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
1,745 Commits 66 Branches 705 Tags
debug
Commit Graph

161 Commits

Author SHA1 Message Date
PhaneeshB
ee0009d4b8 pythonize uname for cpu target triple in windows 2023-01-12 22:39:49 +05:30
PhaneeshB
9d851c3346 small fixes 2023-01-12 22:32:24 +05:30
George Petterson
6ad9b213b9 Add GCN4 
(cherry picked from commit 3be072b3c09c9b38bc2d79ad6e6900eefee49a1c)
 2023-01-09 21:09:50 +05:30
PhaneeshB
e4375e8195 Add support for vulkan target env 2023-01-09 21:09:50 +05:30
jinchen62
017dcab685 Add target triple support for TITAN RTX (#756) 2023-01-04 15:39:00 -08:00
Abhishek Varma
e60b4568c6 [SharkInference] Make SharkInference compile the entire module (#708) 
* [SharkInference] Make SharkInference compile the entire module

-- Previously SharkInference was compiling and providing run APIs
   for a harcoded function with function name "forward".
-- This commit makes the compiling functionality generic and now
   any function being defined within the module can be run.
-- It also creates an API to fetch all the function names defined
   within the compiled module.
-- This commit updates both web and command-line execution of Stable
   Diffusion to use new API of  SharkInference.

Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
 2023-01-03 23:25:23 +05:30
powderluv
cc6fbdb0c3 Add sm_89 and point to nvcuda.dll (#731) 2022-12-26 10:54:38 -08:00
Stanley Winata
5cf4976054 [Vulkan][utils] Add GTX Pascal support. (#709) 2022-12-22 15:24:15 -08:00
PhaneeshB
2befe771b3 Add support for automatic target triple selection for SD 2022-12-21 22:38:06 +05:30
Stella Laurenzo
10630ab597 Add config stanza for NVIDIA RTX 2080. (#658) 
Just happened to have this card on my Windows machine and verified that the SD demo works on it.

```
Average step time: 144.26142692565918ms/it
Clip Inference Avg time (ms) = (205.001 + 44.000) / 2 = 124.501
VAE Inference time (ms): 281.001

Total image generation time: 7.856997728347778sec
```

I'd love to add an API upstream to derive compiler tuning flags from a host device.
 2022-12-18 16:40:47 -08:00
Quinn Dawkins
2bc6de650d [SD] Add support for a compiled version of the discrete Euler scheduler (#657) 
* Add Shark version of euler scheduler

* Add Shark version of euler scheduler to web ui
 2022-12-17 19:25:43 -08:00
Phaneesh Barwaria
831f206cd0 Revert "Add target triple selection for multiple cards" (#655) 
This reverts commit acb905f0cc.
 2022-12-16 15:01:45 -08:00
PhaneeshB
acb905f0cc Add target triple selection for multiple cards 2022-12-17 02:24:37 +05:30
nirvedhmeshram
2928179331 Add more NVIDIA targets (#640) 2022-12-15 11:24:38 -06:00
Boian Petkantchin
bc17c29b2e In get_iree_runtime_config get the specific device instead of the default 2022-12-13 13:21:51 -08:00
Boian Petkantchin
aaf60bdee6 Simplify iree_device_map 2022-12-13 13:21:51 -08:00
Stanley Winata
57c94f8f80 [vulkan] Add "radeon" check to the default AMD triple (#604) 2022-12-10 09:05:48 -08:00
Ean Garvey
0225292a44 Remove print statements from compile utils (#593) 2022-12-08 13:40:47 -08:00
Ean Garvey
1699db79b5 Disable SHARK-Runtime flags if USE_IREE=1 specified during setup. (#588) 
* Disable SHARK-Runtime flags if USE_IREE=1 specified during setup.

* Update setup_venv.sh

* Autodetect cpu count for runtime flags.
 2022-12-08 02:31:31 -06:00
Ean Garvey
40eea21863 Enable conv nchw-to-nhwc flag by default for most models + minor fixes (#584) 2022-12-07 16:24:02 -08:00
Stanley Winata
6049f86bc4 [Vulkan][Utils] Automatic platform/OS detection (#569) 
To enable AMD gpus on macOS, we need this detection to let the compiler know that we would be needing moltenVK to use this GPU.
 2022-12-07 12:05:00 +07:00
Stanley Winata
c4444ff695 [vulkan][utils] Add rdna3 detection (#565) 2022-12-05 23:56:06 -08:00
Quinn Dawkins
e19a97f316 Don't do a numpy copy on the results from compiled vm (#543) 2022-12-05 14:21:47 -05:00
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
Phaneesh Barwaria
e67bcffea7 add vulkan-heap-block-size flag (#498) 2022-11-22 13:30:25 +05:30
Phaneesh Barwaria
d9f4a9954a modify to get correct target triple (#485) 2022-11-13 20:13:44 -08:00
Mehdi Amini
a526f7d5b8 Fix dispatch saving code after 749a2c2d (#483) 
In 749a2c2d iree_device_map and iree_target_map have been made functions
but not all of the uses were updated.
 2022-11-14 05:39:01 +05:30
Phaneesh Barwaria
749a2c2dec add support for choosing vulkan device (#439) 2022-11-12 14:00:41 -08:00
yzhang93
9a86e5c476 Fix dispatch benchmarking tool (#460) 2022-11-08 09:37:12 -08:00
Eliasj42
32d3f4bd5f added ordered benchmarks to dispatch benchmarking tool (#450) 
* added ordered benchmarks to dispatch benchmarking tool

* saved changes

* updated readme

Co-authored-by: Elias Joseph <elias@nod-labs.com>
 2022-11-07 09:36:21 -08:00
Gaurav Shukla
25931d48a3 [WEB] Update stable diffusion UI and enable live preview (#447) 
This commit enables live preview feature and also updates stable
diffusion web UI.

Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>

Signed-off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-10-31 04:10:15 -07:00
powderluv
fd89b06641 Drop RDNA1 for now 2022-10-29 14:29:09 -07:00
Gaurav Shukla
f8dc996004 Update vulkan-target-triple for Radeon devices. (#446) 
Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>

Signed-off-by: Gaurav Shukla <gaurav@nod-labs.com>
 2022-10-29 14:27:20 -07:00
Phaneesh Barwaria
e6a964088b Add os agnostic vulkan device name check (#445) 2022-10-29 13:19:14 -07:00
Eliasj42
7f37599a60 Added a dispatch benchmarking tool (#441) 
To produce benchmarks of individual dispatches, you can add --dispatch_benchmarks=All --dispatch_benchmarks_dir=<output_dir> to your command line argument.

Co-authored-by: Elias Joseph <elias@nod-labs.com>
 2022-10-28 14:31:03 -07:00
Ean Garvey
fd7baae548 Serialize torch-mlir CAPI module as bytecode instead of string. (#435) 
* Serialize torch-mlir CAPI as bytecode instead of string.

* Minor fixes to MLIR data handling in SHARK python.
 2022-10-27 14:37:15 -05:00
PhaneeshB
fd578a48a9 add cli args for vulkan target triple 2022-10-25 21:47:26 +05:30
Quinn Dawkins
1d33913d48 Add option to save and load precompiled flatbuffer (#425) 2022-10-23 16:24:09 -07:00
Quinn Dawkins
7be1d7d0be Add option for extra arguments through SharkInference.compile (#408) 2022-10-19 15:32:48 -05:00
gpetters94
53df0620e3 Add OPT to tank (#214) 2022-10-11 11:03:56 -05:00
Ean Garvey
d82b305781 Fix issues with loading .vmfb into SharkInference 2022-09-23 09:53:13 +05:30
Ean Garvey
814a6f8295 Modify vulkan target triple substring searches. (#318) 2022-09-20 01:20:20 -05:00
erman-gurses
fc8aa6ae63 Add ROCM parameters (#335) 2022-09-16 09:12:19 -07:00
Vivek Khandelwal
c43448a826 Update compile_utils.py 2022-09-15 18:28:10 +05:30
Ean Garvey
6cf5564c84 Remove "gpu" device alias and migrate to using "cuda" for NVIDIA GPU. (#325) 
* Replace instances of "gpu" alias for devices with "cuda"
 2022-09-13 01:16:56 -05:00
Stanley Winata
55bcb2eb3c Level Zero Backend (#280) 2022-08-17 19:19:27 -07:00
Ean Garvey
22ff92c48b Add config.VmModule argument to from_flatbuffer call. (#266) 2022-08-14 15:11:19 -07:00
powderluv
db6e2207ed Update _common.py 2022-08-13 13:49:01 -07:00
Daniel Garvey
7975087ee2 change backend name (#265) 2022-08-13 12:01:12 -07:00