diff options
| author | Johannes Gäßler <johannesg@5d6.de> | 2024-06-10 11:45:13 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-06-10 11:45:13 +0200 |
| commit | 1f0dabda8d5c131f9d4632aa41de74317cdd61fb (patch) | |
| tree | 70f8c54d4752f196616adf6081498d8f92992ec0 /examples | |
| parent | af4ae502ddaeb03cd5861273ca2e9a5ae4551db7 (diff) | |
CUDA: use tensor cores for MMQ (#7676)
* CUDA: int8 tensor cores for MMQ (legacy quants)
* fix out-of-bounds writes
* __builtin_assume -> GGML_CUDA_ASSUME
* fix writeback returning too early
Diffstat (limited to 'examples')
0 files changed, 0 insertions, 0 deletions