diff --git a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp
index 121593d3cc..821bbb0051 100644
--- a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp
+++ b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v1_ab_scale.hpp
@@ -308,7 +308,7 @@ struct BlockwiseGemmXdlops_pipeline_v1_ab_scale<BlockGemmPipelineScheduler::Intr
                                 typename vector_type<ComputeDataType,
                                                      xdlops_gemm.K1PerXdlops>::type;
 
-                            xdlops_gemm.template Run(
+                            xdlops_gemm.template Run<>(
                                 a_thread_vec.template AsType<mfma_input_type>(),
                                 b_thread_vec.template AsType<mfma_input_type>(),
                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
@@ -390,9 +390,10 @@ struct BlockwiseGemmXdlops_pipeline_v1_ab_scale<BlockGemmPipelineScheduler::Intr
                         using mfma_input_type =
                             typename vector_type<ComputeDataType, xdlops_gemm.K1PerXdlops>::type;
 
-                        xdlops_gemm.template Run(a_thread_vec.template AsType<mfma_input_type>(),
-                                                 b_thread_vec.template AsType<mfma_input_type>(),
-                                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
+                        xdlops_gemm.template Run<>(
+                            a_thread_vec.template AsType<mfma_input_type>(),
+                            b_thread_vec.template AsType<mfma_input_type>(),
+                            c_thread_buf_per_scale.GetVectorTypeReference(I0));
                     });
                     static_for<0, xdlops_gemm.GetRegSizePerXdlops(), 1>{}([&](auto t) {
                         constexpr index_t c_offset =
diff --git a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp
index cb7cf605be..40fa776484 100644
--- a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp
+++ b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v2_ab_scale.hpp
@@ -350,7 +350,7 @@ struct BlockwiseGemmXdlops_pipeline_v2_ab_scale<BlockGemmPipelineScheduler::Intr
                                     typename vector_type<ComputeDataType,
                                                          xdlops_gemm.K1PerXdlops>::type;
 
-                                xdlops_gemm.template Run(
+                                xdlops_gemm.template Run<>(
                                     a_thread_vec.template AsType<mfma_input_type>(),
                                     b_thread_vec.template AsType<mfma_input_type>(),
                                     c_thread_buf_per_scale.GetVectorTypeReference(I0));
@@ -443,7 +443,7 @@ struct BlockwiseGemmXdlops_pipeline_v2_ab_scale<BlockGemmPipelineScheduler::Intr
                                 typename vector_type<ComputeDataType,
                                                      xdlops_gemm.K1PerXdlops>::type;
 
-                            xdlops_gemm.template Run(
+                            xdlops_gemm.template Run<>(
                                 a_thread_vec.template AsType<mfma_input_type>(),
                                 b_thread_vec.template AsType<mfma_input_type>(),
                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
@@ -518,9 +518,10 @@ struct BlockwiseGemmXdlops_pipeline_v2_ab_scale<BlockGemmPipelineScheduler::Intr
                         using mfma_input_type =
                             typename vector_type<ComputeDataType, xdlops_gemm.K1PerXdlops>::type;
 
-                        xdlops_gemm.template Run(a_thread_vec.template AsType<mfma_input_type>(),
-                                                 b_thread_vec.template AsType<mfma_input_type>(),
-                                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
+                        xdlops_gemm.template Run<>(
+                            a_thread_vec.template AsType<mfma_input_type>(),
+                            b_thread_vec.template AsType<mfma_input_type>(),
+                            c_thread_buf_per_scale.GetVectorTypeReference(I0));
                     });
                     static_for<0, xdlops_gemm.GetRegSizePerXdlops(), 1>{}([&](auto t) {
                         constexpr index_t c_offset =
@@ -575,9 +576,10 @@ struct BlockwiseGemmXdlops_pipeline_v2_ab_scale<BlockGemmPipelineScheduler::Intr
                         using mfma_input_type =
                             typename vector_type<ComputeDataType, xdlops_gemm.K1PerXdlops>::type;
 
-                        xdlops_gemm.template Run(a_thread_vec.template AsType<mfma_input_type>(),
-                                                 b_thread_vec.template AsType<mfma_input_type>(),
-                                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
+                        xdlops_gemm.template Run<>(
+                            a_thread_vec.template AsType<mfma_input_type>(),
+                            b_thread_vec.template AsType<mfma_input_type>(),
+                            c_thread_buf_per_scale.GetVectorTypeReference(I0));
                     });
                     static_for<0, xdlops_gemm.GetRegSizePerXdlops(), 1>{}([&](auto t) {
                         constexpr index_t c_offset =
diff --git a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp
index 66c9a5c339..de542866a6 100644
--- a/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp
+++ b/include/ck/tensor_operation/gpu/block/blockwise_gemm_pipeline_xdlops_v3_ab_scale.hpp
@@ -427,7 +427,7 @@ struct BlockwiseGemmXdlops_pipeline_v3_ab_scale<BlockGemmPipelineScheduler::Intr
                                 typename vector_type<ComputeDataType,
                                                      xdlops_gemm.K1PerXdlops>::type;
 
-                            xdlops_gemm.template Run(
+                            xdlops_gemm.template Run<>(
                                 a_thread_vec.template AsType<mfma_input_type>(),
                                 b_thread_vec.template AsType<mfma_input_type>(),
                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
@@ -504,9 +504,10 @@ struct BlockwiseGemmXdlops_pipeline_v3_ab_scale<BlockGemmPipelineScheduler::Intr
                         using mfma_input_type =
                             typename vector_type<ComputeDataType, xdlops_gemm.K1PerXdlops>::type;
 
-                        xdlops_gemm.template Run(a_thread_vec.template AsType<mfma_input_type>(),
-                                                 b_thread_vec.template AsType<mfma_input_type>(),
-                                                 c_thread_buf_per_scale.GetVectorTypeReference(I0));
+                        xdlops_gemm.template Run<>(
+                            a_thread_vec.template AsType<mfma_input_type>(),
+                            b_thread_vec.template AsType<mfma_input_type>(),
+                            c_thread_buf_per_scale.GetVectorTypeReference(I0));
                     });
                     static_for<0, xdlops_gemm.GetRegSizePerXdlops(), 1>{}([&](auto t) {
                         constexpr index_t c_offset =