llama.cpp Open PRs: Speeding up CPU and Hybrid Inference
pmttyji · reddit · 2026-08-30
A list of 30+ open PRs in llama.cpp focused on CPU, RAM, disk, and hybrid inference optimizations. Key improvements include MoE expert caching/disk streaming, AVX/VNNI acceleration, quantization kernels (RVV/NEON/AMX), NUMA mirroring, KV cloning, and RAM peak reduction, aiming for significant speedups.
More from Infra
- Next-gen GraalVM benchmarks: 4x faster startup, 2x less memory, no build step needed — RealGeneKim · 2026-10-15
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark AI laptops — tomwarren · 2026-10-15
- How Microsoft optimized Windows for Nvidia's Arm-based RTX Spark laptops — The Verge AI · 2026-10-15
- Global smartphone shipments plunge 22% in worst quarterly decline on record as AI demand fuels memory costs — Polymarket · 2026-10-15
- SK Hynix drops 11% again, now below 3x FY2027 earnings as AI memory selloff deepens — JOBhakdi · 2026-10-15
- ASML Sells Zero Lithography Machines into Europe as US Hosts 75%+ of AI Compute — NinaDSchick · 2026-10-15