DGX Spark driver fixes reclaim 64GB of 'missing' GPU memory and boost page faults 49x
QuixiAI · x · 2026-09-06
QuixiAI (Eric Hartford) shares two fixes for the NVIDIA open kernel driver, both verified on a DGX Spark (GB10):
- Memory reclaim: freed GPU memory now returns to the OS when a process exits — previously 64 GB went 'missing' and vLLM refused to start.
- Huge pages: GPU page faults on system memory now use huge pages, raising first-touch bandwidth from 0.4 to 19.6 GiB/s (49x).
More from Infra
- SurfSense: an open-source local NotebookLM with MCP agents and a prompt-caching lesson — FurtiveMirth · 2026-10-23
- Meta researcher: horrifying infra, not talent, is why OSS can't catch closed frontier — A_K_Nain · 2026-10-23
- Jane Street and NVIDIA speed up Mamba-2 with custom cuTile kernels — blelbach · 2026-10-23
- LatentLoop: asymmetric compute allocation cuts looped-MoE training cost by 19% — pmttyji · 2026-10-23
- Same AI request emits 8.5x more CO2 in Texas than in Sweden — jlundberg · 2026-10-23
- SK Hynix Pulls Plug on Solidigm US IPO as Chip Stocks Slide, Shares Down ~30% This Quarter — rohanpaul_ai · 2026-10-23