]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: ad39cca)
sycl : Make GGML_SYCL_F16=ON the default (#23996)
authorTodd Malsbary <redacted>
Tue, 16 Jun 2026 05:34:02 +0000 (22:34 -0700)
committerGitHub <redacted>
Tue, 16 Jun 2026 05:34:02 +0000 (08:34 +0300)
commit4196b477da74a9a4d24b4ea40ec99b63df788c45
tree2c049b26afcef36bfcd4f31d1f058da87416d65btree
parentad39ccaa1951debc5174c3ce22c0ffe328442f2ccommit | diff
sycl : Make GGML_SYCL_F16=ON the default (#23996)

* Add -cl-fp32-correctly-rounded-divide-sqrt to F16=ON builds

Signed-off-by: Todd Malsbary <redacted>
* Make GGML_SYCL_F16=ON the default

Signed-off-by: Todd Malsbary <redacted>
* Leave F32 the default

F16 remains explictly set for example and Dockerfile builds.

Signed-off-by: Todd Malsbary <redacted>
* Revert changes to examples/sycl/build scripts

Signed-off-by: Todd Malsbary <redacted>
---------

Signed-off-by: Todd Malsbary <redacted>
.devops/intel.Dockerfile diff | blob | history
docs/backend/SYCL.md diff | blob | history
Packaging of ggml-org/llama.cpp