whisper.cpp/ggml/src/ggml-opencl
Hongqiang Wang 4673c4bc32 opencl: route large q6_K lm_head to the flat GEMV (llama/26427)
* add a direct size condition for `large` weights; the original
  dimension condition is insufficient -- q6_K lm_head for gemma-4 E2B
  has [1536, 262144], which is big enough to slowdown gemv_noshuffle but
  does not satisfy the dimension condition (ne0 >= 2048)
2026-08-04 13:37:47 +03:00
..
kernels sync : ggml (#3962) 2026-07-31 09:11:28 +02:00
CMakeLists.txt sync : ggml (#3962) 2026-07-31 09:11:28 +02:00
cl-program-cache.cpp sync : ggml (#3962) 2026-07-31 09:11:28 +02:00
cl-program-cache.h sync : ggml (#3962) 2026-07-31 09:11:28 +02:00
fa_tune.h opencl: general flash attention decode performance optimizations (llama/25366) 2026-07-10 13:06:42 +03:00
ggml-opencl.cpp opencl: route large q6_K lm_head to the flat GEMV (llama/26427) 2026-08-04 13:37:47 +03:00
libdl.h opencl: allow loading precompiled binary kernels from library (llama/23042) 2026-07-10 13:06:42 +03:00