]> git.djapps.eu Git - pkg/ggml/sources/whisper.cpp/commit
vulkan: support GET_ROWS for k-quants (llama/16235)
authorJeff Bolz <redacted>
Sat, 27 Sep 2025 10:36:11 +0000 (06:36 -0400)
committerGeorgi Gerganov <redacted>
Mon, 29 Sep 2025 12:18:11 +0000 (15:18 +0300)
commit97bd65f90f9521e1944f97864184d50e5bb146f5
tree98722fe23e3986906bee609c8b27ecd7626a709b
parent23b359895266a60c99c80e44c475cae023d231bf
vulkan: support GET_ROWS for k-quants (llama/16235)

The dequantize functions are copy/pasted from mul_mm_funcs.comp with very few
changes - add a_offset and divide iqs by 2. It's probably possible to call
these functions from mul_mm_funcs and avoid the duplication, but I didn't go
that far in this change.
ggml/src/ggml-vulkan/ggml-vulkan.cpp
ggml/src/ggml-vulkan/vulkan-shaders/dequant_funcs.comp
ggml/src/ggml-vulkan/vulkan-shaders/types.comp
ggml/src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp