Tianlei Wu 72186bbb71 [CUDA] Build nhwc ops by default (#22648) ... ### Description * Build cuda nhwc ops by default. * Deprecate `--enable_cuda_nhwc_ops` in build.py and add `--disable_cuda_nhwc_ops` option Note that it requires cuDNN 9.x. If you build with cuDNN 8, NHWC ops will be disabled automatically. ### Motivation and Context In general, NHWC is faster than NCHW for convolution in Nvidia GPUs with Tensor Cores, and this could improve performance for vision models. This is the first step to prefer NHWC for CUDA in 1.21 release. Next step is to do some tests on popular vision models. If it help in most models and devices, set `prefer_nhwc=1` as default cuda provider option.		2024-11-06 09:54:55 -08:00
..
github	[CUDA] Build nhwc ops by default (#22648)	2024-11-06 09:54:55 -08:00
requirements	Replace inline pip install with pip install from requirements*.txt (#21106)	2024-07-22 12:39:10 -07:00
__init__.py
amd_hipify.py	fix issue when build with hipblasLt on rocm6.1 (#22553)	2024-10-28 13:57:08 +08:00
build.py	[CUDA] Build nhwc ops by default (#22648)	2024-11-06 09:54:55 -08:00
compile_triton.py
coverage.py
gen_def.py	Initial WebGPU EP checkin (#22318)	2024-10-08 16:10:46 -07:00
get_docker_image.py	Fix docker image layer caching to avoid redundant docker building and transient connection exceptions. (#21612)	2024-08-06 21:37:09 +08:00
logger.py
op_registration_utils.py	Bump ruff to 0.3.2 and black to 24 (#19878)	2024-03-13 10:00:32 -07:00
op_registration_validator.py	Bump ruff to 0.3.2 and black to 24 (#19878)	2024-03-13 10:00:32 -07:00
patch_manylinux.py
policheck_exclusions.xml
reduce_op_kernels.py	Update ruff and clang-format versions (#21479)	2024-07-24 11:50:11 -07:00
replace_urls_in_deps.py	Update ruff and clang-format versions (#21479)	2024-07-24 11:50:11 -07:00
set-trigger-rules.py	Initial WebGPU EP checkin (#22318)	2024-10-08 16:10:46 -07:00
update_tsaoptions.py
upload_python_package_to_azure_storage.py	Update ruff and clang-format versions (#21479)	2024-07-24 11:50:11 -07:00