wrapper class for ip_adapters, loras, textual_inversion and onnx runtime
-AnyModel = Union[ConfigMixin, ModelMixin, RawModel, torch.nn.Module, Dict[str, torch.Tensor]]
+AnyModel = Union[ModelMixin, RawModel, torch.nn.Module, Dict[str, torch.Tensor]]
 
 
 class InvalidModelConfigException(Exception):
@@ -178,7 +177,6 @@ class ModelConfigBase(BaseModel):
 
     @staticmethod
     def json_schema_extra(schema: dict[str, Any], model_class: Type[BaseModel]) -> None:
-        """Extend the pydantic schema from a json."""
         schema["required"].extend(["key", "type", "format"])
 
     model_config = ConfigDict(validate_assignment=True, json_schema_extra=json_schema_extra)
@@ -445,7 +443,7 @@ class ModelConfigFactory(object):
             model = dest_class.model_validate(model_data)
         else:
             # mypy doesn't typecheck TypeAdapters well?
-            model = AnyModelConfigValidator.validate_python(model_data)
+            model = AnyModelConfigValidator.validate_python(model_data)  # type: ignore
         assert model is not None
         if key:
             model.key = key

invokeai/backend/model_manager/load/load_base.py

@@ -65,7 +65,8 @@ class LoadedModelWithoutConfig:
 
     def __enter__(self) -> AnyModel:
         """Context entry."""
-        return self._locker.lock()
+        self._locker.lock()
+        return self.model
 
     def __exit__(self, *args: Any, **kwargs: Any) -> None:
         """Context exit."""

invokeai/backend/model_manager/load/model_cache/model_cache_base.py

@@ -8,10 +8,9 @@ model will be cleared and (re)loaded from disk when next needed.
 """
 
 from abc import ABC, abstractmethod
-from contextlib import contextmanager
 from dataclasses import dataclass, field
 from logging import Logger
-from typing import Dict, Generator, Generic, Optional, Set, TypeVar
+from typing import Dict, Generic, Optional, TypeVar
 
 import torch
 
@@ -52,13 +51,44 @@ class CacheRecord(Generic[T]):
     Elements of the cache:
 
     key: Unique key for each model, same as used in the models database.
-    model: Read-only copy of the model *without weights* residing in the "meta device"
+    model: Model in memory.
+    state_dict: A read-only copy of the model's state dict in RAM. It will be
+                used as a template for creating a copy in the VRAM.
     size: Size of the model
+    loaded: True if the model's state dict is currently in VRAM
+
+    Before a model is executed, the state_dict template is copied into VRAM,
+    and then injected into the model. When the model is finished, the VRAM
+    copy of the state dict is deleted, and the RAM version is reinjected
+    into the model.
+
+    The state_dict should be treated as a read-only attribute. Do not attempt
+    to patch or otherwise modify it. Instead, patch the copy of the state_dict
+    after it is loaded into the execution device (e.g. CUDA) using the `LoadedModel`
+    context manager call `model_on_device()`.
     """
 
     key: str
-    size: int
     model: T
+    device: torch.device
+    state_dict: Optional[Dict[str, torch.Tensor]]
+    size: int
+    loaded: bool = False
+    _locks: int = 0
+
+    def lock(self) -> None:
+        """Lock this record."""
+        self._locks += 1
+
+    def unlock(self) -> None:
+        """Unlock this record."""
+        self._locks -= 1
+        assert self._locks >= 0
+
+    @property
+    def locked(self) -> bool:
+        """Return true if record is locked."""
+        return self._locks > 0
 
 
 @dataclass
@@ -85,27 +115,14 @@ class ModelCacheBase(ABC, Generic[T]):
 
     @property
     @abstractmethod
-    def execution_devices(self) -> Set[torch.device]:
-        """Return the set of available execution devices."""
+    def execution_device(self) -> torch.device:
+        """Return the exection device (e.g. "cuda" for VRAM)."""
         pass
 
-    @contextmanager
+    @property
     @abstractmethod
-    def reserve_execution_device(self, timeout: int = 0) -> Generator[torch.device, None, None]:
-        """Reserve an execution device (GPU) under the current thread id."""
-        pass
-
-    @abstractmethod
-    def get_execution_device(self) -> torch.device:
-        """
-        Return an execution device that has been reserved for current thread.
-
-        Note that reservations are done using the current thread's TID.
-        It might be better to do this using the session ID, but that involves
-        too many detailed changes to model manager calls.
-
-        May generate a ValueError if no GPU has been reserved.
-        """
+    def lazy_offloading(self) -> bool:
+        """Return true if the cache is configured to lazily offload models in VRAM."""
         pass
 
     @property
@@ -114,6 +131,16 @@ class ModelCacheBase(ABC, Generic[T]):
         """Return true if the cache is configured to lazily offload models in VRAM."""
         pass
 
+    @abstractmethod
+    def offload_unlocked_models(self, size_required: int) -> None:
+        """Offload from VRAM any models not actively in use."""
+        pass
+
+    @abstractmethod
+    def move_model_to_device(self, cache_entry: CacheRecord[AnyModel], target_device: torch.device) -> None:
+        """Move model into the indicated device."""
+        pass
+
     @property
     @abstractmethod
     def stats(self) -> Optional[CacheStats]:
@@ -175,11 +202,6 @@ class ModelCacheBase(ABC, Generic[T]):
         """Return true if the model identified by key and submodel_type is in the cache."""
         pass
 
-    @abstractmethod
-    def model_to_device(self, cache_entry: CacheRecord[AnyModel], target_device: torch.device) -> AnyModel:
-        """Move a copy of the model into the indicated device and return it."""
-        pass
-
     @abstractmethod
     def cache_size(self) -> int:
         """Get the total size of the models currently cached."""

invokeai/backend/model_manager/load/model_cache/model_cache_default.py

@@ -18,19 +18,17 @@ context. Use like this:
 
 """
 
-import copy
 import gc
-import sys
-import threading
-from contextlib import contextmanager, suppress
+import math
+import time
+from contextlib import suppress
 from logging import Logger
-from threading import BoundedSemaphore
-from typing import Dict, Generator, List, Optional, Set
+from typing import Dict, List, Optional
 
 import torch
 
 from invokeai.backend.model_manager import AnyModel, SubModelType
-from invokeai.backend.model_manager.load.memory_snapshot import MemorySnapshot
+from invokeai.backend.model_manager.load.memory_snapshot import MemorySnapshot, get_pretty_snapshot_diff
 from invokeai.backend.model_manager.load.model_util import calc_model_size_by_data
 from invokeai.backend.util.devices import TorchDevice
 from invokeai.backend.util.logging import InvokeAILogger
@@ -41,7 +39,9 @@ from .model_locker import ModelLocker
 # Maximum size of the cache, in gigs
 # Default is roughly enough to hold three fp16 diffusers models in RAM simultaneously
 DEFAULT_MAX_CACHE_SIZE = 6.0
-DEFAULT_MAX_VRAM_CACHE_SIZE = 0.25
+
+# amount of GPU memory to hold in reserve for use by generations (GB)
+DEFAULT_MAX_VRAM_CACHE_SIZE = 2.75
 
 # actual size of a gig
 GIG = 1073741824
@@ -57,8 +57,12 @@ class ModelCache(ModelCacheBase[AnyModel]):
         self,
         max_cache_size: float = DEFAULT_MAX_CACHE_SIZE,
         max_vram_cache_size: float = DEFAULT_MAX_VRAM_CACHE_SIZE,
+        execution_device: torch.device = torch.device("cuda"),
         storage_device: torch.device = torch.device("cpu"),
         precision: torch.dtype = torch.float16,
+        sequential_offload: bool = False,
+        lazy_offloading: bool = True,
+        sha_chunksize: int = 16777216,
         log_memory_usage: bool = False,
         logger: Optional[Logger] = None,
     ):
@@ -66,19 +70,23 @@ class ModelCache(ModelCacheBase[AnyModel]):
         Initialize the model RAM cache.
 
         :param max_cache_size: Maximum size of the RAM cache [6.0 GB]
+        :param execution_device: Torch device to load active model into [torch.device('cuda')]
         :param storage_device: Torch device to save inactive model in [torch.device('cpu')]
         :param precision: Precision for loaded models [torch.float16]
+        :param lazy_offloading: Keep model in VRAM until another model needs to be loaded
         :param sequential_offload: Conserve VRAM by loading and unloading each stage of the pipeline sequentially
         :param log_memory_usage: If True, a memory snapshot will be captured before and after every model cache
             operation, and the result will be logged (at debug level). There is a time cost to capturing the memory
             snapshots, so it is recommended to disable this feature unless you are actively inspecting the model cache's
             behaviour.
         """
+        # allow lazy offloading only when vram cache enabled
+        self._lazy_offloading = lazy_offloading and max_vram_cache_size > 0
         self._precision: torch.dtype = precision
         self._max_cache_size: float = max_cache_size
         self._max_vram_cache_size: float = max_vram_cache_size
+        self._execution_device: torch.device = execution_device
         self._storage_device: torch.device = storage_device
-        self._ram_lock = threading.Lock()
         self._logger = logger or InvokeAILogger.get_logger(self.__class__.__name__)
         self._log_memory_usage = log_memory_usage
         self._stats: Optional[CacheStats] = None
@@ -86,87 +94,25 @@ class ModelCache(ModelCacheBase[AnyModel]):
         self._cached_models: Dict[str, CacheRecord[AnyModel]] = {}
         self._cache_stack: List[str] = []
 
-        # device to thread id
-        self._device_lock = threading.Lock()
-        self._execution_devices: Dict[torch.device, int] = {x: 0 for x in TorchDevice.execution_devices()}
-        self._free_execution_device = BoundedSemaphore(len(self._execution_devices))
-
-        self.logger.info(
-            f"Using rendering device(s): {', '.join(sorted([str(x) for x in self._execution_devices.keys()]))}"
-        )
-
     @property
     def logger(self) -> Logger:
         """Return the logger used by the cache."""
         return self._logger
 
+    @property
+    def lazy_offloading(self) -> bool:
+        """Return true if the cache is configured to lazily offload models in VRAM."""
+        return self._lazy_offloading
+
     @property
     def storage_device(self) -> torch.device:
         """Return the storage device (e.g. "CPU" for RAM)."""
         return self._storage_device
 
     @property
-    def execution_devices(self) -> Set[torch.device]:
-        """Return the set of available execution devices."""
-        devices = self._execution_devices.keys()
-        return set(devices)
-
-    def get_execution_device(self) -> torch.device:
-        """
-        Return an execution device that has been reserved for current thread.
-
-        Note that reservations are done using the current thread's TID.
-        It would be better to do this using the session ID, but that involves
-        too many detailed changes to model manager calls.
-
-        May generate a ValueError if no GPU has been reserved.
-        """
-        current_thread = threading.current_thread().ident
-        assert current_thread is not None
-        assigned = [x for x, tid in self._execution_devices.items() if current_thread == tid]
-        if not assigned:
-            raise ValueError(f"No GPU has been reserved for the use of thread {current_thread}")
-        return assigned[0]
-
-    @contextmanager
-    def reserve_execution_device(self, timeout: Optional[int] = None) -> Generator[torch.device, None, None]:
-        """Reserve an execution device (e.g. GPU) for exclusive use by a generation thread.
-
-        Note that the reservation is done using the current thread's TID.
-        It would be better to do this using the session ID, but that involves
-        too many detailed changes to model manager calls.
-        """
-        device = None
-        with self._device_lock:
-            current_thread = threading.current_thread().ident
-            assert current_thread is not None
-
-            # look for a device that has already been assigned to this thread
-            assigned = [x for x, tid in self._execution_devices.items() if current_thread == tid]
-            if assigned:
-                device = assigned[0]
-
-        # no device already assigned. Get one.
-        if device is None:
-            self._free_execution_device.acquire(timeout=timeout)
-            with self._device_lock:
-                free_device = [x for x, tid in self._execution_devices.items() if tid == 0]
-                self._execution_devices[free_device[0]] = current_thread
-                device = free_device[0]
-
-        # we are outside the lock region now
-        self.logger.info(f"{current_thread} Reserved torch device {device}")
-
-        # Tell TorchDevice to use this object to get the torch device.
-        TorchDevice.set_model_cache(self)
-        try:
-            yield device
-        finally:
-            with self._device_lock:
-                self.logger.info(f"{current_thread} Released torch device {device}")
-                self._execution_devices[device] = 0
-                self._free_execution_device.release()
-                torch.cuda.empty_cache()
+    def execution_device(self) -> torch.device:
+        """Return the exection device (e.g. "cuda" for VRAM)."""
+        return self._execution_device
 
     @property
     def max_cache_size(self) -> float:
@@ -211,16 +157,16 @@ class ModelCache(ModelCacheBase[AnyModel]):
         submodel_type: Optional[SubModelType] = None,
     ) -> None:
         """Store model under key and optional submodel_type."""
-        with self._ram_lock:
-            key = self._make_cache_key(key, submodel_type)
-            if key in self._cached_models:
-                return
-            size = calc_model_size_by_data(model)
-            self.make_room(size)
+        key = self._make_cache_key(key, submodel_type)
+        if key in self._cached_models:
+            return
+        size = calc_model_size_by_data(model)
+        self.make_room(size)
 
-            cache_record = CacheRecord(key=key, model=model, size=size)
-            self._cached_models[key] = cache_record
-            self._cache_stack.append(key)
+        state_dict = model.state_dict() if isinstance(model, torch.nn.Module) else None
+        cache_record = CacheRecord(key=key, model=model, device=self.storage_device, state_dict=state_dict, size=size)
+        self._cached_models[key] = cache_record
+        self._cache_stack.append(key)
 
     def get(
         self,
@@ -238,37 +184,36 @@ class ModelCache(ModelCacheBase[AnyModel]):
 
         This may raise an IndexError if the model is not in the cache.
         """
-        with self._ram_lock:
-            key = self._make_cache_key(key, submodel_type)
-            if key in self._cached_models:
-                if self.stats:
-                    self.stats.hits += 1
-            else:
-                if self.stats:
-                    self.stats.misses += 1
-                raise IndexError(f"The model with key {key} is not in the cache.")
-
-            cache_entry = self._cached_models[key]
-
-            # more stats
+        key = self._make_cache_key(key, submodel_type)
+        if key in self._cached_models:
             if self.stats:
-                stats_name = stats_name or key
-                self.stats.cache_size = int(self._max_cache_size * GIG)
-                self.stats.high_watermark = max(self.stats.high_watermark, self.cache_size())
-                self.stats.in_cache = len(self._cached_models)
-                self.stats.loaded_model_sizes[stats_name] = max(
-                    self.stats.loaded_model_sizes.get(stats_name, 0), cache_entry.size
-                )
+                self.stats.hits += 1
+        else:
+            if self.stats:
+                self.stats.misses += 1
+            raise IndexError(f"The model with key {key} is not in the cache.")
 
-            # this moves the entry to the top (right end) of the stack
-            with suppress(Exception):
-                self._cache_stack.remove(key)
-            self._cache_stack.append(key)
-            return ModelLocker(
-                cache=self,
-                cache_entry=cache_entry,
+        cache_entry = self._cached_models[key]
+
+        # more stats
+        if self.stats:
+            stats_name = stats_name or key
+            self.stats.cache_size = int(self._max_cache_size * GIG)
+            self.stats.high_watermark = max(self.stats.high_watermark, self.cache_size())
+            self.stats.in_cache = len(self._cached_models)
+            self.stats.loaded_model_sizes[stats_name] = max(
+                self.stats.loaded_model_sizes.get(stats_name, 0), cache_entry.size
             )
 
+        # this moves the entry to the top (right end) of the stack
+        with suppress(Exception):
+            self._cache_stack.remove(key)
+        self._cache_stack.append(key)
+        return ModelLocker(
+            cache=self,
+            cache_entry=cache_entry,
+        )
+
     def _capture_memory_snapshot(self) -> Optional[MemorySnapshot]:
         if self._log_memory_usage:
             return MemorySnapshot.capture()
@@ -280,34 +225,127 @@ class ModelCache(ModelCacheBase[AnyModel]):
         else:
             return model_key
 
-    def model_to_device(self, cache_entry: CacheRecord[AnyModel], target_device: torch.device) -> AnyModel:
-        """Move a copy of the model into the indicated device and return it.
+    def offload_unlocked_models(self, size_required: int) -> None:
+        """Move any unused models from VRAM."""
+        reserved = self._max_vram_cache_size * GIG
+        vram_in_use = torch.cuda.memory_allocated() + size_required
+        self.logger.debug(f"{(vram_in_use/GIG):.2f}GB VRAM needed for models; max allowed={(reserved/GIG):.2f}GB")
+        for _, cache_entry in sorted(self._cached_models.items(), key=lambda x: x[1].size):
+            if vram_in_use <= reserved:
+                break
+            if not cache_entry.loaded:
+                continue
+            if not cache_entry.locked:
+                self.move_model_to_device(cache_entry, self.storage_device)
+                cache_entry.loaded = False
+                vram_in_use = torch.cuda.memory_allocated() + size_required
+                self.logger.debug(
+                    f"Removing {cache_entry.key} from VRAM to free {(cache_entry.size/GIG):.2f}GB; vram free = {(torch.cuda.memory_allocated()/GIG):.2f}GB"
+                )
+
+        TorchDevice.empty_cache()
+
+    def move_model_to_device(self, cache_entry: CacheRecord[AnyModel], target_device: torch.device) -> None:
+        """Move model into the indicated device.
 
         :param cache_entry: The CacheRecord for the model
         :param target_device: The torch.device to move the model into
 
         May raise a torch.cuda.OutOfMemoryError
         """
-        with self._ram_lock:
-            self.logger.debug(f"Called to move {cache_entry.key} ({type(cache_entry.model)=}) to {target_device}")
+        self.logger.debug(f"Called to move {cache_entry.key} to {target_device}")
+        source_device = cache_entry.device
 
-            # Some models don't have a state dictionary, in which case the
-            # stored model will still reside in CPU
-            if hasattr(cache_entry.model, "to"):
-                model_in_gpu = copy.deepcopy(cache_entry.model)
-                assert hasattr(model_in_gpu, "to")
-                model_in_gpu.to(target_device)
-                return model_in_gpu
-            else:
-                return cache_entry.model  # what happens in CPU stays in CPU
+        # Note: We compare device types only so that 'cuda' == 'cuda:0'.
+        # This would need to be revised to support multi-GPU.
+        if torch.device(source_device).type == torch.device(target_device).type:
+            return
+
+        # Some models don't have a `to` method, in which case they run in RAM/CPU.
+        if not hasattr(cache_entry.model, "to"):
+            return
+
+        # This roundabout method for moving the model around is done to avoid
+        # the cost of moving the model from RAM to VRAM and then back from VRAM to RAM.
+        # When moving to VRAM, we copy (not move) each element of the state dict from
+        # RAM to a new state dict in VRAM, and then inject it into the model.
+        # This operation is slightly faster than running `to()` on the whole model.
+        #
+        # When the model needs to be removed from VRAM we simply delete the copy
+        # of the state dict in VRAM, and reinject the state dict that is cached
+        # in RAM into the model. So this operation is very fast.
+        start_model_to_time = time.time()
+        snapshot_before = self._capture_memory_snapshot()
+
+        try:
+            if cache_entry.state_dict is not None:
+                assert hasattr(cache_entry.model, "load_state_dict")
+                if target_device == self.storage_device:
+                    cache_entry.model.load_state_dict(cache_entry.state_dict, assign=True)
+                else:
+                    new_dict: Dict[str, torch.Tensor] = {}
+                    for k, v in cache_entry.state_dict.items():
+                        new_dict[k] = v.to(torch.device(target_device), copy=True, non_blocking=True)
+                    cache_entry.model.load_state_dict(new_dict, assign=True)
+            cache_entry.model.to(target_device, non_blocking=True)
+            cache_entry.device = target_device
+        except Exception as e:  # blow away cache entry
+            self._delete_cache_entry(cache_entry)
+            raise e
+
+        snapshot_after = self._capture_memory_snapshot()
+        end_model_to_time = time.time()
+        self.logger.debug(
+            f"Moved model '{cache_entry.key}' from {source_device} to"
+            f" {target_device} in {(end_model_to_time-start_model_to_time):.2f}s."
+            f"Estimated model size: {(cache_entry.size/GIG):.3f} GB."
+            f"{get_pretty_snapshot_diff(snapshot_before, snapshot_after)}"
+        )
+
+        if (
+            snapshot_before is not None
+            and snapshot_after is not None
+            and snapshot_before.vram is not None
+            and snapshot_after.vram is not None
+        ):
+            vram_change = abs(snapshot_before.vram - snapshot_after.vram)
+
+            # If the estimated model size does not match the change in VRAM, log a warning.
+            if not math.isclose(
+                vram_change,
+                cache_entry.size,
+                rel_tol=0.1,
+                abs_tol=10 * MB,
+            ):
+                self.logger.debug(
+                    f"Moving model '{cache_entry.key}' from {source_device} to"
+                    f" {target_device} caused an unexpected change in VRAM usage. The model's"
+                    " estimated size may be incorrect. Estimated model size:"
+                    f" {(cache_entry.size/GIG):.3f} GB.\n"
+                    f"{get_pretty_snapshot_diff(snapshot_before, snapshot_after)}"
+                )
 
     def print_cuda_stats(self) -> None:
         """Log CUDA diagnostics."""
         vram = "%4.2fG" % (torch.cuda.memory_allocated() / GIG)
         ram = "%4.2fG" % (self.cache_size() / GIG)
 
-        in_ram_models = len(self._cached_models)
-        self.logger.debug(f"Current VRAM/RAM usage for {in_ram_models} models: {vram}/{ram}")
+        in_ram_models = 0
+        in_vram_models = 0
+        locked_in_vram_models = 0
+        for cache_record in self._cached_models.values():
+            if hasattr(cache_record.model, "device"):
+                if cache_record.model.device == self.storage_device:
+                    in_ram_models += 1
+                else:
+                    in_vram_models += 1
+                if cache_record.locked:
+                    locked_in_vram_models += 1
+
+                self.logger.debug(
+                    f"Current VRAM/RAM usage: {vram}/{ram}; models_in_ram/models_in_vram(locked) ="
+                    f" {in_ram_models}/{in_vram_models}({locked_in_vram_models})"
+                )
 
     def make_room(self, size: int) -> None:
         """Make enough room in the cache to accommodate a new model of indicated size."""
@@ -330,14 +368,12 @@ class ModelCache(ModelCacheBase[AnyModel]):
         while current_size + bytes_needed > maximum_size and pos < len(self._cache_stack):
             model_key = self._cache_stack[pos]
             cache_entry = self._cached_models[model_key]
+            device = cache_entry.model.device if hasattr(cache_entry.model, "device") else None
+            self.logger.debug(
+                f"Model: {model_key}, locks: {cache_entry._locks}, device: {device}, loaded: {cache_entry.loaded}"
+            )
 
-            refs = sys.getrefcount(cache_entry.model)
-
-            # Expected refs:
-            # 1 from cache_entry
-            # 1 from getrefcount function
-            # 1 from onnx runtime object
-            if refs <= (3 if "onnx" in model_key else 2):
+            if not cache_entry.locked:
                 self.logger.debug(
                     f"Removing {model_key} from RAM cache to free at least {(size/GIG):.2f} GB (-{(cache_entry.size/GIG):.2f} GB)"
                 )
@@ -364,26 +400,10 @@ class ModelCache(ModelCacheBase[AnyModel]):
             if self.stats:
                 self.stats.cleared = models_cleared
             gc.collect()
+
         TorchDevice.empty_cache()
         self.logger.debug(f"After making room: cached_models={len(self._cached_models)}")
 
-    def _check_free_vram(self, target_device: torch.device, needed_size: int) -> None:
-        if target_device.type != "cuda":
-            return
-        vram_device = (  # mem_get_info() needs an indexed device
-            target_device if target_device.index is not None else torch.device(str(target_device), index=0)
-        )
-        free_mem, _ = torch.cuda.mem_get_info(torch.device(vram_device))
-        if needed_size > free_mem:
-            raise torch.cuda.OutOfMemoryError
-
     def _delete_cache_entry(self, cache_entry: CacheRecord[AnyModel]) -> None:
-        try:
-            self._cache_stack.remove(cache_entry.key)
-            del self._cached_models[cache_entry.key]
-        except ValueError:
-            pass
-
-    @staticmethod
-    def _device_name(device: torch.device) -> str:
-        return f"{device.type}:{device.index}"
+        self._cache_stack.remove(cache_entry.key)
+        del self._cached_models[cache_entry.key]

invokeai/backend/model_manager/load/model_cache/model_locker.py

@@ -10,8 +10,6 @@ from invokeai.backend.model_manager import AnyModel
 
 from .model_cache_base import CacheRecord, ModelCacheBase, ModelLockerBase
 
-MAX_GPU_WAIT = 600  # wait up to 10 minutes for a GPU to become free
-
 
 class ModelLocker(ModelLockerBase):
     """Internal class that mediates movement in and out of GPU."""
@@ -31,29 +29,33 @@ class ModelLocker(ModelLockerBase):
         """Return the model without moving it around."""
         return self._cache_entry.model
 
+    def get_state_dict(self) -> Optional[Dict[str, torch.Tensor]]:
+        """Return the state dict (if any) for the cached model."""
+        return self._cache_entry.state_dict
+
     def lock(self) -> AnyModel:
         """Move the model into the execution device (GPU) and lock it."""
+        self._cache_entry.lock()
         try:
-            device = self._cache.get_execution_device()
-            model_on_device = self._cache.model_to_device(self._cache_entry, device)
-            self._cache.logger.debug(f"Moved {self._cache_entry.key} to {device}")
+            if self._cache.lazy_offloading:
+                self._cache.offload_unlocked_models(self._cache_entry.size)
+            self._cache.move_model_to_device(self._cache_entry, self._cache.execution_device)
+            self._cache_entry.loaded = True
+            self._cache.logger.debug(f"Locking {self._cache_entry.key} in {self._cache.execution_device}")
             self._cache.print_cuda_stats()
         except torch.cuda.OutOfMemoryError:
             self._cache.logger.warning("Insufficient GPU memory to load model. Aborting")
+            self._cache_entry.unlock()
             raise
         except Exception:
+            self._cache_entry.unlock()
             raise
 
-        return model_on_device
+        return self.model
 
-    # It is no longer necessary to move the model out of VRAM
-    # because it will be removed when it goes out of scope
-    # in the caller's context
     def unlock(self) -> None:
         """Call upon exit from context."""
-        self._cache.print_cuda_stats()
-
-    # This is no longer in use in MGPU.
-    def get_state_dict(self) -> Optional[Dict[str, torch.Tensor]]:
-        """Return the state dict (if any) for the cached model."""
-        return None
+        self._cache_entry.unlock()
+        if not self._cache.lazy_offloading:
+            self._cache.offload_unlocked_models(0)
+            self._cache.print_cuda_stats()

invokeai/backend/model_manager/load/model_loaders/vae.py

@@ -22,7 +22,8 @@ from .generic_diffusers import GenericDiffusersLoader
 
 
 @ModelLoaderRegistry.register(base=BaseModelType.Any, type=ModelType.VAE, format=ModelFormat.Diffusers)
-@ModelLoaderRegistry.register(base=BaseModelType.Any, type=ModelType.VAE, format=ModelFormat.Checkpoint)
+@ModelLoaderRegistry.register(base=BaseModelType.StableDiffusion1, type=ModelType.VAE, format=ModelFormat.Checkpoint)
+@ModelLoaderRegistry.register(base=BaseModelType.StableDiffusion2, type=ModelType.VAE, format=ModelFormat.Checkpoint)
 class VAELoader(GenericDiffusersLoader):
     """Class to load VAE models."""
 
@@ -39,8 +40,12 @@ class VAELoader(GenericDiffusersLoader):
             return True
 
     def _convert_model(self, config: AnyModelConfig, model_path: Path, output_path: Optional[Path] = None) -> AnyModel:
-        assert isinstance(config, CheckpointConfigBase)
-        config_file = self._app_config.legacy_conf_path / config.config_path
+        # TODO(MM2): check whether sdxl VAE models convert.
+        if config.base not in {BaseModelType.StableDiffusion1, BaseModelType.StableDiffusion2}:
+            raise Exception(f"VAE conversion not supported for model type: {config.base}")
+        else:
+            assert isinstance(config, CheckpointConfigBase)
+            config_file = self._app_config.legacy_conf_path / config.config_path
 
         if model_path.suffix == ".safetensors":
             checkpoint = safetensors_load_file(model_path, device="cpu")

invokeai/backend/model_manager/probe.py

@@ -451,16 +451,8 @@ class PipelineCheckpointProbe(CheckpointProbeBase):
 
 class VaeCheckpointProbe(CheckpointProbeBase):
     def get_base_type(self) -> BaseModelType:
-        # VAEs of all base types have the same structure, so we wimp out and
-        # guess using the name.
-        for regexp, basetype in [
-            (r"xl", BaseModelType.StableDiffusionXL),
-            (r"sd2", BaseModelType.StableDiffusion2),
-            (r"vae", BaseModelType.StableDiffusion1),
-        ]:
-            if re.search(regexp, self.model_path.name, re.IGNORECASE):
-                return basetype
-        raise InvalidModelConfigException("Cannot determine base type")
+        # I can't find any standalone 2.X VAEs to test with!
+        return BaseModelType.StableDiffusion1
 
 
 class LoRACheckpointProbe(CheckpointProbeBase):

invokeai/backend/model_patcher.py

@@ -4,7 +4,6 @@
 from __future__ import annotations
 
 import pickle
-import threading
 from contextlib import contextmanager
 from typing import Any, Dict, Generator, Iterator, List, Optional, Tuple, Union
 
@@ -35,8 +34,6 @@ with LoRAHelper.apply_lora_unet(unet, loras):
 
 # TODO: rename smth like ModelPatcher and add TI method?
 class ModelPatcher:
-    _thread_lock = threading.Lock()
-
     @staticmethod
     def _resolve_lora_key(model: torch.nn.Module, lora_key: str, prefix: str) -> Tuple[str, torch.nn.Module]:
         assert "." not in lora_key
@@ -109,7 +106,7 @@ class ModelPatcher:
         """
         original_weights = {}
         try:
-            with torch.no_grad(), cls._thread_lock:
+            with torch.no_grad():
                 for lora, lora_weight in loras:
                     # assert lora.device.type == "cpu"
                     for layer_key, layer in lora.layers.items():
@@ -132,7 +129,9 @@ class ModelPatcher:
                         dtype = module.weight.dtype
 
                         if module_key not in original_weights:
-                            if model_state_dict is None:  # no CPU copy of the state dict was provided
+                            if model_state_dict is not None:  # we were provided with the CPU copy of the state dict
+                                original_weights[module_key] = model_state_dict[module_key + ".weight"]
+                            else:
                                 original_weights[module_key] = module.weight.detach().to(device="cpu", copy=True)
 
                         layer_scale = layer.alpha / layer.rank if (layer.alpha and layer.rank) else 1.0

invokeai/backend/stable_diffusion/diffusers_pipeline.py

@@ -10,12 +10,11 @@ import PIL.Image
 import psutil
 import torch
 import torchvision.transforms as T
-from diffusers.models import AutoencoderKL, UNet2DConditionModel
-from diffusers.models.controlnet import ControlNetModel
+from diffusers.models.autoencoders.autoencoder_kl import AutoencoderKL
+from diffusers.models.unets.unet_2d_condition import UNet2DConditionModel
 from diffusers.pipelines.stable_diffusion.pipeline_stable_diffusion import StableDiffusionPipeline
 from diffusers.pipelines.stable_diffusion.safety_checker import StableDiffusionSafetyChecker
-from diffusers.schedulers import KarrasDiffusionSchedulers
-from diffusers.schedulers.scheduling_utils import SchedulerMixin
+from diffusers.schedulers.scheduling_utils import KarrasDiffusionSchedulers, SchedulerMixin
 from diffusers.utils.import_utils import is_xformers_available
 from pydantic import Field
 from transformers import CLIPFeatureExtractor, CLIPTextModel, CLIPTokenizer
@@ -26,6 +25,7 @@ from invokeai.backend.stable_diffusion.diffusion.shared_invokeai_diffusion impor
 from invokeai.backend.stable_diffusion.diffusion.unet_attention_patcher import UNetAttentionPatcher, UNetIPAdapterData
 from invokeai.backend.util.attention import auto_detect_slice_size
 from invokeai.backend.util.devices import TorchDevice
+from invokeai.backend.util.hotfixes import ControlNetModel
 
 
 @dataclass
@@ -38,56 +38,18 @@ class PipelineIntermediateState:
     predicted_original: Optional[torch.Tensor] = None
 
 
-@dataclass
-class AddsMaskLatents:
-    """Add the channels required for inpainting model input.
-
-    The inpainting model takes the normal latent channels as input, _plus_ a one-channel mask
-    and the latent encoding of the base image.
-
-    This class assumes the same mask and base image should apply to all items in the batch.
-    """
-
-    forward: Callable[[torch.Tensor, torch.Tensor, torch.Tensor], torch.Tensor]
-    mask: torch.Tensor
-    initial_image_latents: torch.Tensor
-
-    def __call__(
-        self,
-        latents: torch.Tensor,
-        t: torch.Tensor,
-        text_embeddings: torch.Tensor,
-        **kwargs,
-    ) -> torch.Tensor:
-        model_input = self.add_mask_channels(latents)
-        return self.forward(model_input, t, text_embeddings, **kwargs)
-
-    def add_mask_channels(self, latents):
-        batch_size = latents.size(0)
-        # duplicate mask and latents for each batch
-        mask = einops.repeat(self.mask, "b c h w -> (repeat b) c h w", repeat=batch_size)
-        image_latents = einops.repeat(self.initial_image_latents, "b c h w -> (repeat b) c h w", repeat=batch_size)
-        # add mask and image as additional channels
-        model_input, _ = einops.pack([latents, mask, image_latents], "b * h w")
-        return model_input
-
-
-def are_like_tensors(a: torch.Tensor, b: object) -> bool:
-    return isinstance(b, torch.Tensor) and (a.size() == b.size())
-
-
 @dataclass
 class AddsMaskGuidance:
-    mask: torch.FloatTensor
-    mask_latents: torch.FloatTensor
+    mask: torch.Tensor
+    mask_latents: torch.Tensor
     scheduler: SchedulerMixin
     noise: torch.Tensor
-    gradient_mask: bool
+    is_gradient_mask: bool
 
     def __call__(self, latents: torch.Tensor, t: torch.Tensor) -> torch.Tensor:
         return self.apply_mask(latents, t)
 
-    def apply_mask(self, latents: torch.Tensor, t) -> torch.Tensor:
+    def apply_mask(self, latents: torch.Tensor, t: torch.Tensor) -> torch.Tensor:
         batch_size = latents.size(0)
         mask = einops.repeat(self.mask, "b c h w -> (repeat b) c h w", repeat=batch_size)
         if t.dim() == 0:
@@ -100,7 +62,7 @@ class AddsMaskGuidance:
         # TODO: Do we need to also apply scheduler.scale_model_input? Or is add_noise appropriately scaled already?
         # mask_latents = self.scheduler.scale_model_input(mask_latents, t)
         mask_latents = einops.repeat(mask_latents, "b c h w -> (repeat b) c h w", repeat=batch_size)
-        if self.gradient_mask:
+        if self.is_gradient_mask:
             threshhold = (t.item()) / self.scheduler.config.num_train_timesteps
             mask_bool = mask > threshhold  # I don't know when mask got inverted, but it did
             masked_input = torch.where(mask_bool, latents, mask_latents)
@@ -200,7 +162,6 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
         safety_checker: Optional[StableDiffusionSafetyChecker],
         feature_extractor: Optional[CLIPFeatureExtractor],
         requires_safety_checker: bool = False,
-        control_model: ControlNetModel = None,
     ):
         super().__init__(
             vae=vae,
@@ -214,8 +175,6 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
         )
 
         self.invokeai_diffuser = InvokeAIDiffuserComponent(self.unet, self._unet_forward)
-        self.control_model = control_model
-        self.use_ip_adapter = False
 
     def _adjust_memory_efficient_attention(self, latents: torch.Tensor):
         """
@@ -280,116 +239,131 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
     def to(self, torch_device: Optional[Union[str, torch.device]] = None, silence_dtype_warnings=False):
         raise Exception("Should not be called")
 
+    def add_inpainting_channels_to_latents(
+        self, latents: torch.Tensor, masked_ref_image_latents: torch.Tensor, inpainting_mask: torch.Tensor
+    ):
+        """Given a `latents` tensor, adds the mask and image latents channels required for inpainting.
+
+        Standard (non-inpainting) SD UNet models expect an input with shape (N, 4, H, W). Inpainting models expect an
+        input of shape (N, 9, H, W). The 9 channels are defined as follows:
+        - Channel 0-3: The latents being denoised.
+        - Channel 4: The mask indicating which parts of the image are being inpainted.
+        - Channel 5-8: The latent representation of the masked reference image being inpainted.
+
+        This function assumes that the same mask and base image should apply to all items in the batch.
+        """
+        # Validate assumptions about input tensor shapes.
+        batch_size, latent_channels, latent_height, latent_width = latents.shape
+        assert latent_channels == 4
+        assert masked_ref_image_latents.shape == [1, 4, latent_height, latent_width]
+        assert inpainting_mask == [1, 1, latent_height, latent_width]
+
+        # Repeat original_image_latents and inpainting_mask to match the latents batch size.
+        original_image_latents = masked_ref_image_latents.expand(batch_size, -1, -1, -1)
+        inpainting_mask = inpainting_mask.expand(batch_size, -1, -1, -1)
+
+        # Concatenate along the channel dimension.
+        return torch.cat([latents, inpainting_mask, original_image_latents], dim=1)
+
     def latents_from_embeddings(
         self,
         latents: torch.Tensor,
-        num_inference_steps: int,
         scheduler_step_kwargs: dict[str, Any],
         conditioning_data: TextConditioningData,
-        *,
         noise: Optional[torch.Tensor],
+        seed: int,
         timesteps: torch.Tensor,
         init_timestep: torch.Tensor,
-        additional_guidance: List[Callable] = None,
-        callback: Callable[[PipelineIntermediateState], None] = None,
-        control_data: List[ControlNetData] = None,
+        callback: Callable[[PipelineIntermediateState], None],
+        control_data: list[ControlNetData] | None = None,
         ip_adapter_data: Optional[list[IPAdapterData]] = None,
         t2i_adapter_data: Optional[list[T2IAdapterData]] = None,
         mask: Optional[torch.Tensor] = None,
         masked_latents: Optional[torch.Tensor] = None,
-        gradient_mask: Optional[bool] = False,
-        seed: int,
+        is_gradient_mask: bool = False,
     ) -> torch.Tensor:
-        if init_timestep.shape[0] == 0:
-            return latents
+        """Denoise the latents.
 
-        if additional_guidance is None:
-            additional_guidance = []
+        Args:
+            latents: The latent-space image to denoise.
+                - If we are inpainting, this is the initial latent image before noise has been added.
+                - If we are generating a new image, this should be initialized to zeros.
+                - In some cases, this may be a partially-noised latent image (e.g. when running the SDXL refiner).
+            scheduler_step_kwargs: kwargs forwarded to the scheduler.step() method.
+            conditioning_data: Text conditionging data.
+            noise: Noise used for two purposes:
+                1. Used by the scheduler to noise the initial `latents` before denoising.
+                2. Used to noise the `masked_latents` when inpainting.
+                `noise` should be None if the `latents` tensor has already been noised.
+            seed: The seed used to generate the noise for the denoising process.
+                HACK(ryand): seed is only used in a particular case when `noise` is None, but we need to re-generate the
+                same noise used earlier in the pipeline. This should really be handled in a clearer way.
+            timesteps: The timestep schedule for the denoising process.
+            init_timestep: The first timestep in the schedule.
+                TODO(ryand): I'm pretty sure this should always be the same as timesteps[0:1]. Confirm that that is the
+                case, and remove this duplicate param.
+            callback: A callback function that is called to report progress during the denoising process.
+            control_data: ControlNet data.
+            ip_adapter_data: IP-Adapter data.
+            t2i_adapter_data: T2I-Adapter data.
+            mask: A mask indicating which parts of the image are being inpainted. The presence of mask is used to
+                determine whether we are inpainting or not. `mask` should have the same spatial dimensions as the
+                `latents` tensor.
+                TODO(ryand): Check and document the expected dtype, range, and values used to represent
+                foreground/background.
+            masked_latents: A latent-space representation of a masked inpainting reference image. This tensor is only
+                used if an *inpainting* model is being used i.e. this tensor is not used when inpainting with a standard
+                SD UNet model.
+            is_gradient_mask: A flag indicating whether `mask` is a gradient mask or not.
+        """
+        # TODO(ryand): Figure out why this condition is necessary, and document it. My guess is that it's to handle
+        # cases where densoisings_start and denoising_end are set such that there are no timesteps.
+        if init_timestep.shape[0] == 0 or timesteps.shape[0] == 0:
+            return latents
 
         orig_latents = latents.clone()
 
         batch_size = latents.shape[0]
-        batched_t = init_timestep.expand(batch_size)
+        batched_init_timestep = init_timestep.expand(batch_size)
 
+        # noise can be None if the latents have already been noised (e.g. when running the SDXL refiner).
         if noise is not None:
+            # TODO(ryand): I'm pretty sure we should be applying init_noise_sigma in cases where we are starting with
+            # full noise. Investigate the history of why this got commented out.
             # latents = noise * self.scheduler.init_noise_sigma # it's like in t2l according to diffusers
-            latents = self.scheduler.add_noise(latents, noise, batched_t)
+            latents = self.scheduler.add_noise(latents, noise, batched_init_timestep)
 
-        if mask is not None:
-            if is_inpainting_model(self.unet):
-                if masked_latents is None:
-                    raise Exception("Source image required for inpaint mask when inpaint model used!")
-
-                self.invokeai_diffuser.model_forward_callback = AddsMaskLatents(
-                    self._unet_forward, mask, masked_latents
-                )
-            else:
-                # if no noise provided, noisify unmasked area based on seed
-                if noise is None:
-                    noise = torch.randn(
-                        orig_latents.shape,
-                        dtype=torch.float32,
-                        device="cpu",
-                        generator=torch.Generator(device="cpu").manual_seed(seed),
-                    ).to(device=orig_latents.device, dtype=orig_latents.dtype)
-
-                additional_guidance.append(AddsMaskGuidance(mask, orig_latents, self.scheduler, noise, gradient_mask))
-
-        try:
-            latents = self.generate_latents_from_embeddings(
-                latents,
-                timesteps,
-                conditioning_data,
-                scheduler_step_kwargs=scheduler_step_kwargs,
-                additional_guidance=additional_guidance,
-                control_data=control_data,
-                ip_adapter_data=ip_adapter_data,
-                t2i_adapter_data=t2i_adapter_data,
-                callback=callback,
-            )
-        finally:
-            self.invokeai_diffuser.model_forward_callback = self._unet_forward
-
-        # restore unmasked part after the last step is completed
-        # in-process masking happens before each step
-        if mask is not None:
-            if gradient_mask:
-                latents = torch.where(mask > 0, latents, orig_latents)
-            else:
-                latents = torch.lerp(
-                    orig_latents, latents.to(dtype=orig_latents.dtype), mask.to(dtype=orig_latents.dtype)
-                )
-
-        return latents
-
-    def generate_latents_from_embeddings(
-        self,
-        latents: torch.Tensor,
-        timesteps,
-        conditioning_data: TextConditioningData,
-        scheduler_step_kwargs: dict[str, Any],
-        *,
-        additional_guidance: List[Callable] = None,
-        control_data: List[ControlNetData] = None,
-        ip_adapter_data: Optional[list[IPAdapterData]] = None,
-        t2i_adapter_data: Optional[list[T2IAdapterData]] = None,
-        callback: Callable[[PipelineIntermediateState], None] = None,
-    ) -> torch.Tensor:
         self._adjust_memory_efficient_attention(latents)
-        if additional_guidance is None:
-            additional_guidance = []
 
-        batch_size = latents.shape[0]
+        # Handle mask guidance (a.k.a. inpainting).
+        mask_guidance: AddsMaskGuidance | None = None
+        if mask is not None and not is_inpainting_model(self.unet):
+            # We are doing inpainting, since a mask is provided, but we are not using an inpainting model, so we will
+            # apply mask guidance to the latents.
 
-        if timesteps.shape[0] == 0:
-            return latents
+            # 'noise' might be None if the latents have already been noised (e.g. when running the SDXL refiner).
+            # We still need noise for inpainting, so we generate it from the seed here.
+            if noise is None:
+                noise = torch.randn(
+                    orig_latents.shape,
+                    dtype=torch.float32,
+                    device="cpu",
+                    generator=torch.Generator(device="cpu").manual_seed(seed),
+                ).to(device=orig_latents.device, dtype=orig_latents.dtype)
+
+            mask_guidance = AddsMaskGuidance(
+                mask=mask,
+                mask_latents=orig_latents,
+                scheduler=self.scheduler,
+                noise=noise,
+                is_gradient_mask=is_gradient_mask,
+            )
 
         use_ip_adapter = ip_adapter_data is not None
         use_regional_prompting = (
             conditioning_data.cond_regions is not None or conditioning_data.uncond_regions is not None
         )
         unet_attention_patcher = None
-        self.use_ip_adapter = use_ip_adapter
         attn_ctx = nullcontext()
 
         if use_ip_adapter or use_regional_prompting:
@@ -402,28 +376,28 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
             attn_ctx = unet_attention_patcher.apply_ip_adapter_attention(self.invokeai_diffuser.model)
 
         with attn_ctx:
-            if callback is not None:
-                callback(
-                    PipelineIntermediateState(
-                        step=-1,
-                        order=self.scheduler.order,
-                        total_steps=len(timesteps),
-                        timestep=self.scheduler.config.num_train_timesteps,
-                        latents=latents,
-                    )
+            callback(
+                PipelineIntermediateState(
+                    step=-1,
+                    order=self.scheduler.order,
+                    total_steps=len(timesteps),
+                    timestep=self.scheduler.config.num_train_timesteps,
+                    latents=latents,
                 )
+            )
 
-            # print("timesteps:", timesteps)
             for i, t in enumerate(self.progress_bar(timesteps)):
                 batched_t = t.expand(batch_size)
                 step_output = self.step(
-                    batched_t,
-                    latents,
-                    conditioning_data,
+                    t=batched_t,
+                    latents=latents,
+                    conditioning_data=conditioning_data,
                     step_index=i,
                     total_step_count=len(timesteps),
                     scheduler_step_kwargs=scheduler_step_kwargs,
-                    additional_guidance=additional_guidance,
+                    mask_guidance=mask_guidance,
+                    mask=mask,
+                    masked_latents=masked_latents,
                     control_data=control_data,
                     ip_adapter_data=ip_adapter_data,
                     t2i_adapter_data=t2i_adapter_data,
@@ -431,19 +405,28 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
                 latents = step_output.prev_sample
                 predicted_original = getattr(step_output, "pred_original_sample", None)
 
-                if callback is not None:
-                    callback(
-                        PipelineIntermediateState(
-                            step=i,
-                            order=self.scheduler.order,
-                            total_steps=len(timesteps),
-                            timestep=int(t),
-                            latents=latents,
-                            predicted_original=predicted_original,
-                        )
+                callback(
+                    PipelineIntermediateState(
+                        step=i,
+                        order=self.scheduler.order,
+                        total_steps=len(timesteps),
+                        timestep=int(t),
+                        latents=latents,
+                        predicted_original=predicted_original,
                     )
+                )
 
-            return latents
+        # restore unmasked part after the last step is completed
+        # in-process masking happens before each step
+        if mask is not None:
+            if is_gradient_mask:
+                latents = torch.where(mask > 0, latents, orig_latents)
+            else:
+                latents = torch.lerp(
+                    orig_latents, latents.to(dtype=orig_latents.dtype), mask.to(dtype=orig_latents.dtype)
+                )
+
+        return latents
 
     @torch.inference_mode()
     def step(
@@ -454,19 +437,20 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
         step_index: int,
         total_step_count: int,
         scheduler_step_kwargs: dict[str, Any],
-        additional_guidance: List[Callable] = None,
-        control_data: List[ControlNetData] = None,
+        mask_guidance: AddsMaskGuidance | None,
+        mask: torch.Tensor | None,
+        masked_latents: torch.Tensor | None,
+        control_data: list[ControlNetData] | None = None,
         ip_adapter_data: Optional[list[IPAdapterData]] = None,
         t2i_adapter_data: Optional[list[T2IAdapterData]] = None,
     ):
         # invokeai_diffuser has batched timesteps, but diffusers schedulers expect a single value
         timestep = t[0]
-        if additional_guidance is None:
-            additional_guidance = []
 
-        # one day we will expand this extension point, but for now it just does denoise masking
-        for guidance in additional_guidance:
-            latents = guidance(latents, timestep)
+        # Handle masked image-to-image (a.k.a inpainting).
+        if mask_guidance is not None:
+            # NOTE: This is intentionally done *before* self.scheduler.scale_model_input(...).
+            latents = mask_guidance(latents, timestep)
 
         # TODO: should this scaling happen here or inside self._unet_forward?
         #     i.e. before or after passing it to InvokeAIDiffuserComponent
@@ -514,6 +498,31 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
 
             down_intrablock_additional_residuals = accum_adapter_state
 
+        # Handle inpainting models.
+        if is_inpainting_model(self.unet):
+            # NOTE: These calls to add_inpainting_channels_to_latents(...) are intentionally done *after*
+            # self.scheduler.scale_model_input(...) so that the scaling is not applied to the mask or reference image
+            # latents.
+            if mask is not None:
+                if masked_latents is None:
+                    raise ValueError("Source image required for inpaint mask when inpaint model used!")
+                latent_model_input = self.add_inpainting_channels_to_latents(
+                    latents=latent_model_input, masked_ref_image_latents=masked_latents, inpainting_mask=mask
+                )
+            else:
+                # We are using an inpainting model, but no mask was provided, so we are not really "inpainting".
+                # We generate a global mask and empty original image so that we can still generate in this
+                # configuration.
+                # TODO(ryand): Should we just raise an exception here instead? I can't think of a use case for wanting
+                # to do this.
+                # TODO(ryand): If we decide that there is a good reason to keep this, then we should generate the 'fake'
+                # mask and original image once rather than on every denoising step.
+                latent_model_input = self.add_inpainting_channels_to_latents(
+                    latents=latent_model_input,
+                    masked_ref_image_latents=torch.zeros_like(latent_model_input[:1]),
+                    inpainting_mask=torch.ones_like(latent_model_input[:1, :1]),
+                )
+
         uc_noise_pred, c_noise_pred = self.invokeai_diffuser.do_unet_step(
             sample=latent_model_input,
             timestep=t,  # TODO: debug how handled batched and non batched timesteps
@@ -542,17 +551,18 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
         # compute the previous noisy sample x_t -> x_t-1
         step_output = self.scheduler.step(noise_pred, timestep, latents, **scheduler_step_kwargs)
 
-        # TODO: discuss injection point options. For now this is a patch to get progress images working with inpainting again.
-        for guidance in additional_guidance:
-            # apply the mask to any "denoised" or "pred_original_sample" fields
+        # TODO: discuss injection point options. For now this is a patch to get progress images working with inpainting
+        # again.
+        if mask_guidance is not None:
+            # Apply the mask to any "denoised" or "pred_original_sample" fields.
             if hasattr(step_output, "denoised"):
-                step_output.pred_original_sample = guidance(step_output.denoised, self.scheduler.timesteps[-1])
+                step_output.pred_original_sample = mask_guidance(step_output.denoised, self.scheduler.timesteps[-1])
             elif hasattr(step_output, "pred_original_sample"):
-                step_output.pred_original_sample = guidance(
+                step_output.pred_original_sample = mask_guidance(
                     step_output.pred_original_sample, self.scheduler.timesteps[-1]
                 )
             else:
-                step_output.pred_original_sample = guidance(latents, self.scheduler.timesteps[-1])
+                step_output.pred_original_sample = mask_guidance(latents, self.scheduler.timesteps[-1])
 
         return step_output
 
@@ -575,17 +585,6 @@ class StableDiffusionGeneratorPipeline(StableDiffusionPipeline):
         **kwargs,
     ):
         """predict the noise residual"""
-        if is_inpainting_model(self.unet) and latents.size(1) == 4:
-            # Pad out normal non-inpainting inputs for an inpainting model.
-            # FIXME: There are too many layers of functions and we have too many different ways of
-            #     overriding things! This should get handled in a way more consistent with the other
-            #     use of AddsMaskLatents.
-            latents = AddsMaskLatents(
-                self._unet_forward,
-                mask=torch.ones_like(latents[:1, :1], device=latents.device, dtype=latents.dtype),
-                initial_image_latents=torch.zeros_like(latents[:1], device=latents.device, dtype=latents.dtype),
-            ).add_mask_channels(latents)
-
         # First three args should be positional, not keywords, so torch hooks can see them.
         return self.unet(
             latents,

invokeai/backend/stable_diffusion/diffusion/conditioning_data.py

@@ -32,11 +32,8 @@ class SDXLConditioningInfo(BasicConditioningInfo):
 
     def to(self, device, dtype=None):
         self.pooled_embeds = self.pooled_embeds.to(device=device, dtype=dtype)
-        assert self.pooled_embeds.device == device
         self.add_time_ids = self.add_time_ids.to(device=device, dtype=dtype)
-        result = super().to(device=device, dtype=dtype)
-        assert self.embeds.device == device
-        return result
+        return super().to(device=device, dtype=dtype)
 
 
 @dataclass

invokeai/backend/stable_diffusion/diffusion/shared_invokeai_diffusion.py

@@ -1,7 +1,6 @@
 from __future__ import annotations
 
 import math
-import threading
 from typing import Any, Callable, Optional, Union
 
 import torch
@@ -294,31 +293,24 @@ class InvokeAIDiffuserComponent:
             cross_attention_kwargs["regional_ip_data"] = regional_ip_data
 
         added_cond_kwargs = None
-        try:
-            if conditioning_data.is_sdxl():
-                # tid = threading.current_thread().ident
-                # print(f'DEBUG {tid} {conditioning_data.uncond_text.pooled_embeds.device=} {conditioning_data.cond_text.pooled_embeds.device=}', flush=True),
-                added_cond_kwargs = {
-                    "text_embeds": torch.cat(
-                        [
-                            # TODO: how to pad? just by zeros? or even truncate?
-                            conditioning_data.uncond_text.pooled_embeds,
-                            conditioning_data.cond_text.pooled_embeds,
-                        ],
-                        dim=0,
-                    ),
-                    "time_ids": torch.cat(
-                        [
-                            conditioning_data.uncond_text.add_time_ids,
-                            conditioning_data.cond_text.add_time_ids,
-                        ],
-                        dim=0,
-                    ),
-                }
-        except Exception as e:
-            tid = threading.current_thread().ident
-            print(f"DEBUG: {tid} {str(e)}")
-            raise e
+        if conditioning_data.is_sdxl():
+            added_cond_kwargs = {
+                "text_embeds": torch.cat(
+                    [
+                        # TODO: how to pad? just by zeros? or even truncate?
+                        conditioning_data.uncond_text.pooled_embeds,
+                        conditioning_data.cond_text.pooled_embeds,
+                    ],
+                    dim=0,
+                ),
+                "time_ids": torch.cat(
+                    [
+                        conditioning_data.uncond_text.add_time_ids,
+                        conditioning_data.cond_text.add_time_ids,
+                    ],
+                    dim=0,
+                ),
+            }
 
         if conditioning_data.cond_regions is not None or conditioning_data.uncond_regions is not None:
             # TODO(ryand): We currently initialize RegionalPromptData for every denoising step. The text conditionings

invokeai/backend/stable_diffusion/multi_diffusion_pipeline.py

@@ -0,0 +1,242 @@
+from __future__ import annotations
+
+import copy
+from dataclasses import dataclass
+from typing import Any, Callable, Optional
+
+import torch
+from diffusers.schedulers.scheduling_utils import SchedulerMixin
+
+from invokeai.backend.stable_diffusion.diffusers_pipeline import (
+    ControlNetData,
+    PipelineIntermediateState,
+    StableDiffusionGeneratorPipeline,
+)
+from invokeai.backend.stable_diffusion.diffusion.conditioning_data import TextConditioningData
+from invokeai.backend.tiles.utils import TBLR
+
+# The maximum number of regions with compatible sizes that will be batched together.
+# Larger batch sizes improve speed, but require more device memory.
+MAX_REGION_BATCH_SIZE = 4
+
+
+@dataclass
+class MultiDiffusionRegionConditioning:
+    # Region coords in latent space.
+    region: TBLR
+    text_conditioning_data: TextConditioningData
+    control_data: list[ControlNetData]
+
+
+class MultiDiffusionPipeline(StableDiffusionGeneratorPipeline):
+    """A Stable Diffusion pipeline that uses Multi-Diffusion (https://arxiv.org/pdf/2302.08113) for denoising."""
+
+    def _split_into_region_batches(
+        self, multi_diffusion_conditioning: list[MultiDiffusionRegionConditioning]
+    ) -> list[list[MultiDiffusionRegionConditioning]]:
+        # Group the regions by shape. Only regions with the same shape can be batched together.
+        conditioning_by_shape: dict[tuple[int, int], list[MultiDiffusionRegionConditioning]] = {}
+        for region_conditioning in multi_diffusion_conditioning:
+            shape_hw = (
+                region_conditioning.region.bottom - region_conditioning.region.top,
+                region_conditioning.region.right - region_conditioning.region.left,
+            )
+            # In python, a tuple of hashable objects is hashable, so can be used as a key in a dict.
+            if shape_hw not in conditioning_by_shape:
+                conditioning_by_shape[shape_hw] = []
+            conditioning_by_shape[shape_hw].append(region_conditioning)
+
+        # Split the regions into batches, respecting the MAX_REGION_BATCH_SIZE constraint.
+        region_conditioning_batches = []
+        for region_conditioning_batch in conditioning_by_shape.values():
+            for i in range(0, len(region_conditioning_batch), MAX_REGION_BATCH_SIZE):
+                region_conditioning_batches.append(region_conditioning_batch[i : i + MAX_REGION_BATCH_SIZE])
+
+        return region_conditioning_batches
+
+    def _check_regional_prompting(self, multi_diffusion_conditioning: list[MultiDiffusionRegionConditioning]):
+        """Check the input conditioning and confirm that regional prompting is not used."""
+        for region_conditioning in multi_diffusion_conditioning:
+            if (
+                region_conditioning.text_conditioning_data.cond_regions is not None
+                or region_conditioning.text_conditioning_data.uncond_regions is not None
+            ):
+                raise NotImplementedError("Regional prompting is not yet supported in Multi-Diffusion.")
+
+    def multi_diffusion_denoise(
+        self,
+        multi_diffusion_conditioning: list[MultiDiffusionRegionConditioning],
+        latents: torch.Tensor,
+        scheduler_step_kwargs: dict[str, Any],
+        noise: Optional[torch.Tensor],
+        timesteps: torch.Tensor,
+        init_timestep: torch.Tensor,
+        callback: Callable[[PipelineIntermediateState], None],
+    ) -> torch.Tensor:
+        self._check_regional_prompting(multi_diffusion_conditioning)
+
+        # TODO(ryand): Figure out why this condition is necessary, and document it. My guess is that it's to handle
+        # cases where densoisings_start and denoising_end are set such that there are no timesteps.
+        if init_timestep.shape[0] == 0 or timesteps.shape[0] == 0:
+            return latents
+
+        batch_size, _, latent_height, latent_width = latents.shape
+        batched_init_timestep = init_timestep.expand(batch_size)
+
+        # noise can be None if the latents have already been noised (e.g. when running the SDXL refiner).
+        if noise is not None:
+            # TODO(ryand): I'm pretty sure we should be applying init_noise_sigma in cases where we are starting with
+            # full noise. Investigate the history of why this got commented out.
+            # latents = noise * self.scheduler.init_noise_sigma # it's like in t2l according to diffusers
+            latents = self.scheduler.add_noise(latents, noise, batched_init_timestep)
+
+        # TODO(ryand): Look into the implications of passing in latents here that are larger than they will be after
+        # cropping into regions.
+        self._adjust_memory_efficient_attention(latents)
+
+        # Populate a weighted mask that will be used to combine the results from each region after every step.
+        # For now, we assume that each region has the same weight (1.0).
+        region_weight_mask = torch.zeros(
+            (1, 1, latent_height, latent_width), device=latents.device, dtype=latents.dtype
+        )
+        for region_conditioning in multi_diffusion_conditioning:
+            region = region_conditioning.region
+            region_weight_mask[:, :, region.top : region.bottom, region.left : region.right] += 1.0
+
+        # Group the region conditioning into batches for faster processing.
+        # region_conditioning_batches[b][r] is the r'th region in the b'th batch.
+        region_conditioning_batches = self._split_into_region_batches(multi_diffusion_conditioning)
+
+        # Many of the diffusers schedulers are stateful (i.e. they update internal state in each call to step()). Since
+        # we are calling step() multiple times at the same timestep (once for each region batch), we must maintain a
+        # separate scheduler state for each region batch.
+        region_batch_schedulers: list[SchedulerMixin] = [
+            copy.deepcopy(self.scheduler) for _ in region_conditioning_batches
+        ]
+
+        callback(
+            PipelineIntermediateState(
+                step=-1,
+                order=self.scheduler.order,
+                total_steps=len(timesteps),
+                timestep=self.scheduler.config.num_train_timesteps,
+                latents=latents,
+            )
+        )
+
+        for i, t in enumerate(self.progress_bar(timesteps)):
+            batched_t = t.expand(batch_size)
+
+            merged_latents = torch.zeros_like(latents)
+            merged_pred_original: torch.Tensor | None = None
+            for region_batch_idx, region_conditioning_batch in enumerate(region_conditioning_batches):
+                # Switch to the scheduler for the region batch.
+                self.scheduler = region_batch_schedulers[region_batch_idx]
+
+                # TODO(ryand): This logic has not yet been tested with input latents with a batch_size > 1.
+
+                # Prepare the latents for the region batch.
+                batch_latents = torch.cat(
+                    [
+                        latents[
+                            :,
+                            :,
+                            region_conditioning.region.top : region_conditioning.region.bottom,
+                            region_conditioning.region.left : region_conditioning.region.right,
+                        ]
+                        for region_conditioning in region_conditioning_batch
+                    ],
+                )
+
+                # TODO(ryand): Do we have to repeat the text_conditioning_data to match the batch size? Or does step()
+                # handle broadcasting properly?
+
+                # TODO(ryand): Resume here!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
+                # Run the denoising step on the region.
+                step_output = self.step(
+                    t=batched_t,
+                    latents=batch_latents,
+                    conditioning_data=region_conditioning.text_conditioning_data,
+                    step_index=i,
+                    total_step_count=total_step_count,
+                    scheduler_step_kwargs=scheduler_step_kwargs,
+                    mask_guidance=None,
+                    mask=None,
+                    masked_latents=None,
+                    control_data=region_conditioning.control_data,
+                )
+                # Run a denoising step on the region.
+                # step_output = self._region_step(
+                #     region_conditioning=region_conditioning,
+                #     t=batched_t,
+                #     latents=latents,
+                #     step_index=i,
+                #     total_step_count=len(timesteps),
+                #     scheduler_step_kwargs=scheduler_step_kwargs,
+                # )
+
+                # Store the results from the region.
+                region = region_conditioning.region
+                merged_latents[:, :, region.top : region.bottom, region.left : region.right] += step_output.prev_sample
+                pred_orig_sample = getattr(step_output, "pred_original_sample", None)
+                if pred_orig_sample is not None:
+                    # If one region has pred_original_sample, then we can assume that all regions will have it, because
+                    # they all use the same scheduler.
+                    if merged_pred_original is None:
+                        merged_pred_original = torch.zeros_like(latents)
+                    merged_pred_original[:, :, region.top : region.bottom, region.left : region.right] += (
+                        pred_orig_sample
+                    )
+
+            # Normalize the merged results.
+            latents = torch.where(region_weight_mask > 0, merged_latents / region_weight_mask, merged_latents)
+            predicted_original = None
+            if merged_pred_original is not None:
+                predicted_original = torch.where(
+                    region_weight_mask > 0, merged_pred_original / region_weight_mask, merged_pred_original
+                )
+
+            callback(
+                PipelineIntermediateState(
+                    step=i,
+                    order=self.scheduler.order,
+                    total_steps=len(timesteps),
+                    timestep=int(t),
+                    latents=latents,
+                    predicted_original=predicted_original,
+                )
+            )
+
+        return latents
+
+    @torch.inference_mode()
+    def _region_batch_step(
+        self,
+        region_conditioning: MultiDiffusionRegionConditioning,
+        t: torch.Tensor,
+        latents: torch.Tensor,
+        step_index: int,
+        total_step_count: int,
+        scheduler_step_kwargs: dict[str, Any],
+    ):
+        # Crop the inputs to the region.
+        region_latents = latents[
+            :,
+            :,
+            region_conditioning.region.top : region_conditioning.region.bottom,
+            region_conditioning.region.left : region_conditioning.region.right,
+        ]
+
+        # Run the denoising step on the region.
+        return self.step(
+            t=t,
+            latents=region_latents,
+            conditioning_data=region_conditioning.text_conditioning_data,
+            step_index=step_index,
+            total_step_count=total_step_count,
+            scheduler_step_kwargs=scheduler_step_kwargs,
+            mask_guidance=None,
+            mask=None,
+            masked_latents=None,
+            control_data=region_conditioning.control_data,
+        )

invokeai/backend/util/devices.py

@@ -1,16 +1,10 @@
-"""Torch Device class provides torch device selection services."""
-
-from typing import TYPE_CHECKING, Dict, Literal, Optional, Set, Union
+from typing import Dict, Literal, Optional, Union
 
 import torch
 from deprecated import deprecated
 
 from invokeai.app.services.config.config_default import get_config
 
-if TYPE_CHECKING:
-    from invokeai.backend.model_manager.config import AnyModel
-    from invokeai.backend.model_manager.load.model_cache.model_cache_base import ModelCacheBase
-
 # legacy APIs
 TorchPrecisionNames = Literal["float32", "float16", "bfloat16"]
 CPU_DEVICE = torch.device("cpu")
@@ -48,23 +42,9 @@ PRECISION_TO_NAME: Dict[torch.dtype, TorchPrecisionNames] = {v: k for k, v in NA
 class TorchDevice:
     """Abstraction layer for torch devices."""
 
-    _model_cache: Optional["ModelCacheBase[AnyModel]"] = None
-
-    @classmethod
-    def set_model_cache(cls, cache: "ModelCacheBase[AnyModel]"):
-        """Set the current model cache."""
-        cls._model_cache = cache
-
     @classmethod
     def choose_torch_device(cls) -> torch.device:
         """Return the torch.device to use for accelerated inference."""
-        if cls._model_cache:
-            return cls._model_cache.get_execution_device()
-        else:
-            return cls._choose_device()
-
-    @classmethod
-    def _choose_device(cls) -> torch.device:
         app_config = get_config()
         if app_config.device != "auto":
             device = torch.device(app_config.device)
@@ -76,19 +56,11 @@ class TorchDevice:
             device = CPU_DEVICE
         return cls.normalize(device)
 
-    @classmethod
-    def execution_devices(cls) -> Set[torch.device]:
-        """Return a list of torch.devices that can be used for accelerated inference."""
-        app_config = get_config()
-        if app_config.devices is None:
-            return cls._lookup_execution_devices()
-        return {torch.device(x) for x in app_config.devices}
-
     @classmethod
     def choose_torch_dtype(cls, device: Optional[torch.device] = None) -> torch.dtype:
         """Return the precision to use for accelerated inference."""
+        device = device or cls.choose_torch_device()
         config = get_config()
-        device = device or cls._choose_device()
         if device.type == "cuda" and torch.cuda.is_available():
             device_name = torch.cuda.get_device_name(device)
             if "GeForce GTX 1660" in device_name or "GeForce GTX 1650" in device_name:
@@ -136,13 +108,3 @@ class TorchDevice:
     @classmethod
     def _to_dtype(cls, precision_name: TorchPrecisionNames) -> torch.dtype:
         return NAME_TO_PRECISION[precision_name]
-
-    @classmethod
-    def _lookup_execution_devices(cls) -> Set[torch.device]:
-        if torch.cuda.is_available():
-            devices = {torch.device(f"cuda:{x}") for x in range(0, torch.cuda.device_count())}
-        elif torch.backends.mps.is_available():
-            devices = {torch.device("mps")}
-        else:
-            devices = {torch.device("cpu")}
-        return devices

scripts/populate_model_db_from_yaml.py

@@ -1,54 +0,0 @@
-#!/bin/env python
-
-from argparse import ArgumentParser, Namespace
-from pathlib import Path
-
-from invokeai.app.services.config import InvokeAIAppConfig, get_config
-from invokeai.app.services.download import DownloadQueueService
-from invokeai.app.services.model_install import ModelInstallService
-from invokeai.app.services.model_records import ModelRecordServiceSQL
-from invokeai.app.services.shared.sqlite.sqlite_database import SqliteDatabase
-from invokeai.backend.util.logging import InvokeAILogger
-
-
-def get_args() -> Namespace:
-    parser = ArgumentParser(description="Update models database from yaml file")
-    parser.add_argument("--root", type=Path, required=False, default=None)
-    parser.add_argument("--yaml_file", type=Path, required=False, default=None)
-    return parser.parse_args()
-
-
-def populate_config() -> InvokeAIAppConfig:
-    args = get_args()
-    config = get_config()
-    if args.root:
-        config._root = args.root
-    if args.yaml_file:
-        config.legacy_models_yaml_path = args.yaml_file
-    else:
-        config.legacy_models_yaml_path = config.root_path / "configs/models.yaml"
-    return config
-
-
-def initialize_installer(config: InvokeAIAppConfig) -> ModelInstallService:
-    logger = InvokeAILogger.get_logger(config=config)
-    db = SqliteDatabase(config.db_path, logger)
-    record_store = ModelRecordServiceSQL(db)
-    queue = DownloadQueueService()
-    queue.start()
-    installer = ModelInstallService(app_config=config, record_store=record_store, download_queue=queue)
-    return installer
-
-
-def main() -> None:
-    config = populate_config()
-    installer = initialize_installer(config)
-    installer._migrate_yaml(rename_yaml=False, overwrite_db=True)
-    print("\n<INSTALLED MODELS>")
-    print("\t".join(["key", "name", "type", "path"]))
-    for model in installer.record_store.all_models():
-        print("\t".join([model.key, model.name, model.type, (config.models_path / model.path).as_posix()]))
-
-
-if __name__ == "__main__":
-    main()

tests/backend/model_manager/model_loading/test_model_load.py

@@ -14,14 +14,13 @@ def test_loading(mm2_model_manager: ModelManagerServiceBase, embedding_file: Pat
     matches = store.search_by_attr(model_name="test_embedding")
     assert len(matches) == 0
     key = mm2_model_manager.install.register_path(embedding_file)
-    with mm2_model_manager.load.ram_cache.reserve_execution_device():
-        loaded_model = mm2_model_manager.load.load_model(store.get_model(key))
-        assert loaded_model is not None
-        assert loaded_model.config.key == key
-        with loaded_model as model:
-            assert isinstance(model, TextualInversionModelRaw)
+    loaded_model = mm2_model_manager.load.load_model(store.get_model(key))
+    assert loaded_model is not None
+    assert loaded_model.config.key == key
+    with loaded_model as model:
+        assert isinstance(model, TextualInversionModelRaw)
 
-        config = mm2_model_manager.store.get_model(key)
-        loaded_model_2 = mm2_model_manager.load.load_model(config)
+    config = mm2_model_manager.store.get_model(key)
+    loaded_model_2 = mm2_model_manager.load.load_model(config)
 
-        assert loaded_model.config.key == loaded_model_2.config.key
+    assert loaded_model.config.key == loaded_model_2.config.key

tests/backend/model_manager/model_manager_fixtures.py

@@ -89,10 +89,11 @@ def mm2_download_queue(mm2_session: Session) -> DownloadQueueServiceBase:
 
 
 @pytest.fixture
-def mm2_loader(mm2_app_config: InvokeAIAppConfig) -> ModelLoadServiceBase:
+def mm2_loader(mm2_app_config: InvokeAIAppConfig, mm2_record_store: ModelRecordServiceBase) -> ModelLoadServiceBase:
     ram_cache = ModelCache(
         logger=InvokeAILogger.get_logger(),
         max_cache_size=mm2_app_config.ram,
+        max_vram_cache_size=mm2_app_config.vram,
     )
     convert_cache = ModelConvertCache(mm2_app_config.convert_cache_path)
     return ModelLoadService(

tests/backend/util/test_devices.py

@@ -8,9 +8,7 @@ import pytest
 import torch
 
 from invokeai.app.services.config import get_config
-from invokeai.backend.model_manager.load import ModelCache
 from invokeai.backend.util.devices import TorchDevice, choose_precision, choose_torch_device, torch_dtype
-from tests.backend.model_manager.model_manager_fixtures import *  # noqa F403
 
 devices = ["cpu", "cuda:0", "cuda:1", "mps"]
 device_types_cpu = [("cpu", torch.float32), ("cuda:0", torch.float32), ("mps", torch.float32)]
@@ -22,7 +20,6 @@ device_types_mps = [("cpu", torch.float32), ("cuda:0", torch.float32), ("mps", t
 def test_device_choice(device_name):
     config = get_config()
     config.device = device_name
-    TorchDevice.set_model_cache(None)  # disable dynamic selection of GPU device
     torch_device = TorchDevice.choose_torch_device()
     assert torch_device == torch.device(device_name)
 
@@ -133,32 +130,3 @@ def test_legacy_precision_name():
         assert "float16" == choose_precision(torch.device("cuda"))
         assert "float16" == choose_precision(torch.device("mps"))
         assert "float32" == choose_precision(torch.device("cpu"))
-
-
-def test_multi_device_support_1():
-    config = get_config()
-    config.devices = ["cuda:0", "cuda:1"]
-    assert TorchDevice.execution_devices() == {torch.device("cuda:0"), torch.device("cuda:1")}
-
-
-def test_multi_device_support_2():
-    config = get_config()
-    config.devices = None
-    with (
-        patch("torch.cuda.device_count", return_value=3),
-        patch("torch.cuda.is_available", return_value=True),
-    ):
-        assert TorchDevice.execution_devices() == {
-            torch.device("cuda:0"),
-            torch.device("cuda:1"),
-            torch.device("cuda:2"),
-        }
-
-
-def test_multi_device_support_3():
-    config = get_config()
-    config.devices = ["cuda:0", "cuda:1"]
-    cache = ModelCache()
-    with cache.reserve_execution_device() as gpu:
-        assert gpu in [torch.device(x) for x in config.devices]
-        assert TorchDevice.choose_torch_device() == gpu

tests/conftest.py

@@ -17,6 +17,7 @@ from invokeai.app.services.config.config_default import InvokeAIAppConfig
 from invokeai.app.services.images.images_default import ImageService
 from invokeai.app.services.invocation_cache.invocation_cache_memory import MemoryInvocationCache
 from invokeai.app.services.invocation_services import InvocationServices
+from invokeai.app.services.invocation_stats.invocation_stats_default import InvocationStatsService
 from invokeai.app.services.invoker import Invoker
 from invokeai.backend.util.logging import InvokeAILogger
 from tests.backend.model_manager.model_manager_fixtures import *  # noqa: F403
@@ -48,13 +49,13 @@ def mock_services() -> InvocationServices:
         model_manager=None,  # type: ignore
         download_queue=None,  # type: ignore
         names=None,  # type: ignore
+        performance_statistics=InvocationStatsService(),
         session_processor=None,  # type: ignore
         session_queue=None,  # type: ignore
         urls=None,  # type: ignore
         workflow_records=None,  # type: ignore
         tensors=None,  # type: ignore
         conditioning=None,  # type: ignore
-        performance_statistics=None,  # type: ignore
     )
 
 

tests/test_config.py

@@ -92,6 +92,7 @@ def test_migrate_v3_config_from_file(tmp_path: Path, patch_rootdir: None):
     assert config.host == "192.168.1.1"
     assert config.port == 8080
     assert config.ram == 100
+    assert config.vram == 50
     assert config.legacy_models_yaml_path == Path("/custom/models.yaml")
     # This should be stripped out
     assert not hasattr(config, "esrgan")