From 98c58fd94bb8bce406ef07fb6c8564015f1fff09 Mon Sep 17 00:00:00 2001 From: Jin Zhou Date: Mon, 8 Sep 2025 02:46:41 +0000 Subject: [PATCH] refine --- ...device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/include/ck/tensor_operation/gpu/device/impl/device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp b/include/ck/tensor_operation/gpu/device/impl/device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp index 55e8929237..faaea60fea 100644 --- a/include/ck/tensor_operation/gpu/device/impl/device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp +++ b/include/ck/tensor_operation/gpu/device/impl/device_gemm_multiple_d_xdl_cshuffle_v3_b_preshuffle.hpp @@ -324,6 +324,7 @@ struct DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle Run(kernel); } } + else { if(GridwiseGemm::CalculateKBlockLoopTailNum(K_split) == TailNumber::Odd) { @@ -336,6 +337,7 @@ struct DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle TailNumber::Odd>; Run(kernel); } + else { const auto kernel = kernel_gemm_xdl_cshuffle_v3_multi_d_b_preshuffle_2lds< @@ -353,10 +355,12 @@ struct DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle throw std::runtime_error("todo: only v1 v2 and v3 support now"); } } +#if 0 else { if constexpr(BlkGemmPipelineVer == BlockGemmPipelineVersion::v1) { +#if 0 if(arg.KBatch > 1) { if(GridwiseGemm::CalculateKBlockLoopTailNum(K_split) == TailNumber::Odd) @@ -403,6 +407,7 @@ struct DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle Run(kernel); } } +#endif if(GridwiseGemm::CalculateKBlockLoopTailNum(K_split) == TailNumber::Odd) { const auto kernel = kernel_gemm_xdl_cshuffle_v3_multi_d_b_preshuffle< @@ -482,7 +487,7 @@ struct DeviceGemmMultiD_Xdl_CShuffle_V3_BPreshuffle throw std::runtime_error("todo: only v3 support now"); } } - +#endif return ave_time; }