UPSTREAM PR #20424: common/parser: add proper reasoning tag prefill reading #1247
LOCI Review / Performance Per Binary #1247
succeeded
Mar 19, 2026 in 0s
Performance varied across binaries, overall acceptable
1 binary improved · 10 binaries unchanged · 4 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 | 0 | unchanged |
build.bin.libggml-cpu.so |
0 | 0 | unchanged |
build.bin.libggml.so |
0 | 0 | unchanged |
build.bin.libllama.so |
0 | 0 | unchanged |
build.bin.libmtmd.so |
0 | 0 | unchanged |
build.bin.llama-bench |
-0.01 | 0.11 | improved |
build.bin.llama-cvector-generator |
8.26 | 0.12 | stable |
build.bin.llama-gemma3-cli |
0 | 0 | unchanged |
build.bin.llama-gguf-split |
0 | 0 | unchanged |
build.bin.llama-llava-cli |
0 | 0 | unchanged |
build.bin.llama-minicpmv-cli |
0 | 0 | unchanged |
build.bin.llama-quantize |
1.71 | 0.88 | stable |
build.bin.llama-qwen2vl-cli |
0 | 0 | unchanged |
build.bin.llama-tokenize |
2.18 | 0.7 | stable |
build.bin.llama-tts |
8.48 | 0.1 | 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