llama.cpp
metal : allow ops to run concurrently
#15929
Merged

metal : allow ops to run concurrently #15929

ggerganov merged 12 commits into master from gg/metal-concurrent-graphs
ggerganov
github-actions github-actions added ggml
github-actions github-actions added Apple Metal
calvin2021y
ggerganov ggerganov force pushed to 417df406 116 days ago
ggerganov
calvin2021y
ggerganov metal : run graphs ops concurrently
2fb15521
ggerganov cont : add flags for debugging and disabling concurrency
a3519fd5
ggerganov cont : refactor and handle fusing
74d2961e
ggerganov cont : simplify - no need to use GPU address
acd1404a
ggerganov cont : prepare mem ranges for reuse + add ggml-metal-common.cpp
f7aeab93
ggerganov cont : avoid redundant keywords in cpp [no ci]
1c9d3f3b
ggerganov metal : reorder graph for better concurrency
89cca2ad
ggerganov metal : fix race on mem pool buffers
a3f17d61
ggerganov cont : add env GGML_METAL_GRAPH_OPTIMIZE_DISABLE
0b586365
ggerganov cont : refactor, optimize, add comments
907616de
ggerganov cont : refactor ggml-metal.m
faffbec3
ggerganov ggerganov force pushed from 17cf93d2 to faffbec3 115 days ago
ggerganov minor : update logs [no ci]
e502db1f
ggerganov ggerganov merged f161463a into master 115 days ago
ggerganov ggerganov deleted the gg/metal-concurrent-graphs branch 115 days ago

Login to write a write a comment.

Login via GitHub

Reviewers
No reviews
Assignees
No one assigned
Labels
Milestone