diff options
author | Johannes Gäßler <johannesg@5d6.de> | 2023-06-19 10:23:56 +0200 |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-06-19 10:23:56 +0200 |
commit | 16b9cd193965769089881bb8ec012fccca7b37b6 (patch) | |
tree | 2ee329793e782f253966fd81f89ea05f5a1a2495 /README.md | |
parent | b24c3049d96557c24782e4d32feaae65f47277af (diff) |
Convert vector to f16 for dequantize mul mat vec (#1913)
* Convert vector to f16 for dmmv
* compile option
* Added compilation option description to README
* Changed cmake CUDA_ARCHITECTURES from "OFF" to "native"
Diffstat (limited to 'README.md')
-rw-r--r-- | README.md | 9 |
1 files changed, 8 insertions, 1 deletions
@@ -337,7 +337,14 @@ Building the program with BLAS support may lead to some performance improvements cmake --build . --config Release ``` - The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. + The environment variable [`CUDA_VISIBLE_DEVICES`](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#env-vars) can be used to specify which GPU(s) will be used. The following compilation options are also available to tweak performance: + + | Option | Legal values | Default | Description | + |-------------------------|------------------------|---------|-------------| + | LLAMA_CUDA_DMMV_X | Positive integer >= 32 | 32 | Number of values in x direction processed by the CUDA dequantization + matrix vector multiplication kernel per iteration. Increasing this value can improve performance on fast GPUs. Power of 2 heavily recommended. Does not affect k-quants. | + | LLAMA_CUDA_DMMV_Y | Positive integer | 1 | Block size in y direction for the CUDA dequantization + mul mat vec kernels. Increasing this value can improve performance on fast GPUs. Power of 2 recommended. Does not affect k-quants. | + | LLAMA_CUDA_DMMV_F16 | Boolean | false | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels. Can improve performance on relatively recent GPUs. | + | LLAMA_CUDA_KQUANTS_ITER | 1 or 2 | 2 | Number of values processed per iteration and per CUDA thread for Q2_K and Q6_K quantization formats. Setting this value 2 1 can improve performance for slow GPUs. | - #### CLBlast |