#gpu-performance

The latest AI and developer news about #gpu-performance, with the original source and a short summary.

InfraRSI, an LLM distributed learning optimization agent, achieves a geometric mean speedup of 1.83x

InfraRSI achieved a geometric mean speedup of 1.83x on H100/H200 by simultaneously exploring configurations and code.

·2026.11.11 16:00

Decagon's AutoResearch Agent Automates Inference Stack Optimization, Achieving Up to 2.5x Throughput Gains

The system uses high-fidelity production replay and parallel sub-agents to diagnose bottlenecks and test speculative decoding, quantization, and scheduling changes.

·1·2026.10.29 03:00

doubleAI Introduces Maxwell: A Code Execution Model Using Formal Verification for Runtime Binary Optimization

Maxwell, a Code Execution Model by doubleAI available in early access, rewrites binaries at runtime with formal proof of semantic equivalence, achieving 1.86x speedups in LTX-2.5 video generation.

·3·2026.10.19 22:00

llama.cpp PR #29353 introduces GDN chunked kernel for CUDA and HIP, boosting Qwen inference speed by up to 20.6%

The new GDN chunked kernel implementation in llama.cpp delivers performance gains ranging from 0.9% to 20.6% across various Qwen models and hardware configurations.

·1·2026.10.14 23:00

mistral.rs Significantly Improves CUDA Inference Performance

mistral.rs v0.8.2 has been released, delivering up to 2.8x faster CUDA inference performance than llama.cpp on the latest GPUs such as H100 and B200.

·1·2026.06.01 23:00