]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
[SYCL] supprt Flash Attention for fp32/fp16/Q4/Q5/Q8 (#20190)
authorNeo Zhang <redacted>
Sun, 8 Mar 2026 04:00:07 +0000 (12:00 +0800)
committerGitHub <redacted>
Sun, 8 Mar 2026 04:00:07 +0000 (12:00 +0800)
commit213c4a0b81788e058c30479842954fb0815be61a
tree6bd9c2e785ccc48c9aeaf4c5b755543606ddcfbb
parentc5a778891ba0ddbd4cbb507c823f970595b1adc2
[SYCL] supprt Flash Attention for fp32/fp16/Q4/Q5/Q8 (#20190)

* support flash-attention for fp32/fp16/Q4/Q5/Q8

* rm warining

* update for JIT
65 files changed:
docs/backend/SYCL.md
docs/ops.md
docs/ops/SYCL.csv
ggml/src/ggml-sycl/CMakeLists.txt
ggml/src/ggml-sycl/backend.hpp
ggml/src/ggml-sycl/common.hpp
ggml/src/ggml-sycl/convert.cpp
ggml/src/ggml-sycl/convert.hpp
ggml/src/ggml-sycl/count-equal.cpp
ggml/src/ggml-sycl/dpct/helper.hpp
ggml/src/ggml-sycl/fattn-common.hpp [new file with mode: 0644]
ggml/src/ggml-sycl/fattn-tile.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/fattn-tile.hpp [new file with mode: 0644]
ggml/src/ggml-sycl/fattn-vec.hpp [new file with mode: 0644]
ggml/src/ggml-sycl/fattn.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/fattn.hpp [new file with mode: 0644]
ggml/src/ggml-sycl/ggml-sycl.cpp
ggml/src/ggml-sycl/presets.hpp
ggml/src/ggml-sycl/softmax.cpp
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq112-dv112.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq128-dv128.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq256-dv256.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq40-dv40.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq576-dv512.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq64-dv64.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq72-dv72.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq80-dv80.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-tile-instance-dkq96-dv96.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-f16-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_0-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q4_1-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_0-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q5_1-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-f16.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-q4_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-q4_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-q5_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-q5_1.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/template-instances/fattn-vec-instance-q8_0-q8_0.cpp [new file with mode: 0644]
ggml/src/ggml-sycl/vecdotq.hpp