Commit 4f285b35 authored by Tri Dao's avatar Tri Dao
Browse files

FlashAttention-2 release

parent 6d48e14a
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
// Copyright (c) 2022, Tri Dao.
// Splitting the different head dimensions to different files to speed up compilation.
#include "fmha_bwd_launch_template.h"
void run_fmha_bwd_hdim128(FMHA_dgrad_params &params, cudaStream_t stream, const bool configure) {
FP16_SWITCH(params.is_bf16, ([&] {
using Kernel_traits = FMHA_kernel_traits<128, 128, 16, 1, 8, 0x100u, elem_type>;
run_fmha_bwd_loop<Kernel_traits>(params, stream, configure);
}));
}
\ No newline at end of file
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
This diff is collapsed.
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