llama.cpp
vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl
#21029

Merged

vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl #21029

0cc4m merged 1 commit into ggml-org:master from mkoker:vulkan-fa-q4_1-q5_0-q5_1

mkoker requested a review 20 days ago

github-actions added Vulkan

github-actions added ggml

mkoker force pushed from ce0593f8 to 7fbd00e9 20 days ago

mkoker changed the title ~~vulkan: add flash attention dequant for q4_1, q5_0, q5_1 KV cache types~~ vulkan: add FA dequant for q4_1, q5_0, q5_1 20 days ago

mkoker force pushed from 7fbd00e9 to a712a8a7 20 days ago

mkoker changed the title ~~vulkan: add FA dequant for q4_1, q5_0, q5_1~~ vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl 20 days ago

0cc4m requested changes on 2026-03-30

mkoker force pushed from a712a8a7 to 355a65f9 16 days ago

0cc4m commented on 2026-04-02

mkoker force pushed from d6ba11b1 to b9e5364f 13 days ago

vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl

370e90da

mkoker force pushed from b9e5364f to 370e90da 13 days ago

0cc4m approved these changes on 2026-04-04

JohannesGaessler approved these changes on 2026-04-07

0cc4m merged edd4d9bc into master 8 days ago

Reviewers

JohannesGaessler

0cc4m

Assignees

No one assigned

Labels

Vulkan ggml

Milestone

No milestone

llama.cpp vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl #21029 Merged

vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl #21029

llama.cpp
vulkan: add FA dequant for q4_1, q5_0, q5_1, iq4_nl
#21029

Merged