github/AMD-SHARK-Studio
1
1
Fork 0
mirror of https://github.com/nod-ai/AMD-SHARK-Studio.git synced 2026-04-03 03:00:17 -04:00
Code Issues Packages Projects Releases Wiki Activity
976 Commits 66 Branches 705 Tags
enable_caching_alloc
Commit Graph

459 Commits

Author SHA1 Message Date
Quinn Dawkins
e19a97f316 Don't do a numpy copy on the results from compiled vm (#543) 2022-12-05 14:21:47 -05:00
Harish Anand
482ffd6275 Move discord link from advanced instructions (#542) 2022-12-04 06:15:34 -08:00
Quinn Dawkins
5117e50602 Revert "Enable the clip f16 model." until correctness is fixed 2022-12-04 19:17:34 +05:30
Quinn Dawkins
1870cb4557 Add a note to the Stable Diffusion README about clearing vulkan cache (#545) 2022-12-03 15:12:45 -08:00
Prashant Kumar
42ad5b9c5c Enable the clip f16 model. 
-- Enabled the clip f16 model.
-- Updated the location of sdv2 model.
 2022-12-03 18:50:40 +05:30
yzhang93
333975eb8f Update Unet fp16 tuned model and Vae flag (#539) 2022-12-02 23:21:18 -05:00
Gaurav Shukla
aa0195e4ef [SHARK] Add vae encoder wrapper 
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-12-03 08:42:25 +05:30
powderluv
e74046478b Update stable_diffusion_amd.md 2022-12-02 13:57:03 -08:00
Gaurav Shukla
aa5a60812f [SHARK] Fix space issues in download path 
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-12-03 00:52:10 +05:30
mariecwhite
6393dc5d14 Use correct TF device depending on configuration (#492) 2022-12-02 11:33:56 -06:00
powderluv
8c3eabdcee Update stable_diffusion_amd.md 2022-12-02 07:13:10 -08:00
powderluv
8aa0ce6a24 Update stable_diffusion_amd.md 2022-12-02 07:10:31 -08:00
Gaurav Shukla
a27ee141b3 [WEB] Fix few warnings and generate seed faster 
1. Fix gsutil warnings while copying multiple files.
2. Enhance random seed generation speed.
3. Add support for multiple schedulers.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-12-02 17:16:19 +05:30
Quinn Dawkins
8856878cbd Add flag for enabling rgp from the main.py SD script (#533) 2022-12-01 19:01:29 -05:00
Phaneesh Barwaria
a0d0eaa408 add clip and vae timing (#527) 2022-12-01 16:17:40 +05:30
Prashant Kumar
e2bf734b67 Update f32 models. 2022-12-01 14:16:03 +05:30
Prashant Kumar
a333a90441 Update to the latest bytecode. 2022-12-01 12:44:54 +05:30
powderluv
1d32eabd14 Update stable_diffusion_amd.md 2022-11-30 16:52:07 -08:00
powderluv
53d03f06a6 Update stable_diffusion_amd.md 2022-11-30 16:04:53 -08:00
powderluv
a2d8c40455 Update stable_diffusion_amd.md 2022-11-30 15:56:38 -08:00
powderluv
4f7d950c8d Update README.md 2022-11-30 15:54:50 -08:00
Harish Anand
cac54b8c26 Update stable_diffusion_amd.md (#525) 
- Mention `git clone` after installing git in Windows
- Remove the extra . in powershell set-executionpolicy
 2022-11-30 14:48:10 -08:00
powderluv
cd0e881d7d Update stable_diffusion_amd.md 2022-11-30 13:43:24 -08:00
powderluv
128342f47f Update stable_diffusion_amd.md 2022-11-30 13:42:25 -08:00
powderluv
024487c5fe Update stable_diffusion_amd.md 2022-11-30 13:40:00 -08:00
powderluv
879ba27ccb Update stable_diffusion_amd.md 2022-11-30 13:33:04 -08:00
powderluv
6d6d9627e7 Update stable_diffusion_amd.md 2022-11-30 13:31:53 -08:00
powderluv
af4bc82543 Update stable_diffusion_amd.md 2022-11-30 13:30:15 -08:00
powderluv
232ce12e9b Create stable_diffusion_amd.md 2022-11-30 12:10:34 -08:00
aldesilv
9a8638a6d0 dump all isas with amdllpc (#517) 
SHARK/shark/examples/shark_inference/stable_diffusion$ python main.py --precision="fp16" --device="vulkan" --iree-vulkan-target-triple=rdna3-unknown-linux --no-load_vmfb --dispatch_benchmarks="all" --dispatch_benchmarks_dir="SD_dispatches" --dump_isa

Co-authored-by: alexander <alexander@nod-labs.com>
 2022-11-30 11:33:30 -08:00
Gaurav Shukla
a5445866b8 [WEB] Update the iree flag 
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-11-30 18:56:48 +05:30
powderluv
e8ded71a7b Default to 50 steps for SD 2022-11-29 16:45:23 -08:00
Prashant Kumar
a14c615def Update with the new flag. (#522) 2022-11-29 09:39:32 -08:00
Ean Garvey
41bf262482 Update SD README.md (#516) 
* Update README.md

* Create profiling_with_iree.md
 2022-11-29 10:21:28 -06:00
Prashant Kumar
6ee8f61fbe Add the stable diffusion v2 model. 
The f16 version of stable diffusion v2 model is added.
--version="v2" will run the v2 model.
 2022-11-29 18:18:04 +05:30
Prashant Kumar
3c4c4231ce Add new args. 2022-11-29 18:18:04 +05:30
Prashant Kumar
d0eef19eba Remove the lms versions as they were redundant. 
Tested with the DPM scheduler.
 2022-11-29 15:05:05 +05:30
Prashant Kumar
74aeb55733 Add support for different schedulers. 
Initial support for adding schedulers. This verifies the model running
with the PNDM scheduler too.
 2022-11-28 22:12:09 +05:30
aldesilv
f80b85f10c dump spv for dispatches (#509) 2022-11-23 22:34:27 -06:00
Quinn Dawkins
32a2ec432d [Stable Diffusion] Revive the tuned model (#506) 2022-11-23 15:42:24 -05:00
Prashant Kumar
fdf2aa54ef Update the sd models. 2022-11-22 23:09:04 +05:30
Phaneesh Barwaria
e67bcffea7 add vulkan-heap-block-size flag (#498) 2022-11-22 13:30:25 +05:30
Gaurav Shukla
d624940e12 Remove unnecessary torch_mlir import 
Signed-off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-11-21 21:47:21 +05:30
Prashant Kumar
88c58244b9 Update stable diffusion models to point to new location. 2022-11-18 19:39:21 +05:30
yzhang93
0754c6ea20 Update model annotation to take vulkan configs (#495) 
Co-authored-by: vivian <vivian@nod-labs.com>
 2022-11-17 14:34:17 -08:00
Prashant Kumar
7b1f04d121 Changes incorporating the recent torch_mlir compile api changes. 2022-11-15 15:25:37 +05:30
Phaneesh Barwaria
d8a9bee244 Add internet connection check for re-downloading models (#488) 2022-11-14 13:56:42 -06:00
Ean Garvey
45677c1e23 Install torch version required by torch-mlir when setting up importer venv. (#486) 2022-11-14 14:01:01 +05:30
Phaneesh Barwaria
d9f4a9954a modify to get correct target triple (#485) 2022-11-13 20:13:44 -08:00
mariecwhite
ec461a4456 Enable XLA compiler for TF models (#484) 2022-11-13 20:10:47 -08:00