[gfx110x] support Navi3x architectures. (#628)

* enable building on Nav31

* fix syntax

* replace GPU_TARGETS with offload-arch

* add gfx1102 rachitecture

* fix typo

* update changelog

[ROCm/composable_kernel commit: 0ccecc7c31]
This commit is contained in:
Illia Silin
2023-03-09 05:56:40 -08:00
committed by GitHub
parent 0d23b0d1c9
commit 9ce65cae0e
13 changed files with 27 additions and 26 deletions

View File

@@ -38,7 +38,7 @@ add_example_executable_no_testing(example_gemm_xdl_fp64 gemm_xdl_fp64.cpp)
add_dependencies(example_gemm_xdl example_gemm_xdl_skip_b_lds_fp16)
add_dependencies(example_gemm_xdl example_gemm_xdl_fp64)
if(GPU_TARGETS MATCHES "gfx1100")
if(GPU_TARGETS MATCHES "gfx1100" OR GPU_TARGETS MATCHES "gfx1101" OR GPU_TARGETS MATCHES "gfx1102")
add_custom_target(example_gemm_wmma)
add_example_executable(example_gemm_wmma_fp16 gemm_wmma_fp16.cpp)
add_dependencies(example_gemm_wmma example_gemm_wmma_fp16)

View File

@@ -1,4 +1,4 @@
add_example_executable(example_gemm_bilinear_xdl_fp16 gemm_bilinear_xdl_fp16.cpp)
if(GPU_TARGETS MATCHES "gfx1100")
if(GPU_TARGETS MATCHES "gfx1100" OR GPU_TARGETS MATCHES "gfx1101" OR GPU_TARGETS MATCHES "gfx1102")
add_example_executable(example_gemm_bilinear_wmma_fp16 gemm_bilinear_wmma_fp16.cpp)
endif()

View File

@@ -1,5 +1,5 @@
add_example_executable(example_batched_gemm_bias_e_permute_xdl_fp16 batched_gemm_bias_e_permute_xdl_fp16.cpp)
if(GPU_TARGETS MATCHES "gfx1100")
if(GPU_TARGETS MATCHES "gfx1100" OR GPU_TARGETS MATCHES "gfx1101" OR GPU_TARGETS MATCHES "gfx1102")
add_example_executable(example_batched_gemm_bias_e_permute_wmma_fp16 batched_gemm_bias_e_permute_wmma_fp16.cpp)
endif()

View File

@@ -16,7 +16,7 @@ if(USE_BITINT_EXTENSION_INT4)
add_dependencies(example_grouped_conv_fwd_multiple_d example_grouped_conv_fwd_bias_relu_add_xdl_int4)
endif() # USE_BITINT_EXTENSION_INT4
if(GPU_TARGETS MATCHES "gfx1100")
if(GPU_TARGETS MATCHES "gfx1100" OR GPU_TARGETS MATCHES "gfx1101" OR GPU_TARGETS MATCHES "gfx1102")
add_example_executable(example_grouped_conv_fwd_bias_relu_add_wmma_fp16 grouped_conv_fwd_bias_relu_add_wmma_fp16.cpp)
endif()