llama.cpp
fix: use `vm_allocate` to allocate CPU backend buffer on macOS
#9875
Merged
Merged
fix: use `vm_allocate` to allocate CPU backend buffer on macOS #9875
slaren
merged
15 commits into
ggml-org:master
from
giladgd:macosMemalignBackendCpuBuffer
fix: use `vm_allocate` to allocate CPU backend buffer on macOS
06b119a1
fix: switch to `posix_memalign` to keep existing `free()` usages work
cec2d4e2
slaren
commented
on 2024-10-14
feat: move `GGML_ALIGNED_MALLOC` to `ggml-backend-impl.h`, add suppor…
51ccdebc
style: formatting
c2259e3c
fix: move const outside of `#ifndef`
55298d25
style: formatting
fa79e0d0
fix: unused var
19a820fc
fix: transform `GGML_ALIGNED_MALLOC` and `GGML_ALIGNED_FREE` into fun…
8352354a
fix: unused var
df67ae38
giladgd
changed the title
slaren
commented
on 2024-10-14
fix: page align to `GGUF_DEFAULT_ALIGNMENT`
33b43081
fix: page align to `TENSOR_ALIGNMENT`
cb224645
fix: convert `TENSOR_ALIGNMENT` to a macro
1516f7b7
fix: increase page size to `32` on iOS
4455c7f0
fix: iOS page size
de7836a4
slaren
commented
on 2024-10-15
fix: `hbw_posix_memalign` alignment
ecddb245
slaren
approved these changes
on 2024-10-16
Login to write a write a comment.
Login via GitHub