Support b_scale: (#2350)

- extend pipeline v1 and v3
 - add instances
 - add tests
 - add example

Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>

[ROCm/composable_kernel commit: b01a27ff22]
This commit is contained in:
Enrico Degregori
2025-07-25 03:49:58 +02:00
committed by GitHub
parent c721559117
commit 2d68b3f9c0
24 changed files with 3744 additions and 1660 deletions

View File

@@ -72,7 +72,6 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_gemm_add_multiply.cpp)
list(APPEND PROFILER_OPS profile_gemm_bias_add_reduce.cpp)
list(APPEND PROFILER_OPS profile_gemm_splitk.cpp)
list(APPEND PROFILER_OPS profile_gemm_b_scale.cpp)
list(APPEND PROFILER_OPS profile_batched_gemm_b_scale.cpp)
list(APPEND PROFILER_OPS profile_gemm_universal_batched.cpp)
list(APPEND PROFILER_OPS profile_gemm_universal_reduce.cpp)
@@ -93,6 +92,7 @@ endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx11" OR SUPPORTED_GPU_TARGETS MATCHES "gfx12" OR SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_gemm_universal.cpp)
list(APPEND PROFILER_OPS profile_batched_gemm.cpp)
list(APPEND PROFILER_OPS profile_gemm_b_scale.cpp)
list(APPEND PROFILER_OPS profile_grouped_conv_fwd.cpp)
list(APPEND PROFILER_OPS profile_grouped_conv_bwd_data.cpp)
list(APPEND PROFILER_OPS profile_grouped_conv_bwd_weight.cpp)
@@ -178,7 +178,6 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND DEVICE_INSTANCES device_gemm_mx_instance)
endif()
list(APPEND DEVICE_INSTANCES device_gemm_splitk_instance)
list(APPEND DEVICE_INSTANCES device_gemm_b_scale_instance)
list(APPEND DEVICE_INSTANCES device_batched_gemm_b_scale_instance)
list(APPEND DEVICE_INSTANCES device_gemm_universal_batched_instance)
list(APPEND DEVICE_INSTANCES device_gemm_universal_reduce_instance)
@@ -208,6 +207,7 @@ endif()
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx1[12]")
list(APPEND DEVICE_INSTANCES device_gemm_universal_instance)
list(APPEND DEVICE_INSTANCES device_batched_gemm_instance)
list(APPEND DEVICE_INSTANCES device_gemm_b_scale_instance)
list(APPEND DEVICE_INSTANCES device_grouped_conv3d_fwd_instance)
list(APPEND DEVICE_INSTANCES device_grouped_conv2d_bwd_data_instance)
list(APPEND DEVICE_INSTANCES device_grouped_conv3d_bwd_data_instance)