Post
  1. Portrait of Onur Solmaz
    @onusoz · /2026/06/27 · View on
    This. Especially when the whole machine hangs instead of OOMing when my agent accidentally loads too many models into memory (lmk if there is a firmware update or sth that fixes this on the spark now, creating a cgroup doesn’t work) forums.developer.nvidia.com/t/dgx-spark-be…
    @ekzhang1 ·
    shower thought: AI inference engines like vLLM/SGLang might be the hardest “server”-shaped OSS to configure, ever made. Like it’s probably harder to configure them optimally than even Postgres, Nginx, Kubernetes, maybe even Kafka/Zookeeper … just in terms of all the issues, bugs, crashes, scheduler dynamics, kernels etc and it changes every month