Skip to content

Navigation Menu

Explore
By company size
By use case
By industry
View all solutions
Topics
- AI
- DevOps
- Security
- Software Development
- View all
Explore
- GitHub Sponsors
  Fund open source developers
- The ReadME Project
  GitHub community articles
Repositories
- Enterprise platform
  AI-powered developer platform
Available add-ons
Pricing

Search code, repositories, users, issues, pull requests...

Search

Clear

Search syntax tips

Provide feedback

We read every piece of feedback, and take your input very seriously.

Include my email address so I can be contacted

Saved searches

Use saved searches to filter your results more quickly

Name

Query

To see all available qualifiers, see our documentation.

You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session. You switched accounts on another tab or window. Reload to refresh your session.

Dismiss alert

ggml-org / llama.cpp Public

Notifications You must be signed in to change notification settings
Fork 10.9k
Star 75.3k

Code
Issues 344
Pull requests 387
Discussions
Actions
Projects 9
Wiki
Security
Insights

Additional navigation options

Code
Issues
Pull requests
Discussions
Actions
Projects
Wiki
Security
Insights

Add GGML_HIP_ROCWMMA_FATTN to enable rocWMMA for FlashAttention#12032

Open

hjc4869 wants to merge 11 commits intoggml-org:masterggml-org/llama.cpp:masterfrom hjc4869:prhjc4869/llama.cpp:prCopy head branch name to clipboard

+58-21

Conversation Commits11 (11)Checks Files changed

Commits

Commits on Feb 22, 2025

Add GGML_HIP_ROCWMMA_FATTN and rocwmma header check
hjc4869
committed
Add rocWMMA support
hjc4869
committed
Merge branch 'master' into pr
hjc4869
authored

Commits on Feb 23, 2025

Update ggml/src/ggml-hip/CMakeLists.txt

hjc4869
and
JohannesGaessler
authored
Move comments to reduce confusion.
hjc4869
committed
Use namespace alias wmma instead of lots of ifdefs.
hjc4869
committed
Fix: FP16_MMA_AVAILABLE should not be checked in host code.
hjc4869
committed

Commits on Feb 25, 2025

Always return false in fp16_mma_available when compiling for HIP and GGML_HIP_ROCWMMA_FATTN is disabled.
hjc4869
committed
Remove the Q->ne[1] > 8 check
hjc4869
committed
Also always return false in fp16_mma_hardware_available when compiled for AMD and GGML_HIP_ROCWMMA_FATTN not enabled.
hjc4869
committed
Revert "Also always return false in fp16_mma_hardware_available when compiled for AMD and GGML_HIP_ROCWMMA_FATTN not enabled."
hjc4869
committed

Footer

© 2025 GitHub, Inc.

Footer navigation

Terms
Privacy
Security
Status
Docs
Contact

You can’t perform that action at this time.