Urgent.News

One page, thousands of outlets. See who else covered it.

Editions

Tech

AI-enriched Linux 7.2 delivers cache-aware scheduling - here's everything new

The latest stable kernel also brings filesystem and I/O improvements, and substantial new support across AMD, Intel, Apple, Nvidia, USB4, and laptop hardware.

Linus Torvalds released Linux kernel 7.2, highlighting several improvements. Torvalds stated that the size of this release was larger than anticipated, but it was necessary due to numerous fixes, many discovered through AI. The most significant new feature is Cache Aware Scheduling, which aggregates tasks sharing data to the same LLC cache domain, reducing cache bouncing and improving data access efficiency.

This feature is particularly beneficial for workloads like multithreaded servers, AI and ML compute jobs, and some games. Other enhancements include new and improved drivers, such as initial AMDGPU support for HDMI 2.1 Fixed Rate Link (FRL), which allows uncompressed 4K at 120Hz and 8K video playback. The release also introduces Btrfs large folios as default, resulting in 15% improvements in sequential writes.

ExFAT performance on devices like SD cards and USB sticks has been improved by up to 87% with the new kernel's iomap feature. Additionally, Wi-Fi Aware and early Wi-Fi 8/UHR groundwork has been added, enabling direct device-to-device Wi-Fi connections without traditional network infrastructure. However, a significant rollback was made due to a performance regression with the new GPU scheduler on certain AMD graphics configurations, reverting to FIFO scheduling instead.

Written by urgent.news from ZDNet's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at zdnet.com →

More in Tech

Feedly attributes weeklong slowdown to bug, not its AI pivot

Feedly says a bug is behind the performance issues that have made its web app nearly "unusable" for some users, while complaints about its mobile apps and customer support are adding to frustrations.

  • Feedly experienced a weeklong slowdown due to a bug, not AI pivot
  • Web app issues caused frustration, rendering browsing feeds unusable
  • CEO assured RSS priority, fix released Friday, investigating all user issues

20 Native macOS Apps That Replaced My Electron Bloat for Good

I opened Activity Monitor last year on a Thursday afternoon, just to check something quick. Slack was sitting at 780MB. Discord was at 410MB. Notion was at 490MB. Spotify was lurking at 280MB.

  • Electron framework causes high RAM usage in popular macOS apps
  • Mimestream, Zed, Craft, and Reeder are faster alternatives
  • Native apps are more integrated with macOS than Electron apps

From HTTP Requests to Atomic On-Chain Transactions

Coming from Web2, I initially thought of a Solana transaction like an API request: send request -> server processes it -> get response But after working with Solana transactions, I realized the mental…

  • Solana transaction differs from API request
  • Atomicity ensures transaction success or failure
  • Validity window tied to recent blockhash

More from Monday 17 August →