]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
server : refactor prompt cache state ownership (#25649)
authorGeorgi Gerganov <redacted>
Tue, 14 Jul 2026 15:25:52 +0000 (18:25 +0300)
committerGitHub <redacted>
Tue, 14 Jul 2026 15:25:52 +0000 (18:25 +0300)
commitbf2c86ddc0685f580595954056c2e77ebabfab4f
tree2b9ac9c85f40b758b747dd1d027694471e815d59
parent6e52db5b72b643cad908458b918ddb4f4b7974d9
server : refactor prompt cache state ownership (#25649)

* server : clear checkpoints upon prompt clear

* server : move the prompt state data to the server_prompt_cache

Assisted-by: pi:llama.cpp/Qwen3.6-27B
* server : handle batched slot being cleared
tools/server/server-context.cpp
tools/server/server-task.cpp
tools/server/server-task.h