mirror of
https://github.com/Comfy-Org/comfy-kitchen.git
synced 2026-11-04 23:37:42 +00:00
[PR #186] [MERGED] Support 256 head dim in Flash attention kernel. #204
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
Comfy-Org/comfy-kitchen#204
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
📋 Pull Request Information
Original PR: https://github.com/Comfy-Org/comfy-kitchen/pull/186
Author: @comfyanonymous
Created: 9/18/2026
Status: ✅ Merged
Merged: 9/24/2026
Merged by: @comfyanonymous
Base:
main← Head:dev/fa256📝 Commits (2)
9d20fd2Flash attention 256 head dim.a098d2bFix flash attention is_available returning True on CPU device.📊 Changes
7 files changed (+195 additions, -124 deletions)
View changed files
📝
comfy_kitchen/backends/cuda/dlpack_bindings.cpp(+7 -6)📝
comfy_kitchen/backends/cuda/ops/flash_decode.cu(+21 -15)📝
comfy_kitchen/backends/hip/dlpack_bindings.cpp(+7 -7)📝
comfy_kitchen/backends/hip/launchers.h(+2 -2)📝
comfy_kitchen/backends/hip/ops/flash_decode.hip(+78 -60)📝
comfy_kitchen/flash_attention.py(+11 -5)📝
tests/test_flash_attention.py(+69 -29)📄 Description
No description provided
🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.