Skip to content

llama : add --hugepages flag for anonymous HugeTLB-backed weight load…

6aef930
Select commit
Loading
Failed to load commit list.
Open

UPSTREAM PR #21821: llama : add --hugepages for HugeTLB-backed weight loading (Linux) #1347

llama : add --hugepages flag for anonymous HugeTLB-backed weight load…
6aef930
Select commit
Loading
Failed to load commit list.
LOCI Review / Performance Per Binary #1347 succeeded Apr 13, 2026

Performance varied across binaries, overall acceptable

3 binaries improved · 10 binaries unchanged · 2 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 0.48 -0.04 stable
build.bin.libmtmd.so 0 0 unchanged
build.bin.llama-bench -0.06 -0.04 improved
build.bin.llama-cvector-generator 0.36 0.06 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 0 0.02 unchanged
build.bin.llama-qwen2vl-cli 0 0 unchanged
build.bin.llama-tokenize -0.08 0.01 improved
build.bin.llama-tts -0.12 0.09 improved

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.