Merge pull request #1838 from ROCm/cka8w8_uc_newpipe

Cka8w8 uc newpipe

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_compute_default_instance.cpp

+// SPDX-License-Identifier: MIT
+// Copyright (c) 2024, Advanced Micro Devices, Inc. All rights reserved.
+
+#include "device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn.hpp"
+
+namespace ck {
+namespace tensor_operation {
+namespace device {
+namespace instance {
+
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_compute_default_instances(
+    std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
+                                                                     Col,
+                                                                     Tuple<Row, Col>,
+                                                                     Row,
+                                                                     F8,
+                                                                     F8,
+                                                                     Tuple<F32, F32>,
+                                                                     F16,
+                                                                     PassThrough,
+                                                                     PassThrough,
+                                                                     MultiplyMultiply>>>& instances)
+{
+    add_device_operation_instances(
+        instances,
+        device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_compute_instances<
+            GemmDefault>{});
+}
+
+} // namespace instance
+} // namespace device
+} // namespace tensor_operation
+} // namespace ck

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_default_instance.cpp

@@ -8,7 +8,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_default_instances(
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_default_instances_v2(
     std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
                                                                      Col,
                                                                      Tuple<Row, Col>,
@@ -24,6 +24,7 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_instances<
+            v2,
             GemmDefault>{});
 }
 

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_padding_instance.cpp → library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_default_instance_v2.cpp

@@ -8,7 +8,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_padding_instances(
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_default_instances(
     std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
                                                                      Col,
                                                                      Tuple<Row, Col>,
@@ -24,7 +24,8 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p1_instances<
-            GemmKPadding>{});
+            v1,
+            GemmDefault>{});
 }
 
 } // namespace instance

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_default_instance.cpp

@@ -24,6 +24,7 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_instances<
+            v1,
             GemmDefault>{});
 }
 

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_padding_instance.cpp → library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_default_instance_v2.cpp

@@ -8,7 +8,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_padding_instances(
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_default_instances_v2(
     std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
                                                                      Col,
                                                                      Tuple<Row, Col>,
@@ -24,7 +24,8 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p2_instances<
-            GemmKPadding>{});
+            v2,
+            GemmDefault>{});
 }
 
 } // namespace instance

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_default_instance.cpp

@@ -24,6 +24,7 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_instances<
+            v1,
             GemmDefault>{});
 }
 

library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_padding_instance.cpp → library/src/tensor_operation_instance/gpu/gemm_multiply_multiply_weight_preshuffle/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16/device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_default_instance_v2.cpp

@@ -8,7 +8,7 @@ namespace tensor_operation {
 namespace device {
 namespace instance {
 
-void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_padding_instances(
+void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_default_instances_v2(
     std::vector<std::unique_ptr<DeviceGemmMultipleDSplitKBPreShuffle<Row,
                                                                      Col,
                                                                      Tuple<Row, Col>,
@@ -24,7 +24,8 @@ void add_device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_m
     add_device_operation_instances(
         instances,
         device_gemm_multiply_multiply_weight_preshuffle_xdl_f8_f8_f16_mk_mfma_mn_p3_instances<
-            GemmKPadding>{});
+            v2,
+            GemmDefault>{});
 }
 
 } // namespace instance

script/cmake-ck-dev.sh

@@ -17,7 +17,7 @@ fi
 cmake                                                                                             \
 -D CMAKE_PREFIX_PATH=/opt/rocm                                                                    \
 -D CMAKE_CXX_COMPILER=/opt/rocm/bin/hipcc                                                         \
--D CMAKE_CXX_FLAGS="-Xclang -mllvm -Xclang -enable-post-misched=0 -std=c++17 -O3 -ftemplate-backtrace-limit=0  -fPIE  -Wno-gnu-line-marker"     \
+-D CMAKE_CXX_FLAGS="-std=c++17 -O3 -ftemplate-backtrace-limit=0  -fPIE  -Wno-gnu-line-marker"     \
 -D CMAKE_BUILD_TYPE=Release                                                                       \
 -D BUILD_DEV=ON                                                                                   \
 -D GPU_TARGETS=$GPU_TARGETS                                                                       \