]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
ggml-cpu: Support Q5_0 and Q5_1 on s390x (#15486)
authorAaron Teo <redacted>
Fri, 22 Aug 2025 08:11:04 +0000 (16:11 +0800)
committerGitHub <redacted>
Fri, 22 Aug 2025 08:11:04 +0000 (16:11 +0800)
commitad5c975c2d0297124fad210776ef8eed6b90d578
tree380a46b440f38d082ce968ce04863d31e9ca96ac
parent4afb0a746f22abaa545b3ebdb76a400d7da3a713
ggml-cpu: Support Q5_0 and Q5_1 on s390x (#15486)

* ggml-cpu: initial q5_0 impl for s390x

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: updated q5_0 code for better performance

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: use optimised hsum for better performance

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: introduce q5_1 simd + refactor q5_0

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: fix incorrect return type vec_hsum

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: q5_0 incomplete refactor + table_b2b_0 activation

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: refactor q5_1

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: q5_1 update loop unroll to 4

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: update q5_0 unroll to 4

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: update build-s390x docs

Signed-off-by: Aaron Teo <redacted>
* ggml-cpu: update unused variables q5_0

Signed-off-by: Aaron Teo <redacted>
* docs: update the last update date

Signed-off-by: Aaron Teo <redacted>
---------

Signed-off-by: Aaron Teo <redacted>
docs/build-s390x.md
ggml/src/ggml-cpu/arch-fallback.h
ggml/src/ggml-cpu/arch/s390/quants.c
ggml/src/ggml-cpu/ggml-cpu-impl.h