Use the instance I want to optimize

example/01_gemm/CMakeLists.txt

@@ -17,6 +17,7 @@ endif(USE_BITINT_EXTENSION_INT4)
 add_custom_target(example_gemm_xdl)
 
 add_example_executable(example_gemm_xdl_fp16 gemm_xdl_fp16.cpp)
+set_source_files_properties(example_gemm_xdl_fp16 gemm_xdl_fp16.cpp PROPERTIES COMPILE_OPTIONS ";--save-temps;-Wno-gnu-line-marker;")
 add_example_executable(example_gemm_xdl_wavelet_fp16 gemm_xdl_wavelet_fp16.cpp)
 add_example_executable(example_gemm_xdl_bf16 gemm_xdl_bf16.cpp)
 add_example_executable(example_gemm_xdl_int8 gemm_xdl_int8.cpp)

example/01_gemm/common.hpp

@@ -24,20 +24,20 @@
 
 struct ProblemSize final
 {
-    ck::index_t M = 3840;
-    ck::index_t N = 4096;
-    ck::index_t K = 4096;
+    ck::index_t M = 832;
+    ck::index_t N = 1024;
+    ck::index_t K = 1024;
 
-    ck::index_t StrideA = 4096;
-    ck::index_t StrideB = 4096;
-    ck::index_t StrideC = 4096;
+    ck::index_t StrideA = 832;
+    ck::index_t StrideB = 1024;
+    ck::index_t StrideC = 1024;
 };
 
 struct ExecutionConfig final
 {
     bool do_verification = true;
     int init_method      = 1;
-    bool time_kernel     = false;
+    bool time_kernel     = true;
 };
 
 template <ck::index_t... Is>

example/01_gemm/gemm_xdl_fp16.cpp

@@ -30,7 +30,7 @@ using DeviceGemmInstance0 = ck::tensor_operation::device::DeviceGemmXdl
 // ######|      Type|      Type|      Type|        Type|        |        |        | Elementwise| Elementwise| Elementwise|Spacialization|  Size| Block| Block| Block|   |  XDL|  XDL|  Per|  Per|   ThreadCluster|  ThreadCluster| SrcAccessOrder|   SrcVectorDim|      SrcScalar|      DstScalar| AddExtraM|   ThreadCluster|  ThreadCluster| SrcAccessOrder|  SrcVectorDim|      SrcScalar|      DstScalar| AddExtraN| SrcDstVectorDim|       DstScalar|
 // ######|          |          |          |            |        |        |        |   Operation|   Operation|   Operation|              |      |      |      |      |   |     |     | Wave| Wave| Lengths_K0_M_K1|   ArrangeOrder|               |               |      PerVector|   PerVector_K1|          | Lengths_K0_N_K1|   ArrangeOrder|               |              |      PerVector|   PerVector_K1|          |                |       PerVector|
 // ######|          |          |          |            |        |        |        |            |            |            |              |      |      |      |      |   |     |     |     |     |                |               |               |               |               |               |          |                |               |               |              |               |               |          |                |                |
-         < ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,   GemmDefault,   256,   256,   128,     4,  8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1>;
+         < ADataType, BDataType, CDataType, AccDataType, ALayout, BLayout, CLayout,  AElementOp,  BElementOp,  CElementOp,   GemmDefault,   256,    64,   128,     8,  8,   32,   32,    1,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,      true,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,      true,               7,               1,           1,  ck::LoopScheduler::Default,        ck::PipelineVersion::v2>;
 // // clang-format on
 
 // clang-format off
@@ -42,7 +42,7 @@ using DeviceGemmInstance1 = ck::tensor_operation::device::DeviceGemm_Xdl_CShuffl
          < ALayout, BLayout, CLayout, ADataType, BDataType, CDataType, AccDataType, CShuffleDataType,  AElementOp,  BElementOp,  CElementOp,    GemmDefault,        1,   256,   256,   128,    32,   8,   8,   32,   32,    4,    2,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,              2,              8,              8,         1,     S<4, 64, 1>,     S<1, 0, 2>,     S<1, 0, 2>,             2,              8,              8,         1,           1,           1,               S<1, 32, 1, 8>,               8>;
 // clang-format on
 
-using DeviceGemmInstance = DeviceGemmInstance1;
+using DeviceGemmInstance = DeviceGemmInstance0;
 
 using ReferenceGemmInstance = ck::tensor_operation::host::
     ReferenceGemm<ADataType, BDataType, CDataType, AccDataType, AElementOp, BElementOp, CElementOp>;