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

Remove deprecated cli option iree-hal-cuda-disable-loop-nounroll-wa (#1235)

Browse Source
This commit is contained in:
Thomas
2023-03-22 10:05:15 -07:00
committed by GitHub
parent 593428bda4
commit 807947c0c8
5 changed files with 4 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
shark/iree_utils/gpu_utils.py
View File

@@ -30,11 +30,10 @@ def get_iree_gpu_args():
in ["sm_70", "sm_72", "sm_75", "sm_80", "sm_84", "sm_86", "sm_89"]
) and (shark_args.enable_tf32 == True):
return [
"--iree-hal-cuda-disable-loop-nounroll-wa",
f"--iree-hal-cuda-llvm-target-arch={sm_arch}",
]
else:
return ["--iree-hal-cuda-disable-loop-nounroll-wa"]
return []
# Get the default gpu args given the architecture.

2
tank/examples/MiniLM_tf/huggingface_MiniLM_run.py
View File

@@ -70,7 +70,7 @@ if __name__ == "__main__":
backend_config = "dylib"
# backend = "cuda"
# backend_config = "cuda"
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-hal-cuda-disable-loop-nounroll-wa", "--iree-enable-fusion-with-reduction-ops"]
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-enable-fusion-with-reduction-ops"]
flatbuffer_blob = compile_str(
compiler_module,
target_backends=[backend],

1
tank/examples/bert_fine_tuning/bert_fine_tune_tf.py
View File

@@ -146,7 +146,6 @@ if __name__ == "__main__":
backend_config = "cuda"
args = [
"--iree-cuda-llvm-target-arch=sm_80",
"--iree-hal-cuda-disable-loop-nounroll-wa",
"--iree-enable-fusion-with-reduction-ops",
]

2
tank/examples/bert_tf/bert_large_run.py
View File

@@ -91,7 +91,7 @@ if __name__ == "__main__":
backend_config = "dylib"
# backend = "cuda"
# backend_config = "cuda"
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-hal-cuda-disable-loop-nounroll-wa", "--iree-enable-fusion-with-reduction-ops"]
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-enable-fusion-with-reduction-ops"]
flatbuffer_blob = compile_str(
compiler_module,
target_backends=[backend],

2
tank/examples/bert_tf/bert_small_run.py
View File

@@ -86,7 +86,7 @@ if __name__ == "__main__":
backend_config = "dylib"
# backend = "cuda"
# backend_config = "cuda"
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-hal-cuda-disable-loop-nounroll-wa", "--iree-enable-fusion-with-reduction-ops"]
# args = ["--iree-cuda-llvm-target-arch=sm_80", "--iree-enable-fusion-with-reduction-ops"]
flatbuffer_blob = compile_str(
compiler_module,
target_backends=[backend],