GKRootWire
AI Google Adds 'Preferred Source' Button to Help Publishers Fight AI Traffic LossesGadgets Linkdaze Launches a Smart Calendar Aimed at Running Your Whole HouseholdSecurity Popular Rust Crate arrayref Hijacked to Spread Infostealer MalwareCloud & Sysadmin GitHub Details Cause of August 17 Outage, Outlines Reliability FixesDev Tools Show HN: 'Huzzah' Proposes a Fresh Take on AI-Assisted CodingCloud & Sysadmin The Weird Science of Cooling Data Centers With UrineAI Google Adds 'Preferred Source' Button to Help Publishers Fight AI Traffic LossesGadgets Linkdaze Launches a Smart Calendar Aimed at Running Your Whole HouseholdSecurity Popular Rust Crate arrayref Hijacked to Spread Infostealer MalwareCloud & Sysadmin GitHub Details Cause of August 17 Outage, Outlines Reliability FixesDev Tools Show HN: 'Huzzah' Proposes a Fresh Take on AI-Assisted CodingCloud & Sysadmin The Weird Science of Cooling Data Centers With Urine
Dev Tools

Linux Kernel Gets Smarter About Running Out of GPU Memory

A new kernel update improves how Linux handles GPU workloads that exceed available video RAM, reducing slowdowns and stutters.

Modern GPUs, especially in laptops and budget cards, often don't have enough dedicated vRAM for demanding workloads like gaming, video editing, or local AI inference. When memory runs out, the system has to swap data between vRAM and system RAM, a process called overcommit that has historically caused major performance hits on Linux.

A recent kernel improvement targets this exact bottleneck, refining how the graphics memory manager decides what to evict and when. The result, according to early testing referenced in the discussion, is smoother behavior when GPU memory pressure spikes, rather than the system grinding to a halt or stuttering badly.

This kind of low-level driver work rarely gets headlines, but it directly affects real-world usability for anyone running GPU-heavy tasks on Linux without a high-end card.

Why it matters: As more developers run local LLMs and AI tools on consumer GPUs, vRAM limits are becoming a everyday bottleneck rather than an edge case. Better overcommit handling means Linux users can push slightly beyond their hardware's limits without a jarring performance cliff, narrowing a longstanding gap with Windows driver behavior.

Sources: Hacker News