Linux fréttir

CISA adds high-severity flaw to KEV list, urges swift updating

Uncle Sam's cyber wardens have warned that a high-severity flaw in Microsoft's Windows SMB client is now being actively exploited – months after it was patched.…

Alibaba Cloud claims its new Aegaeon GPU pooling system cuts Nvidia GPU use by 82%, letting 213 H20 accelerators handle workloads that previously required 1,192. The advancements have been detailed in a paper (PDF) at the 2025 ACM Symposium on Operating Systems (SOSP) in Seoul. Tom's Hardware reports: Unlike training-time breakthroughs that chase model quality or speed, Aegaeon is an inference-time scheduler designed to maximize GPU utilization across many models with bursty or unpredictable demand. Instead of pinning one accelerator to one model, Aegaeon virtualizes GPU access at the token level, allowing it to schedule tiny slices of work across a shared pool. This means one H20 could serve several different models simultaneously, with system-wide "goodput" -- a measure of effective output -- rising by as much as nine times compared to older serverless systems. The system was tested in production over several months, according to the paper, which lists authors from both Peking University and Alibaba's infrastructure division, including CTO Jingren Zhou. During that window, the number of GPUs needed to support dozens of different LLMs -- ranging in size up to 72 billion parameters -- fell from 1,192 to just 213. While the paper does not break down which models contributed most to the savings, reporting by the South China Morning Post says the tests were conducted using Nvidia's H20, one of the few accelerators still legally available to Chinese buyers under current U.S. export controls.

Read more of this story at Slashdot.

DevOps guru and ex-Googler say vibes beat reading diffs but there are risks

"Accept All. Always. Don't read the diffs anymore."…

Security pros explore whether infection-spoofing code can immunize Windows systems against attack

Feature What's better, prevention or cure? For a long time the global cybersecurity industry has operated by reacting to attacks and computer viruses. But given that ransomware has continued to escalate, more proactive action is needed.…

Another phone Linux? The Reg attempts to disentangle the options

The latest version of Mobian, an edition of Debian aimed at mobile devices, is here, based on Debian 13 "Trixie".…

Calendar cock-up exposed recipients' details

Anti-fraud nonprofit Cifas was left red-faced after sending out a calendar invite that exposed the email addresses of dozens of individuals working across the fraud space.…

SpaceX surpassed the 10,000-satellite milestone for its Starlink constellation after two Falcon 9 launches on Oct. 19 added 56 more satellites to orbit. The company now operates about two-thirds of all active satellites worldwide and continues to break reuse records. Space.com reports: A Falcon 9 rocket carrying 28 Starlink internet satellites lifted off from California's Vandenberg Space Force Base today at 3:24 p.m. EDT (1924 GMT; 12:24 p.m. local California time). Those 28 included the 10,000th Starlink spacecraft ever to reach orbit, which a SpaceX employee noted on the company's launch webcast: "From Tintin to 10,000! Go Starlink, go Falcon, go SpaceX!" It was also the 132nd Falcon 9 liftoff of the year, equaling the mark set by the rocket last year -- and there are still nearly 2.5 months to go in 2025. [...] This launch was the second of the day for SpaceX; less than two hours earlier, another Falcon 9 sent 28 more Starlink satellites up from Florida's Space Coast. That earlier liftoff was the 31st for that Falcon 9's first stage, setting a new reuse record.

Read more of this story at Slashdot.

12 more hours of pain followed initial outage

Amazon Web Services has revealed that its efforts to recover from the massive mess at its US-EAST-1 region caused other services to fail.…

Better scheduling and resource-sharing for inferencing workloads using multiple models, not a training breakthrough

Chinese tech giant Alibaba has published a paper detailing scheduling tech it has used to achieve impressive utilization improvements across the GPU fleet it uses to power inferencing workloads – which is nice, but not a breakthrough that will worry AI investors.…

An anonymous reader quotes a report from Wired: The National Transportation Safety Board confirmed Sunday that it is investigating an airliner that was struck by an object in its windscreen, mid-flight, over Utah. "NTSB gathering radar, weather, flight recorder data," the federal agency said on the social media site X. "Windscreen being sent to NTSB laboratories for examination." The strike occurred Thursday, during a United Airlines flight from Denver to Los Angeles. Images shared on social media showed that one of the two large windows at the front of a 737 MAX aircraft was significantly cracked. Related images also reveal a pilot's arm that has been cut multiple times by what appear to be small shards of glass. The captain of the flight reportedly described the object that hit the plane as "space debris." This has not been confirmed, however. After the impact, the aircraft safely landed at Salt Lake City International Airport after being diverted. Images of the strike showed that an object made a forceful impact near the upper-right part of the window, showing damage to the metal frame. Because aircraft windows are multiple layers thick, with laminate in between, the window pane did not shatter completely. The aircraft was flying above 30,000 feet -- likely around 36,000 feet -- and the cockpit apparently maintained its cabin pressure.

Read more of this story at Slashdot.