Merge branch 'develop' into add_fp16_wmma_conv_instance

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gkm_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gkn_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gkn_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gnk_gmn_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/batched_gemm_multi_d/device_batched_gemm_multi_d_dl_i8_i8_i8_gmk_gnk_gmn_irregular_instance.cpp

 // SPDX-License-Identifier: MIT
-// Copyright (c) 2018-2022, Advanced Micro Devices, Inc. All rights reserved.
+// Copyright (c) 2018-2023, Advanced Micro Devices, Inc. All rights reserved.
 
 #include <cstdlib>
 

library/src/tensor_operation_instance/gpu/conv2d_bwd_data/CMakeLists.txt

-add_instance_library(device_conv2d_bwd_data_instance
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp
-   device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp
-
-   device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f16_instance.cpp
-   device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f32_instance.cpp
-   device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_int8_instance.cpp
-) 
+set(CONV2D_BWD_DATA_INSTANCES)
+if(DTYPES MATCHES "fp32" OR NOT DEFINED DTYPES)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f32_instance.cpp)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f32_instance.cpp)
+endif()
+if(DTYPES MATCHES "bf16" OR NOT DEFINED DTYPES)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_bf16_instance.cpp)
+endif()
+if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_f16_instance.cpp)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_f16_instance.cpp)
+endif()
+if(DTYPES MATCHES "int8" OR NOT DEFINED DTYPES)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp)
+   list(APPEND CONV2D_BWD_DATA_INSTANCES device_conv2d_bwd_data_dl_nhwc_kyxc_nhwk_int8_instance.cpp)
+endif()
+add_instance_library(device_conv2d_bwd_data_instance ${CONV2D_BWD_DATA_INSTANCES})

library/src/tensor_operation_instance/gpu/conv2d_bwd_data/device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instance.cpp

@@ -11,7 +11,7 @@
 #include "ck/tensor_operation/gpu/element/element_wise_operation.hpp"
 
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -151,3 +151,4 @@ void add_device_conv2d_bwd_data_xdl_nhwc_kyxc_nhwk_int8_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/CMakeLists.txt

-add_instance_library(device_gemm_instance
-   device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp
-   device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_km_kn_mn_instance.cpp
-   device_gemm_xdl_f16_f16_f16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp
-   device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp
-   device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instance.cpp
-   device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp
-   device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp
-   device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp
-   device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp
+set(GEMM_INSTANCES)
+if(DTYPES MATCHES "fp64" OR NOT DEFINED DTYPES)
+    list(APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_mk_kn_mn_instance.cpp)
+    list(APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_mk_nk_mn_instance.cpp)
+    list(APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_km_kn_mn_instance.cpp)
+    list(APPEND GEMM_INSTANCES device_gemm_xdl_f64_f64_f64_km_nk_mn_instance.cpp)
+endif()
+if(DTYPES MATCHES "fp32" OR NOT DEFINED DTYPES)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f32_f32_f32_km_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f32_f32_f32_km_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f32_f32_f32_km_nk_mn_instance.cpp)
+endif()
+if(DTYPES MATCHES "fp16" OR NOT DEFINED DTYPES)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_f16_f16_f16_km_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_2_stage_f16_f16_f16_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_kn_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_mk_nk_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_kn_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_f16_f16_f16_km_nk_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_add_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_kn_mn_irregular_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_add_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/km_nk_mn_irregular_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_add_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_kn_mn_irregular_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_add_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_interwave_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v1_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_default_pipeline_v2_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_f16_f16_f16/mk_nk_mn_irregular_interwave_pipeline_v1_instance.cpp)
+endif()
+if(DTYPES MATCHES "int8" OR NOT DEFINED DTYPES)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp)
+endif()
+if(DTYPES MATCHES "bf16" OR NOT DEFINED DTYPES)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_mk_nk_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_kn_mn_instance.cpp)
+   list(APPEND GEMM_INSTANCES device_gemm_xdl_c_shuffle_bf16_bf16_bf16_km_nk_mn_instance.cpp)
+endif()
+
+add_instance_library(device_gemm_instance ${GEMM_INSTANCES})
+
+set(ENABLE_PIPELINE_V2_OPT OFF)
+
+if (ENABLE_PIPELINE_V2_OPT)
+set(MAX_ILP_OPTS
+   -mllvm
+   -amdgpu-enable-max-ilp-scheduling-strategy
 )
+set(WAVES_PER_EU_DEFS
+   CK_USE_WAVES_PER_EU=1
+   CK_MIN_WAVES_PER_EU=1
+   CK_MAX_WAVES_PER_EU=1
+)
+set(IGLP_OPT_DEFS
+   CK_EXPERIMENTAL_PIPELINE_V2_IGLP_OPT=1
+)
+
+# layout=NT
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/km_kn_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS ";;"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+# layout=NN
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/km_nk_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+# layout=TT
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/mk_kn_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS}")
+# layout=TN
+set_source_files_properties(device_gemm_xdl_f16_f16_f16/mk_nk_mn_default_pipeline_v2_opt_instance.cpp PROPERTIES
+   COMPILE_OPTIONS "${MAX_ILP_OPTS}"
+   COMPILE_DEFINITIONS "${WAVES_PER_EU_DEFS};${IGLP_OPT_DEFS}")
+endif(ENABLE_PIPELINE_V2_OPT)

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_kn_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -80,3 +80,4 @@ void add_device_gemm_dl_i8_i8_i8_km_kn_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -78,3 +78,4 @@ void add_device_gemm_dl_i8_i8_i8_km_kn_mn_irregular_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_nk_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -80,3 +80,4 @@ void add_device_gemm_dl_i8_i8_i8_km_nk_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -78,3 +78,4 @@ void add_device_gemm_dl_i8_i8_i8_km_nk_mn_irregular_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_kn_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -80,3 +80,4 @@ void add_device_gemm_dl_i8_i8_i8_mk_kn_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -78,3 +78,4 @@ void add_device_gemm_dl_i8_i8_i8_mk_kn_mn_irregular_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_nk_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -80,3 +80,4 @@ void add_device_gemm_dl_i8_i8_i8_mk_nk_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_dl.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -78,3 +78,4 @@ void add_device_gemm_dl_i8_i8_i8_mk_nk_mn_irregular_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl_cshuffle.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -66,3 +66,4 @@ void add_device_gemm_xdl_c_shuffle_i8_i8_i8_km_kn_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl_cshuffle.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -66,3 +66,4 @@ void add_device_gemm_xdl_c_shuffle_i8_i8_i8_km_nk_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif

library/src/tensor_operation_instance/gpu/gemm/device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instance.cpp

@@ -8,7 +8,7 @@
 #include "ck/tensor_operation/gpu/device/gemm_specialization.hpp"
 #include "ck/tensor_operation/gpu/device/impl/device_gemm_xdl_cshuffle.hpp"
 #include "ck/library/tensor_operation_instance/add_device_operation_instance.hpp"
-
+#ifdef __int8__
 namespace ck {
 namespace tensor_operation {
 namespace device {
@@ -66,3 +66,4 @@ void add_device_gemm_xdl_c_shuffle_i8_i8_i8_mk_kn_mn_instances(
 } // namespace device
 } // namespace tensor_operation
 } // namespace ck
+#endif