tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-12 11:27:43 -04:00
Code Issues Packages Projects Releases Wiki Activity

HIP: Add support for RDNA4 targets (#12372)

Browse Source
This commit is contained in:
Slobodan Josic
2025-03-26 23:46:30 +01:00
committed by GitHub
parent b3298fa47a
commit bd40678df7
7 changed files with 25 additions and 17 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/build.md
View File

@@ -191,7 +191,7 @@ The following compilation options are also available to tweak performance:
| Option | Legal values | Default | Description |
|-------------------------------|------------------------|---------|-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| GGML_CUDA_FORCE_MMQ | Boolean | false | Force the use of custom matrix multiplication kernels for quantized models instead of FP16 cuBLAS even if there is no int8 tensor core implementation available (affects V100, RDNA3). MMQ kernels are enabled by default on GPUs with int8 tensor core support. With MMQ force enabled, speed for large batch sizes will be worse but VRAM consumption will be lower. |
| GGML_CUDA_FORCE_MMQ | Boolean | false | Force the use of custom matrix multiplication kernels for quantized models instead of FP16 cuBLAS even if there is no int8 tensor core implementation available (affects V100, CDNA and RDNA3+). MMQ kernels are enabled by default on GPUs with int8 tensor core support. With MMQ force enabled, speed for large batch sizes will be worse but VRAM consumption will be lower. |
| GGML_CUDA_FORCE_CUBLAS | Boolean | false | Force the use of FP16 cuBLAS instead of custom matrix multiplication kernels for quantized models |
| GGML_CUDA_F16 | Boolean | false | If enabled, use half-precision floating point arithmetic for the CUDA dequantization + mul mat vec kernels and for the q4_1 and q5_1 matrix matrix multiplication kernels. Can improve performance on relatively recent GPUs. |
| GGML_CUDA_PEER_MAX_BATCH_SIZE | Positive integer | 128 | Maximum batch size for which to enable peer access between multiple GPUs. Peer access requires either Linux or NVLink. When using NVLink enabling peer access for larger batch sizes is potentially beneficial. |