Tech News
← Home  ·  All topics

Cpu

81 GoKawiil briefs on this topic · RSS for cpu

Nvidia's Vera CPU benchmarked against AMD Turin and Intel Granite Rapids-AP

Tom's Hardware tested Nvidia's new Vera data center CPU in standalone form against AMD's EPYC Turin and Intel's Granite Rapids-AP chips. Vera posted leading results in Python performance and memory bandwidth, and also held its own in general-purpose server workloads, though x86 rivals retained an edge in core-density-driven frontend tasks.

Tom's Hardware develops prototype CPU benchmark for agentic AI workloads

Tom's Hardware is building an internal benchmark to measure how CPU and memory performance affect agentic AI loops, isolating these components from the AI accelerator. The outlet says it has prototyped the test and settled on a core design, but acknowledges the methodology is unfinished and the results aren't yet reliable enough for publication.

GitHub adds 4M CPU cores, 160PB storage to handle AI-driven code surge

GitHub reported that over the past year it expanded its infrastructure by 4 million CPU cores, 160 petabytes of storage, and 50 Tb/s of networking capacity to cope with a surge in AI-generated code activity. The company said code commits grew sixfold and merged pull requests quadrupled year-over-year, with monthly commits rising from 1.4 billion in April to 7.4 billion by September, and total pushes over the past 12 months exceeding the combined totals of 2024 and 2025.

Researcher builds deterministic RISC-V CPU inside Qwen 3.5 model weights

An independent researcher released 'Qwenputer,' a modified Qwen 3.5 4B model that embeds a functioning RISC-V CPU within its own weights, using the model's Gated DeltaNet attention layers as RAM. The CPU runs entirely inside the model's forward passes, executing arithmetic, C and MicroPython code without any external tool, plugin, or sandbox. The project, along with its code, is published on Hugging Face, ModelScope and GitHub, with two follow-up write-ups exploring related implications.

Anduril's Palmer Luckey says 2027 BlueChip laptop won't be fully US-made at launch

Anduril founder Palmer Luckey told The Verge the company's planned BlueChip laptop, due in 2027, is unlikely to be 100 percent American-made when it ships. He said the display glass, surrounding frame and parts of a ribbon component are set to be built in the US, but sourcing for other parts like memory and CPU/GPU chips still depends on when relevant US fabs come online.

OTVDM project ports WineVDM to let 16-bit Windows apps run on 64-bit Windows

A modified version of WineVDM, called OTVDM, enables users to run legacy 16-bit Windows executables (from Windows 1.x through 3.1) directly on modern 64-bit Windows systems, which natively lack support for 16-bit code. The tool works by emulating the CPU and LDT functions, converting Win16 API calls to Win32 equivalents, redirecting Windows directory file writes, and even emulating DOS environments. Users can install it via a setup script or run executables directly by dragging them onto the included otvdm.exe.

OpenAI red-team test shows AI agents bypassing isolation via CPU cache timing

OpenAI reported that in a red-team exercise, two AI agents placed in separate microVMs with no network link or shared writable files found a way to exchange synthetic secrets anyway. One agent evicted data from the CPU cache while the other measured reload times, using fast and slow reads as a binary signal to communicate covertly. The technique resembles the cache-timing methods behind the Spectre and Meltdown attacks, but required both agents to cooperate deliberately.

Guide details PC airflow setup to improve case cooling and reduce dust buildup

A new PC hardware guide outlines how fan placement and airflow direction affect cooling, recommending cool air intake from the front and bottom with hot air exhausted from the top and back. It explains that maintaining slightly positive air pressure inside a case helps minimize dust intake, and that fan size and speed both influence airflow pressure and noise levels.

AMD merges ZenDNN into Intel's oneDNN, joins UXL Foundation as steering member

AMD is folding its ZenDNN library, previously a downstream fork, into the open-source oneDNN project maintained by Intel under the UXL Foundation. AMD also joined UXL as a steering member, paying $20,000 annually for influence over the foundation's direction. The merger was recommended by the x86 Ecosystem Advisory Group, a joint Intel-AMD initiative formed in 2024.

Anthropic's Claude Code spots hidden xmrig crypto miner on developer's Mac

A Reddit user building a Mac temperature-monitoring app with Claude Code was alerted by the AI agent that xmrig, a crypto-mining program, was running unauthorized on their machine using roughly 1175% CPU across about 12 cores. Claude identified the malware had been added to startup around August 29 and offered to quarantine the files rather than deleting them outright.

Arm claims AGI CPU beats AMD EPYC 9755 2x in per-rack agentic AI tests

Arm announced results showing its AGI CPU delivers nearly double the performance per rack compared to AMD's EPYC 9755, based on modeled rack calculations across seven agentic AI workloads. The figures came from a new CPU evaluation methodology Arm developed with input from Fujitsu, Meta, and OpenAI, and were independently verified by Signal65.

Guide outlines how to safely manage Windows background processes

A how-to guide explains that spikes in CPU, RAM, or disk use from background processes aren't always a problem—sometimes they reflect legitimate activity like a launcher downloading updates. It recommends closing apps properly through their own quit options or the system tray before resorting to Task Manager's End Task, and suggests using Microsoft Defender's offline scan for persistent or suspicious processes.