]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
batched-bench : fix unified KV cache handling + pp timing (#15562)
authorGeorgi Gerganov <redacted>
Mon, 25 Aug 2025 10:56:43 +0000 (13:56 +0300)
committerGitHub <redacted>
Mon, 25 Aug 2025 10:56:43 +0000 (13:56 +0300)
commit6b64f74b55628e4193f4fb00313f07dbd8556528
tree394a2b9ed3406b4adc9511657c196d8466130899
parent0d5a470223fc90b6b6807921d68011ff06ae7f9e
batched-bench : fix unified KV cache handling + pp timing (#15562)

* batched-bench : fix unified KV cache handling + pp timing

* cont : run dummy token only with split KV cache
tools/batched-bench/batched-bench.cpp