PR #14482 llama : reuse compute graphs

llama : reuse compute graphs #14482

ggerganov merged 15 commits into master from gg/llama-reuse-graphs

rgerganov marked this pull request as ready for review 194 days ago

Base automatically changed from gg/kv-cache-use-set-rows to master 193 days ago

ggerganov force pushed from f61b0f77 193 days ago

esrakorkmz approved these changes on 2025-07-03

ggerganov marked this pull request as draft 192 days ago

ggerganov force pushed to 6372365b 191 days ago

ggerganov force pushed from 6372365b 191 days ago

ggerganov force pushed 191 days ago

ggerganov force pushed to 76681e3c 191 days ago

ggerganov requested a review from

slaren 191 days ago

compilade commented on 2025-07-05

ggerganov marked this pull request as ready for review 190 days ago

github-actions added examples

slaren commented on 2025-07-05

ggerganov force pushed to 600e69fd 188 days ago

ggerganov commented on 2025-07-08

slaren commented on 2025-07-09

ggerganov force pushed from a28677d1 to ce770410 186 days ago

ggerganov force pushed from ce770410 to 8303a684 185 days ago

ggerganov added hot

llama : reuse compute graphs

594b184e

llama-bench : add graph reuse parameter

ffc7634c

cont : remove the parameter and the sched resets

02cd9574

graph : rename update() to can_reuse()

3b8cef9f

params : remove is_same()

c82a5336

graph : set res->params in llm_graph_context constructor

3d28b3b1

ggerganov force pushed from 8303a684 to 3d28b3b1 184 days ago

slaren commented on 2025-07-16

graph : avoid set_max_nodes in llm_graph_result

b7b6caf3

kv-cache : reuse llama_context's graph result instance

84e0c1f3

slaren approved these changes on 2025-07-16

Merge branch 'master' into gg/llama-reuse-graphs

0ef1f6bb

context : reset the previous graph result upon memory updates

3d7ec2b9

batch : llama_ubatch now carries its data instead of pointing to balloc

acaf4b77

Merge branch 'master' into gg/llama-reuse-graphs

04155f0b

merge : fix build

a872790b

ggerganov commented on 2025-07-17

graph : fix can_reuse() checks when flash-attention is disabled

41366a4d

graph : move llm_graph_result impl in source file + debug env

c7ccf383

ggerganov force pushed to c7ccf383 179 days ago

slaren commented on 2025-07-17

ggerganov merged 01612b74 into master 179 days ago

ggerganov deleted the gg/llama-reuse-graphs branch 179 days ago

Reviewers

slaren

esrakorkmz

compilade

Assignees

No one assigned

Labels

examples hot

Milestone

No milestone

llama.cpp llama : reuse compute graphs #14482 Merged

llama : reuse compute graphs #14482

llama.cpp
llama : reuse compute graphs
#14482

Merged