]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: d5cb868)
metal : single allocation of encode_async block (#9747)
authorPaul Tsochantaris <redacted>
Mon, 7 Oct 2024 12:26:31 +0000 (13:26 +0100)
committerGitHub <redacted>
Mon, 7 Oct 2024 12:26:31 +0000 (15:26 +0300)
commit96b69121033d2b6b951d1b6b1b43f8b4f97dac99
treee1da8edb3002894f339ffde7ed769288044c4352tree
parentd5cb86844f26f600c48bf3643738ea68138f961dcommit | diff
metal : single allocation of encode_async block (#9747)

* Single allocation of encode_async block with non-ARC capture in ggml-metal.m

* Moving Block_release to the deallocation code

* Release encode block when re-setting encoding buffer count if needed

* Update ggml/src/ggml-metal.m

---------

Co-authored-by: Georgi Gerganov <redacted>
ggml/src/ggml-metal.m diff | blob | history
Packaging of ggml-org/llama.cpp