Update GetTypeString function to generate unique kernel IDs. (#638)

* make conv_fwd_bias_activation kernel id unique

* add more parameters to conv and gemm kernel names

* update GetTypeString for conv and gemm kernels

* fix two more kernel strings

include/ck/tensor_operation/gpu/device/impl/device_gemm_xdl_splitk_c_shuffle.hpp

@@ -640,7 +640,16 @@ struct DeviceGemmXdlSplitKCShuffle : public DeviceGemmSplitK<ALayout,
             << BlockSize << ", "
             << MPerBlock << ", "
             << NPerBlock << ", "
-            << K0PerBlock
+            << K0PerBlock << ", "
+            << K1 << ", "
+            << MPerXDL << ", "
+            << NPerXDL << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << ABlockTransferDstScalarPerVector_K1 << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferDstScalarPerVector_K1
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_data_multiple_d_xdl_cshuffle_v1.hpp

@@ -1003,7 +1003,15 @@ struct DeviceGroupedConvBwdDataMultipleD_Xdl_CShuffle_v1
             << KPerBlock << ", "
             << AK1 << ", "
             << BK1 << ", "
-            << getConvBackwardDataSpecializationString(ConvBackwardDataSpecialization)
+            << getConvBackwardDataSpecializationString(ConvBackwardDataSpecialization) << ", "
+            << MPerXDL << ", "
+            << NPerXDL << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle
             << ">";
 
         return str.str();

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_gnwc_gkxc_gnwk_dl.hpp

@@ -1203,7 +1203,8 @@ struct DeviceGroupedConvBwdWeightGnwcGkxcGnwk_Dl
             << MPerBlock << ", "
             << NPerBlock << ", "
             << K0PerBlock << ", "
-            << getConvBackwardWeightSpecializationString(ConvBackwardWeightSpecialization)
+            << getConvBackwardWeightSpecializationString(ConvBackwardWeightSpecialization) << ", "
+            << K1
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_bwd_weight_gnwc_gkxc_gnwk_xdl_cshuffle.hpp

@@ -1231,7 +1231,17 @@ struct DeviceGroupedConvBwdWeightGnwcGkxcGnwk_Xdl_CShuffle
             << MPerBlock << ", "
             << NPerBlock << ", "
             << K0PerBlock << ", "
-            << getConvBackwardWeightSpecializationString(ConvBackwardWeightSpecialization)
+            << getConvBackwardWeightSpecializationString(ConvBackwardWeightSpecialization) << ", "
+            << K1 << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << ABlockTransferDstScalarPerVector_K1 << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferDstScalarPerVector_K1 << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle << ", "
+            << CBlockTransferScalarPerVector_NWaveNPerXdl
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_d_multiple_r_xdl_cshuffle.hpp

@@ -1092,7 +1092,15 @@ struct DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle
             << MPerBlock << ", "
             << NPerBlock << ", "
             << KPerBlock << ", "
-            << getConvForwardSpecializationString(ConvForwardSpecialization)
+            << getConvForwardSpecializationString(ConvForwardSpecialization) << ", "
+            << MPerXDL << ", "
+            << NPerXDL << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_d_wmma_cshuffle.hpp

@@ -838,7 +838,19 @@ struct DeviceGroupedConvFwdMultipleD_Wmma_CShuffle
             << MPerBlock << ", "
             << NPerBlock << ", "
             << KPerBlock << ", "
-            << getConvForwardSpecializationString(ConvForwardSpecialization)
+            << getConvForwardSpecializationString(ConvForwardSpecialization) << ", "
+            << K1 << ", "
+            << MPerXDL << ", "
+            << NPerXDL << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << ABlockTransferDstScalarPerVector_K1 << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferDstScalarPerVector_K1 << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle << ", "
+            << CBlockTransferScalarPerVector_NWaveNPerXdl
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_conv_fwd_multiple_d_xdl_cshuffle.hpp

@@ -939,7 +939,15 @@ struct DeviceGroupedConvFwdMultipleD_Xdl_CShuffle
             << MPerBlock << ", "
             << NPerBlock << ", "
             << KPerBlock << ", "
-            << getConvForwardSpecializationString(ConvForwardSpecialization)
+            << getConvForwardSpecializationString(ConvForwardSpecialization) << ", "
+            << MPerXDL << ", "
+            << NPerXDL << ", "
+            << MXdlPerWave << ", "
+            << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle
             << ">";
         // clang-format on
 

include/ck/tensor_operation/gpu/device/impl/device_grouped_gemm_xdl.hpp

@@ -688,6 +688,10 @@ struct DeviceGroupedGemm_Xdl : public DeviceGroupedGemm<ALayout,
             << NPerXDL << ", "
             << MXdlPerWave << ", "
             << NXdlPerWave << ", "
+            << ABlockTransferSrcScalarPerVector << ", "
+            << BBlockTransferSrcScalarPerVector << ", "
+            << CShuffleMXdlPerWavePerShuffle << ", "
+            << CShuffleNXdlPerWavePerShuffle << ", "
             << getGemmSpecializationString(GemmSpec)
             << ">";
         // clang-format on