llama.cpp
fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS
#7078

Merged

fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS #7078

ggerganov merged 5 commits into ggml-org:master from giladgd:metalPosixMemalignWorkaround

fix: use `malloc` instead of `posix_memalign` in `ggml-metal.m` to ma…

571dca57

fix: typo

a53e5179

fix: use `vm_allocate` instead of `posix_memalign`

bfa4daea

fix: don't call `newBufferWithBytesNoCopy` with `NULL` when `ggml_met…

a92efecb

fix: use `vm_allocate` only on macOS

78214ac5

giladgd changed the title ~~fix: workaround to not crash when running in Electron~~ fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS 1 year ago

slaren approved these changes on 2024-05-07

ggerganov merged 26458af1 into master 1 year ago

giladgd deleted the metalPosixMemalignWorkaround branch 1 year ago

Reviewers

slaren

Assignees

No one assigned

Labels

None yet

Milestone

No milestone

llama.cpp fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS #7078 Merged

fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS #7078

llama.cpp
fix: use `vm_allocate` instead of `posix_memalign` for Metal on macOS
#7078

Merged