init method revert and reformat some files

example/65_gemm_multiply_multiply/gemm_fp16int8_b_scale.cpp

@@ -217,8 +217,7 @@ int main(int argc, char* argv[])
         a0_m_k.GenerateTensorValue(GeneratorTensor_1<A0DataType>{});
         quant_b0_k_n.GenerateTensorValue(GeneratorTensor_1<QuantDataType>{});
         // a1_m_k.GenerateTensorValue(GeneratorTensor_1<A1DataType>{});
-        // b1_k_n.GenerateTensorValue(GeneratorTensor_1<B1DataType>{});
-        b1_k_n.GenerateTensorValue(GeneratorTensor_3<B1DataType>{0, 1.0});
+        b1_k_n.GenerateTensorValue(GeneratorTensor_1<B1DataType>{});
         break;
     case 3:
         a0_m_k.GenerateTensorValue(GeneratorTensor_2<A0DataType>{-2, 2});

include/ck/tensor_operation/gpu/grid/gridwise_gemm_xdl_cshuffle_v3_multi_d_b_scale.hpp

@@ -1306,28 +1306,28 @@ struct GridwiseGemmMultiD_BScale_xdl_cshuffle_v3
 
         // B matrix blockwise copy
         auto b_blockwise_copy = ThreadGroupTensorSliceTransfer_v4r1_b_scale<
-            ThisThreadBlock,
-            BElementwiseOperation,
-            ck::tensor_operation::element_wise::PassThrough,
-            InMemoryDataOperationEnum::Set,
-            Sequence<BK0Number, NPerBlock, BK1Number>,
-            BBlockTransferThreadClusterLengths_BK0_N_BK1,
-            BBlockTransferThreadClusterArrangeOrder,
-            BDataType,
-            LDSTypeB,
-            decltype(b_grid_desc_bk0_n_bk1),
-            decltype(b_block_desc_bk0_n_bk1),
-            BBlockTransferSrcAccessOrder,
-            Sequence<0, 1, 2>,
-            BBlockTransferSrcVectorDim,
-            2,
-            BBlockTransferSrcScalarPerVector,
-            BBlockTransferDstScalarPerVector_BK1,
-            1,
-            1,
-            BThreadTransferSrcResetCoordinateAfterRun,
-            true,
-            BlockwiseGemmPipe::GlobalBufferNum>(b_grid_desc_bk0_n_bk1,
+                                                ThisThreadBlock,
+                                                BElementwiseOperation,
+                                                ck::tensor_operation::element_wise::PassThrough,
+                                                InMemoryDataOperationEnum::Set,
+                                                Sequence<BK0Number, NPerBlock, BK1Number>,
+                                                BBlockTransferThreadClusterLengths_BK0_N_BK1,
+                                                BBlockTransferThreadClusterArrangeOrder,
+                                                BDataType,
+                                                LDSTypeB,
+                                                decltype(b_grid_desc_bk0_n_bk1),
+                                                decltype(b_block_desc_bk0_n_bk1),
+                                                BBlockTransferSrcAccessOrder,
+                                                Sequence<0, 1, 2>,
+                                                BBlockTransferSrcVectorDim,
+                                                2,
+                                                BBlockTransferSrcScalarPerVector,
+                                                BBlockTransferDstScalarPerVector_BK1,
+                                                1,
+                                                1,
+                                                BThreadTransferSrcResetCoordinateAfterRun,
+                                                true,
+                                                BlockwiseGemmPipe::GlobalBufferNum>(b_grid_desc_bk0_n_bk1,
                                                 make_multi_index(0, n_block_data_idx_on_grid, 0),
                                                 b_element_op,
                                                 b_block_desc_bk0_n_bk1,