Updates for CUTLASS 3.5.0 (#1468)

This commit is contained in:
Vijay Thakkar
2024-04-11 21:33:40 -04:00
committed by GitHub
parent a40e08e9d5
commit 7d49e6c7e2
171 changed files with 7526 additions and 1888 deletions
+7
View File
@@ -140,14 +140,19 @@ if (CUTLASS_NVCC_MAX_ARCH GREATER_EQUAL 80)
conv2d_fprop_implicit_gemm_cf32nhwc_cf32nhwc_cf32nhwc_simt_f32_sm80.cu
conv2d_dgrad_implicit_gemm_cf32nhwc_cf32nhwc_cf32nhwc_simt_f32_sm80.cu
conv2d_wgrad_implicit_gemm_cf32nhwc_cf32nhwc_cf32nhwc_simt_f32_sm80.cu
deconv2d_implicit_gemm_f32nhwc_f32nhwc_f32nhwc_simt_f32_sm80.cu
conv2d_fprop_with_broadcast_simt_sm80.cu
deconv2d_with_broadcast_simt_sm80.cu
conv3d_fprop_implicit_gemm_f32ndhwc_f32ndhwc_f32ndhwc_simt_f32_sm80.cu
conv3d_dgrad_implicit_gemm_f32ndhwc_f32ndhwc_f32ndhwc_simt_f32_sm80.cu
conv3d_wgrad_implicit_gemm_f32ndhwc_f32ndhwc_f32ndhwc_simt_f32_sm80.cu
deconv3d_implicit_gemm_f32ndhwc_f32ndhwc_f32ndhwc_simt_f32_sm80.cu
conv3d_fprop_with_broadcast_simt_sm80.cu
deconv3d_with_broadcast_simt_sm80.cu
)
endif()
@@ -176,6 +181,7 @@ cutlass_test_unit_add_executable(
conv2d_fprop_with_broadcast_sm75.cu
conv2d_fprop_with_reduction_sm75.cu
conv3d_fprop_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm75.cu
conv3d_wgrad_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm75.cu
)
@@ -209,6 +215,7 @@ if (CUTLASS_NVCC_MAX_ARCH GREATER_EQUAL 80)
conv2d_strided_dgrad_implicit_gemm_swizzling4_sm80.cu
# Conv3d
conv3d_fprop_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm80.cu
conv3d_wgrad_implicit_gemm_f16ndhwc_f16ndhwc_f32ndhwc_tensor_op_f32_sm80.cu
# Group Conv2d