Skip to content

[CI/Build] tests(v1): feed Triton attention the (num_blocks, 2, …) KV cache layout in backend-correctness tests #26663

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

Merged
merged 3 commits into from
Oct 18, 2025
Merged

[CI/Build] tests(v1): feed Triton attention the (num_blocks, 2, …) KV cache layout in backend-correctness tests #26663

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
81e81de
fix_kv_cache_unbind
hl475 Oct 12, 2025
54baf13
Merge branch 'main' into fix_kv_cache_unbind
yeqcharlotte Oct 14, 2025
722828c
Merge branch 'main' into fix_kv_cache_unbind
hl475 Oct 16, 2025
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
5 changes: 3 additions & 2 deletions tests/v1/attention/test_attention_backends.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -423,13 +423,14 @@ def _test_backend_correctness(
for backend_name in backend_to_test:
# FlashAttentionm + FlexAttention:
# [2, num_blocks, block_size, num_kv_heads, head_size]
# FlashInfer:
# FlashInfer + Triton:
# [num_blocks, 2, block_size, num_kv_heads, head_size]
# Select the appropriate KV cache format for each backend
kv_cache_for_backend = kv_cache
if backend_name == _Backend.FLASHINFER:
if backend_name in (_Backend.FLASHINFER, _Backend.TRITON_ATTN):
kv_cache_for_backend = kv_cache.transpose(0, 1)

if backend_name == _Backend.FLASHINFER:
# For FlashInfer default to HND layout and
kv_cache_for_backend = (
kv_cache_for_backend.transpose(2, 3).contiguous().transpose(2, 3)
Expand Down