Skip to content

GitLab

Commit b60595f9 authored by letaoqin's avatar letaoqin
Browse files

add to device

parent 83d13bff
Expand all Hide whitespace changes
Inline Side-by-side
Showing with 21 additions and 15 deletions
include/ck/tensor_operation/gpu/device/impl/device_batched_mha_bwd_xdl_cshuffle_qloop_v2.hpp
View file @ b60595f9
...@@ -278,6 +278,7 @@ template <index_t NumDimG, ...@@ -278,6 +278,7 @@ template <index_t NumDimG,
index_t BBlockTransferSrcScalarPerVector, index_t BBlockTransferSrcScalarPerVector,
index_t BBlockTransferDstScalarPerVector_BK1, index_t BBlockTransferDstScalarPerVector_BK1,
bool BBlockLdsExtraN, bool BBlockLdsExtraN,
index_t D0BlockTransferSrcScalarPerVector,
typename B1BlockTransferThreadClusterLengths_BK0_N_BK1, typename B1BlockTransferThreadClusterLengths_BK0_N_BK1,
typename B1BlockTransferThreadClusterArrangeOrder, typename B1BlockTransferThreadClusterArrangeOrder,
typename B1BlockTransferSrcAccessOrder, typename B1BlockTransferSrcAccessOrder,
...@@ -302,7 +303,7 @@ struct DeviceBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2 ...@@ -302,7 +303,7 @@ struct DeviceBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2
static constexpr index_t NumAcc1Bias = Acc1BiasDataType::Size(); static constexpr index_t NumAcc1Bias = Acc1BiasDataType::Size();
// TODO: implement bias combination // TODO: implement bias combination
static_assert(NumAcc0Bias == 0 && NumAcc0Bias == 0, "Bias addition is unimplemented"); static_assert(NumAcc1Bias == 0, "Bias addition is unimplemented");
using DeviceOp = DeviceBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2; using DeviceOp = DeviceBatchedMultiheadAttentionBackward_Qloop_Xdl_CShuffle_V2;
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment