ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-07-04 14:17:42 +00:00
Code Issues Packages Projects Releases Wiki Activity

This is work in progress, edited the template parameters in order to build

Browse Source 
(cherry picked from commit b4fde8a3314cb44659c4bbda35f1a0133c63dc41)

Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
This commit is contained in:
Apoorva Kalyani
2025-05-26 21:36:15 +00:00
committed by apoorva
parent 113ea09770
commit b129e731c3
4 changed files with 50 additions and 40 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
library/src/tensor_operation_instance/gpu/gemm_add/CMakeLists.txt
View File

@@ -1,4 +1,4 @@
# ONLY XDL_KERNELS
# ONLY XDL_AND_WMMA_KERNELS
add_instance_library(device_gemm_add_instance
device_gemm_add_xdl_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_instance.cpp
device_gemm_add_xdl_c_shuffle_bf16_i8_bf16_bf16_mk_kn_mn_mn_instance.cpp

4
library/src/tensor_operation_instance/gpu/gemm_add/device_gemm_add_wmma_c_shuffle_bf16_i8_bf16_bf16_mk_kn_mn_mn_instance.cpp
View File

@@ -27,7 +27,8 @@ using device_gemm_add_wmma_c_shuffle_bf16_i8_bf16_bf16_mk_kn_mn_mn_generic_insta
//################################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Stage| Size| Block| Block| Block| | WMMA| WMMA| | | ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//################################| | | | | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//################################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, BF16, I8, I32, F32, BF16_Tuple, BF16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 64, 128, 32, 8, 32, 32, 4, 2, S<4, 64, 1>, S<0, 2, 1>, S<1, 0, 2>, 2, 1, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 8, 1, 1, 1, S<1, 32, 1, 8>, 1>
// TODO: these template variables need to be adjusted
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, BF16, I8, I32, F32, BF16_Tuple, BF16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 64, 128, 32, 8, 32, 32, 1, 2, S<4, 64, 1>, S<0, 2, 1>, S<1, 0, 2>, 2, 1, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 8, 1, 1, 1, S<1, 32, 1, 8>, 1>
// clang-format on
>;
@@ -38,6 +39,7 @@ using device_gemm_add_wmma_c_shuffle_bf16_i8_bf16_bf16_mk_kn_mn_mn_instances = s
//################################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Stage| Size| Block| Block| Block| | WMMA| WMMA| | | ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//################################| | | | | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//################################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
// TODO: these template variables need to be adjusted
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, BF16, I8, F32, F32, BF16_Tuple, BF16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 16, 128, 32, 8, 16, 16, 1, 2, S<4, 16, 4>, S<1, 0, 2>, S<1, 0, 2>, 2, 2, 2, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 2, 8, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, BF16, I8, F32, F32, BF16_Tuple, BF16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 32, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 16, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, BF16, I8, F32, F32, BF16_Tuple, BF16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 64, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 16, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>

30
library/src/tensor_operation_instance/gpu/gemm_add/device_gemm_add_wmma_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_instance.cpp
View File

@@ -23,25 +23,27 @@ static constexpr auto GemmMNKPadding = ck::tensor_operation::device::GemmSpecial
using device_gemm_add_wmma_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_generic_instances = std::tuple<
// clang-format off
// M/N/K padding
//##############################| A| B| Ds| E| AData| BData| AccData| CShuffle| DsData| EData| A| B| CDE| GEMM| NumGemmK| Block| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##############################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| Size| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##############################| | | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##############################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32, F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 64, 128, 32, 8, 8, 32, 32, 1, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 1, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 8, 1, 1, 1, S<1, 32, 1, 8>, 1>
//################################| A| B| Ds| E| AData| BData| AccData| CShuffle| DsData| EData| A| B| CDE| GEMM| Prefetch| Block| MPer| NPer| K0Per| K1| MPer| NPer| MRepeat| NRepeat| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//################################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Stage| Size| Block| Block| Block| | WMMA| WMMA| | | ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//################################| | | | | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//################################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
// TODO: these template variables need to be adjusted
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32,F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 64, 128, 32, 8, 32, 32, 1, 2, S<4, 64, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 1, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 8, 1, 1, 1, S<1, 32, 1, 8>, 1>
// clang-format on
>;
using device_gemm_add_xdl_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_instances = std::tuple<
using device_gemm_add_wmma_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_instances = std::tuple<
// clang-format off
// M/N/K padding
//##############################| A| B| Ds| E| AData| BData| AccData| CShuffle| DsData| EData| A| B| CDE| GEMM| NumGemmK| Block| MPer| NPer| KPer| AK1| BK1| MPer| NPer| MXdl| NXdl| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//##############################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Prefetch| Size| Block| Block| Block| | | XDL| XDL| Per| Per| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//##############################| | | | | | | | | | | Operation| Operation| Operation| | Stage| | | | | | | | | Wave| Wave| Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//##############################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32, F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 16, 128, 32, 8, 8, 16, 16, 1, 2, S<4, 16, 4>, S<1, 0, 2>, S<1, 0, 2>, 2, 2, 2, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 2, 8, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32, F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 32, 8, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 16, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32, F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 64, 8, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 16, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>
// clang-format on
//################################| A| B| Ds| E| AData| BData| AccData| CShuffle| DsData| EData| A| B| CDE| GEMM| Prefetch| Block| MPer| NPer| K0Per| K1| MPer| NPer| MRepeat| NRepeat| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockTransfer| ABlockLds| BBlockTransfer| BBlockTransfer| BBlockTransfer| BlockTransfer| BBlockTransfer| BBlockTransfer| BBlockLds| CShuffle| CShuffle| CBlockTransferClusterLengths| CBlockTransfer|
//################################| Layout| Layout| Layout| Layout| Type| Type| Type| DataType| Type| Type| Elementwise| Elementwise| Elementwise| Specialization| Stage| Size| Block| Block| Block| | WMMA| WMMA| | | ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraM| ThreadCluster| ThreadCluster| SrcAccessOrder| SrcVectorDim| SrcScalar| DstScalar| AddExtraN| MXdlPerWave| NXdlPerWave| _MBlock_MWaveMPerXdl| ScalarPerVector|
//################################| | | | | | | | | | | Operation| Operation| Operation| | | | | | | | | | | | Lengths_K0_M_K1| ArrangeOrder| | | PerVector| PerVector_K1| | Lengths_K0_N_K1| ArrangeOrder| | | PerVector| PerVector_K1| | PerShuffle| PerShuffle| _NBlock_NWaveNPerXdl| _NWaveNPerXdl|
//################################| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
// TODO: these template variables need to be adjusted
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32,F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 256, 16, 128, 32, 8, 16, 16, 1, 2, S<4, 16, 4>, S<1, 0, 2>, S<1, 0, 2>, 2, 2, 2, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 8, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32,F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 32, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>,
DeviceGemmMultipleD_Wmma_CShuffle< Row, Row, Row_Tuple, Row, F16, I8, F32, F32,F16_Tuple, F16, PassThrough, PassThrough, Add, GemmMNKPadding, 1, 64, 16, 16, 64, 8, 16, 16, 1, 1, S<4, 16, 1>, S<1, 0, 2>, S<1, 0, 2>, 2, 8, 8, 1, S<4, 64, 1>, S<0, 2, 1>, S<0, 2, 1>, 1, 1, 1, 1, 1, 1, S<1, 16, 1, 4>, 1, LoopScheduler::Default, PipelineVersion::v1>
// clang-format on
>;
void add_device_gemm_add_wmma_c_shuffle_f16_i8_f16_f16_mk_kn_mn_mn_instances(

54
profiler/src/CMakeLists.txt
View File

@@ -40,19 +40,21 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND PROFILER_OPS profile_contraction_scale.cpp)
endif()
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
list(APPEND PROFILER_OPS profile_gemm_reduce.cpp)
list(APPEND PROFILER_OPS profile_batched_gemm_gemm.cpp)
list(APPEND PROFILER_OPS profile_batched_gemm_add_relu_gemm_add.cpp)
list(APPEND PROFILER_OPS profile_gemm_add.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm.cpp)
list(APPEND PROFILER_OPS profile_gemm_streamk.cpp)
list(APPEND PROFILER_OPS profile_gemm_add_relu.cpp)
list(APPEND PROFILER_OPS profile_gemm_add_silu.cpp)
list(APPEND PROFILER_OPS profile_gemm_add_relu_add_layernorm.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm_fixed_nk.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm_fastgelu.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm_tile_loop.cpp)
list(APPEND PROFILER_OPS profile_grouped_gemm_multiply_tile_loop.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_reduce.cpp)
list(APPEND PROFILER_SOURCES profile_batched_gemm_gemm.cpp)
list(APPEND PROFILER_SOURCES profile_batched_gemm_add_relu_gemm_add.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add_add_fastgelu.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add_fastgelu.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_gemm.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_streamk.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_fastgelu.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add_relu.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add_silu.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add_relu_add_layernorm.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_gemm_fixed_nk.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_gemm_fastgelu.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_gemm_tile_loop.cpp)
list(APPEND PROFILER_SOURCES profile_grouped_gemm_multiply_tile_loop.cpp)
endif()
list(APPEND PROFILER_OPS profile_gemm_multiply_add.cpp)
if(SUPPORTED_GPU_TARGETS MATCHES "gfx9[45]")
@@ -92,6 +94,7 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx1[1
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
list(APPEND PROFILER_OPS profile_gemm_fastgelu.cpp)
list(APPEND PROFILER_OPS profile_gemm_add_add_fastgelu.cpp)
list(APPEND PROFILER_SOURCES profile_gemm_add.cpp)
endif()
endif()
@@ -147,17 +150,19 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9")
list(APPEND DEVICE_INSTANCES device_contraction_scale_instance)
endif()
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
list(APPEND DEVICE_INSTANCES device_gemm_add_instance)
list(APPEND DEVICE_INSTANCES device_batched_gemm_gemm_instance)
list(APPEND DEVICE_INSTANCES device_batched_gemm_add_relu_gemm_add_instance)
list(APPEND DEVICE_INSTANCES device_grouped_gemm_instance)
list(APPEND DEVICE_INSTANCES device_gemm_streamk_instance)
list(APPEND DEVICE_INSTANCES device_gemm_add_relu_instance)
list(APPEND DEVICE_INSTANCES device_gemm_add_silu_instance)
list(APPEND DEVICE_INSTANCES device_gemm_add_relu_add_layernorm_instance)
list(APPEND DEVICE_INSTANCES device_grouped_gemm_fixed_nk_instance)
list(APPEND DEVICE_INSTANCES device_grouped_gemm_fastgelu_instance)
list(APPEND DEVICE_INSTANCES device_grouped_gemm_tile_loop_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_add_fastgelu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_fastgelu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_gemm_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_batched_gemm_add_relu_gemm_add_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_streamk_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_fastgelu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_silu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_relu_add_layernorm_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fixed_nk_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_fastgelu_instance)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_grouped_gemm_tile_loop_instance)
endif()
list(APPEND DEVICE_INSTANCES device_batched_gemm_instance)
list(APPEND DEVICE_INSTANCES device_batched_gemm_reduce_instance)
@@ -203,6 +208,7 @@ if(SUPPORTED_GPU_TARGETS MATCHES "gfx9" OR SUPPORTED_GPU_TARGETS MATCHES "gfx1[1
list(APPEND DEVICE_INSTANCES device_grouped_conv3d_bwd_weight_instance)
list(APPEND DEVICE_INSTANCES device_gemm_add_fastgelu_instance)
if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
target_link_libraries(${PROFILER_EXECUTABLE} PRIVATE device_gemm_add_instance)
list(APPEND DEVICE_INSTANCES device_gemm_fastgelu_instance)
list(APPEND DEVICE_INSTANCES device_gemm_add_add_fastgelu_instance)
endif()