ggerganov/llama.cpp

Pull Requests Commits

allow dmabuf p2p

0cc4m committed 1 day ago

af2146b7

fix compile issue

0cc4m committed 1 day ago

b4acc6d3

Merge branch 'pr-25051' into 0cc4m/vulkan-allreduce

0cc4m committed 1 day ago

6634f71f

reverse check order

0cc4m committed 1 day ago

e46b3ae5

fix compile issues

0cc4m committed 1 day ago

50e0a2dd

llama.cpp : bump version to 0.3.0 (#27696)

ggerganov committed 1 day ago

Verified c1d0e7a0

use memcpy for small copies across host-visible memory

0cc4m committed 1 day ago

4d879b9c

add semi-async staging copy and use it over syncfd for small copies

0cc4m committed 1 day ago

bce45fbe

swap dmabuf check order

0cc4m committed 1 day ago

bac13690

use allocator instead of hop2

0cc4m committed 1 day ago

86ad4417

use sync_fd binary semaphores for cross-driver synchronization

0cc4m committed 1 day ago

5f512ae7

double buffering

0cc4m committed 1 day ago

4f4cea31

fixes, disable semaphore sharing on Nvidia + non-Nvidia

0cc4m committed 1 day ago

3e53e38d

0cc4m committed 1 day ago

bf316ac5

vulkan: add optimized device to device copy function

0cc4m committed 1 day ago

740c1273

ggerganov committed 1 day ago

81191aff

ggml : bump version to 0.22.0 (ggml/1607)

ggerganov committed 1 day ago

93882361

grammar : parse \- in char classes as literal hyphen (#27591)

NIXKnight committed 1 day ago

Verified eb25b726

sycl : mark tq2_0 as not supported (#27660)

arthw committed 1 day ago

Verified 814d84bc

webgpu : fix handling of infinity values during ARGSORT and TOP_K (#27538)

fairydreaming committed 1 day ago

Verified 5ea87dda

metal : per-device tuned (Q, NE) for flash-attn vec (#26570)

forforever73 committed 1 day ago

Verified f280b269

metal: per-op source split + parallel compile (#26561)

forforever73 committed 2 days ago

Verified b615f5b4

misc : read repetition_penalty from generation_config.json (#27659)

tdakhran committed 2 days ago

Verified b3c3b96a

tests : disable DOTS3NOTE arch test for WebGPU (#27654)

fairydreaming committed 2 days ago

Verified 75844307

convert: fix GLM regression in index_tensors (#27655)

jacekpoplawski committed 2 days ago

Verified 71cc86fa

ggml : shorten virtual device naming in CUDA and Metal (#27608)

ggerganov committed 2 days ago

Verified a14dba68

webgpu : reorder includes since V that appears in common_decls.tmpl may be defined as K in flash_attn_decls.tmpl if KV_OVERLAP (#27545)

fairydreaming committed 2 days ago

Verified c1c766da

mtmd: video: fix moov atom at the end of file (#27596)

ngxson committed 2 days ago

Verified 160c6b0b

ci : apply ccache-clear with older/min/dry-run to all ccache jobs (#27602)

ggerganov committed 2 days ago

Verified 985b1491

ggml : fix ggml_clamp (#27644)

ggerganov committed 2 days ago

Verified 6036c635

Older