UPSTREAM PR #16490: graph : reuse SSM graphs#255
Open
DajanaV wants to merge 6 commits into
Open
LOCI Review / Performance Per Binary #255
succeeded
Jan 31, 2026 in 0s
Performance varied across binaries, overall acceptable
3 binaries improved · 7 binaries unchanged · 6 binaries stable ~ within threshold · 0 binaries degraded ~ beyond threshold
| Binary | Δ % Response | Δ % Throughput | Performance (based on response time) |
|---|---|---|---|
build.bin.libggml-base.so |
0.09 | 0.13 | stable |
build.bin.libggml-cpu.so |
0.63 | 0.95 | stable |
build.bin.libggml.so |
0.04 | 0.09 | stable |
build.bin.libllama.so |
-0.04 | 0.22 | improved |
build.bin.libmtmd.so |
0.5 | 0.38 | stable |
build.bin.llama-bench |
0 | 0 | unchanged |
build.bin.llama-cvector-generator |
-0.1 | 0 | improved |
build.bin.llama-gemma3-cli |
0 | 0 | unchanged |
build.bin.llama-gguf-split |
0 | 0.04 | unchanged |
build.bin.llama-llava-cli |
0 | 0 | unchanged |
build.bin.llama-minicpmv-cli |
0 | 0 | unchanged |
build.bin.llama-quantize |
0.01 | 0.01 | stable |
build.bin.llama-qwen2vl-cli |
0 | 0 | unchanged |
build.bin.llama-run |
-0.12 | 0 | improved |
build.bin.llama-tokenize |
0 | 0.02 | unchanged |
build.bin.llama-tts |
0.06 | -0.01 | stable |
Performance threshold: 30%
Default configuration used.
Note: Performance status is evaluated only from Δ% Response. Throughput is displayed for reference.
Explore the complete analysis inside the Loci Inspector.
Open the Pull Request linked to this check-run.
Loading