Skip to content
Open
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions csrc/flash_attn/src/flash_fwd_kernel.h
Original file line number Diff line number Diff line change
Expand Up @@ -391,9 +391,9 @@ inline __device__ void compute_attn_1rowblock(const Params &params, const int bi
Tensor scores = make_tensor(acc_s.data(), flash::convert_layout_acc_rowcol(acc_s.layout()));

if (Is_attn_mask) {
flash::apply_attn_mask<Kernel_traits::TiledMma>(scores, tPgMask, tPcMask,
m_block == m_block_max - 1 ? m_residue : params.seqlen_q,
n_block == n_block_max - 1 ? n_residue : params.seqlen_k,
flash::apply_attn_mask<Kernel_traits::TiledMma>(scores, tPgMask, tPcMask,
params.seqlen_q,
params.seqlen_k,
params.unscale_softmax);
tPgMask.data() = tPgMask.data() + (-kBlockN);
}
Expand Down Expand Up @@ -519,8 +519,8 @@ inline __device__ void compute_attn_1rowblock(const Params &params, const int bi

if (Is_attn_mask) {
flash::apply_attn_mask<Kernel_traits::TiledMma>(scores, tPgMask, tPcMask,
m_block == m_block_max - 1 ? m_residue : params.seqlen_q,
n_block == n_block_max - 1 ? n_residue : params.seqlen_k,
params.seqlen_q,
params.seqlen_k,
params.unscale_softmax);
tPgMask.data() = tPgMask.data() + (-kBlockN);
}
Expand Down