]> git.djapps.eu Git - pkg/ggml/sources/whisper.cpp/commit
vulkan: matmul dequantization improvements (llama/12015)
authorEve <redacted>
Fri, 28 Feb 2025 07:20:08 +0000 (07:20 +0000)
committerGeorgi Gerganov <redacted>
Sat, 8 Mar 2025 13:13:01 +0000 (15:13 +0200)
commit1fbb119b1e31ae96716ad7f6f215745af31f3c45
tree6f238adb680dfc3f1de8cc57c1cbffcfe2aee960
parent40dea850fd3287b41d4afd1188c49dbc1f8807a4
vulkan: matmul dequantization improvements (llama/12015)

* faster dequant for old quants

* dont use unpack for iq4_nl

* vec2 unpack for q8
ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs.comp
ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs_cm2.comp
ggml/src/ggml-vulkan/vulkan-shaders/mul_mm.comp
ggml/src/ggml-vulkan/vulkan-shaders/types.comp
ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp