]> git.djapps.eu Git - pkg/ggml/sources/ggml/commit
ggml-cpu: extend support for RVV floating-point kernels (llama/17318)
authorTaimur Ahmad <redacted>
Thu, 18 Dec 2025 14:02:09 +0000 (19:02 +0500)
committerGeorgi Gerganov <redacted>
Wed, 31 Dec 2025 10:39:43 +0000 (12:39 +0200)
commit117a67747d3ed31119cec6f4d614a2ea92390fa9
treeb80a55b0a75ac7d97530274968deb54e8ed1a969
parent0c4d8f6580cb041aa73af04396ac7097e7f5aeda
ggml-cpu: extend support for RVV floating-point kernels (llama/17318)

* cmake: add BF16 RVV flag for ggml-cpu

* ggml-cpu: add floating-point conversion kernels

* ggml: add floating-point kernels

Co-authored-by: Rehan Qasim <redacted>
* ggml-cpu: fix lmul in vec_dot_bf16

* ggml-cpu: change redsum to lmul 4, fix leftover

---------

Co-authored-by: Rehan Qasim <redacted>
src/ggml-cpu/CMakeLists.txt
src/ggml-cpu/ggml-cpu.c
src/ggml-cpu/vec.cpp
src/ggml-cpu/vec.h