3 papers
cs.LG2026
Ask, Don't Equip: Question-Driven Coding Agents for Deep Kernel Optimization
Gabriele Oliaro, Jaeseong Lee, Zhaoyuan Su +3
LLM agents for GPU kernel optimization sit at two ends of a structure-freedom tradeoff. Scripted agents fix what the model does at each step, via a learned policy, an optimization…
cs.LG2026
FastKernels: Benchmarking GPU Kernel Generation in Production
Gabriele Oliaro, Yichao Fu, Jaeseong Lee +8
LLM-based agents for GPU kernel generation are advancing rapidly, but the benchmarks they optimize against evaluate kernels in isolation, with synthetic inputs and weak baselines,…
cs.DC2025
Nexus:Proactive Intra-GPU Disaggregation of Prefill and Decode in LLM Serving
Xiaoxiang Shi, Colin Cai, Junjia Du +1
Monolithic serving with chunked prefill improves GPU utilization by batching prefill and decode together, but suffers from fine-grained phase interference. Engine-level prefill-dec…