Lumpiasty/llama.cpp
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Revert "CUDA: size routed MoE MMQ N-tiles from typical expert width on RDNA3 (#24546)" (#28551)

Browse Source 
This reverts commit 0c963452ea.

Assisted-by: Claude Fable 5.1
Claude-Session: https://claude.ai/code/session_01Q7rfnkjzgnfhvJsdeXhdoH
This commit is contained in:
Piotr Wilkin (ilintar)
2026-09-07 16:28:19 +02:00
committed by GitHub
parent ccc3646c63
commit e71b80510c
10 changed files with 12 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/mmq-config-blackwell.cuh
-1
View File
@@ -1,5 +1,4 @@
static constexpr __host__ __device__ ggml_cuda_mmq_config ggml_cuda_mmq_get_config_blackwell(ggml_type type, int J, bool fallback) {
constexpr bool use_typical_moe_ncols = false;
CASE(GGML_TYPE_MXFP4, 256, 1, 128, 8, GGML_CUDA_MMQ_SRAM_LAYOUT_FP4, MMQ_ITER_K_FP4, true, true);
CASE(GGML_TYPE_MXFP4, 256, 1, 128, 16, GGML_CUDA_MMQ_SRAM_LAYOUT_FP4, MMQ_ITER_K_FP4, true, true);
CASE(GGML_TYPE_MXFP4, 256, 1, 128, 32, GGML_CUDA_MMQ_SRAM_LAYOUT_FP4, MMQ_ITER_K_FP4, true, true);