Files
llama.cpp/ggml
Nathan Wilson c7bda030e7 vulkan: fix FA dequant path engagement (#28190)
Skip the nb[3] check when ne[3] == 1, the shader never reads it for a
single stream. Cache views carry the full-buffer stride there, so the old
check reduced to n_kv == kv_size and the path only engaged with the
cache full.
2026-09-03 10:40:34 +02:00
..
2024-07-13 18:12:39 +02:00