Skip to content

Fix FMHA bwd_nl kernel issue for batch_size == 3 #1975

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
crcrpar wants to merge 1 commit into
base: master
Choose a base branch
from
Open

Fix FMHA bwd_nl kernel issue for batch_size == 3 #1975

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
12 changes: 8 additions & 4 deletions apex/contrib/fmha/fmha.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,9 +34,13 @@ class FMHAFun(torch.autograd.Function):
@staticmethod
def forward(ctx, qkv, cu_seqlens, p_dropout, max_s, is_training, zero_tensors):
batch_size = cu_seqlens.numel() - 1
if batch_size < 4:
# Note: bwd_nl (no-loop backward) only works correctly for batch_size <= 2.
# For batch_size == 3, bwd_nl produces incorrect results due to num_chunks
# configuration issue in the CUDA kernel, so we use the regular path instead.
use_nl = batch_size <= 2
if use_nl:
max_s = 512
context, S_dmask = mha.fwd_nl(
context, S_dmask = mha.fwd(
qkv, cu_seqlens, p_dropout, max_s, is_training, True, zero_tensors, None
)
else:
Expand All @@ -55,13 +59,13 @@ def forward(ctx, qkv, cu_seqlens, p_dropout, max_s, is_training, zero_tensors):
ctx.p_dropout = p_dropout
ctx.max_s = max_s
ctx.zero_tensors = zero_tensors
ctx.use_nl = use_nl
return context

@staticmethod
def backward(ctx, dout):
qkv, S_dmask = ctx.saved_tensors
batch_size = ctx.cu_seqlens.numel() - 1
if batch_size < 4:
if ctx.use_nl:
dqkv, dp, _ = mha.bwd_nl(
dout,
qkv,
Expand Down
8 changes: 6 additions & 2 deletions apex/contrib/test/fmha/test_fmha.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -91,7 +91,11 @@ def run_test(self, s: int, b: int, zero_tensors: bool):

qkv.requires_grad = True

if b < 4:
# Note: bwd_nl (no-loop backward) only works correctly for batch_size <= 2.
# For batch_size == 3, bwd_nl produces incorrect results due to num_chunks
# configuration issue in the CUDA kernel, so we use the regular bwd instead.
use_nl = b <= 2
if use_nl:
ctx, S_ = mha.fwd(qkv_vs, cu_seqlens, 0.0, s, True, True, zero_tensors, None)
else:
ctx, S_ = mha.fwd(qkv_vs, cu_seqlens, 0.0, s, True, False, zero_tensors, None)
Expand All @@ -109,7 +113,7 @@ def run_test(self, s: int, b: int, zero_tensors: bool):

dw2 = dw.permute(0, 2, 1, 3).clone().detach().contiguous()

if b < 4:
if use_nl:
dqkv2, _, _ = mha.bwd_nl(dw2, qkv_vs, S_, cu_seqlens, 0.0, s, zero_tensors)
else:
dqkv2, _ = mha.bwd(dw2, qkv_vs, S_, cu_seqlens, 0.0, s, zero_tensors)
Expand Down