]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
vulkan: support copy from f32 to q4_0/q4_1/q5_0/q5_1/q8_0/iq4_nl (llama/11166)
authorJeff Bolz <redacted>
Thu, 16 Jan 2025 21:47:10 +0000 (15:47 -0600)
committerGeorgi Gerganov <redacted>
Wed, 29 Jan 2025 10:57:00 +0000 (12:57 +0200)
commit9b8df2211de1441cfb0d15a8f237345dad33cd72
treefdfde7087cebee959aeca2c7f97ce34f9bb90fb4
parent5b62c4013e7db24de63acc6159fb10ff7ba147e0
vulkan: support copy from f32 to q4_0/q4_1/q5_0/q5_1/q8_0/iq4_nl (llama/11166)

* vulkan: support copy from f32 to q4_0/q4_1/q5_0/q5_1/q8_0/iq4_nl

Shaders are based on cpy.cu.

* vulkan: support copy from q4_0/q4_1/q5_0/q5_1/q8_0/iq4_nl to f32

* ggml: copy q->f32 assumes some contiguity in the destination
src/ggml-cpu/ggml-cpu.c
src/ggml-vulkan/ggml-vulkan.cpp
src/ggml-vulkan/vulkan-shaders/copy_from_quant.comp [new file with mode: 0644]
src/ggml-vulkan/vulkan-shaders/copy_to_quant.comp [new file with mode: 0644]
src/ggml-vulkan/vulkan-shaders/generic_unary_head.comp
src/ggml-vulkan/vulkan-shaders/vulkan-shaders-gen.cpp
tests/test-backend-ops.cpp