diff options
| author | Laurent Mazare <laurent.mazare@gmail.com> | 2024-11-12 17:10:12 +0100 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2024-11-12 17:10:12 +0100 |
| commit | 06350c31c780d6ea485f506032aea6ff8809e38a (patch) | |
| tree | 811d3bd8443335b93f17d547d3abb5a9b3fc6ffd /candle-metal-kernels | |
| parent | 9453cc30958dd0e9209aaeba30b15bb97aff0ea9 (diff) | |
| download | candle-06350c31c780d6ea485f506032aea6ff8809e38a.tar.gz candle-06350c31c780d6ea485f506032aea6ff8809e38a.tar.bz2 candle-06350c31c780d6ea485f506032aea6ff8809e38a.zip | |
Add some missing index-select metal kernels. (#2613)
* Add some missing index-select metal kernels.
* Make some matrix contiguous pre-matmul.
Diffstat (limited to 'candle-metal-kernels')
| -rw-r--r-- | candle-metal-kernels/src/indexing.metal | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/candle-metal-kernels/src/indexing.metal b/candle-metal-kernels/src/indexing.metal index 9eee97ca..c14f2c1f 100644 --- a/candle-metal-kernels/src/indexing.metal +++ b/candle-metal-kernels/src/indexing.metal @@ -193,12 +193,16 @@ INDEX_OP(is_i64_f16, int64_t, half) INDEX_OP(is_i64_bf16, int64_t, bfloat) #endif +INDEX_OP(is_u32_u8, uint32_t, uint8_t) +INDEX_OP(is_u32_u32, uint32_t, uint32_t) INDEX_OP(is_u32_f32, uint32_t, float) INDEX_OP(is_u32_f16, uint32_t, half) #if defined(__HAVE_BFLOAT__) INDEX_OP(is_u32_bf16, uint32_t, bfloat) #endif +INDEX_OP(is_u8_u8, uint8_t, uint8_t) +INDEX_OP(is_u8_u32, uint8_t, uint32_t) INDEX_OP(is_u8_f32, uint8_t, float) INDEX_OP(is_u8_f16, uint8_t, half) #if defined(__HAVE_BFLOAT__) |