45k user reviews & 1 million votes, over 29k web hosting brands & 85k plans since 2004!
X
Menu
WHTop → 📋 News → New Open GPU Standard Set to Reshape AI Computing
Apr, 2025 : New Open GPU Standard Set to Reshape AI Computing
📅 - A group of leading technology firms has unveiled a new open GPU interconnect standard designed to dramatically improve the performance, efficiency, and scalability of AI infrastructure. The UALink Consortium has officially ratified the Ultra Accelerator Link (UALink) 200G 1.0 Specification - an open, low-latency, high-bandwidth standard for connecting accelerators within AI computing clusters.
Established last summer, the Ultra Accelerator Link (UALink) Consortium unites some of the leading companies in data centers and semiconductors, such as Intel, AMD, and Microsoft, to create standards to improve linking numerous GPUs.
The UALink 200G 1.0 standard introduces a memory-semantic [...][... Check source for end of article ...]
Tags: AI
Reads: 213 | Category: General | Source: Hosting Jurnalist : Hosting Jurnalist | Author: URL source: https://hostingjournalist.com/news/new-open-gpu-standard-set-to-reshape-ai-computing
Want to add a website news or press release ? Just do it, it's free! Use add web hosting news!
Related news
📅 - IBM-Backed Anderon Wins $1B CHIPS Award for Quantum Wafer Manufacturing - Anderon, an IBM company, has finalized a $1 billion CHIPS and Science Act award with the U.S. Department of Commerce to expand quantum wafer manufacturing in Albany, New York, while IBM is committing another $1 billion. The 300-millimeter foundry is already processing its first quantum wafers, marking a concrete step toward industrializing a supply chain still dominated by specialized research. The federal money is aimed at research and development around domestic quantum wafer manufacturing, not construction of a conventional high-volume logic fab. Still, the scale of the funding is difficult to ignore. Anderon is being positioned as a pure-play manufacturing operation that can fabricate [...]
📅 - Swiss Quantum Hub Targets Industry With IBM Quantum System Two - IBM, Lockheed Martin and ETH Zurich are establishing a Swiss quantum innovation hub that will give researchers and industry access to an IBM Quantum System Two in Lugano. The machine, due by the end of 2026, will be Switzerland's first dedicated IBM quantum computer and is intended to support research, training and early industrial experimentation on site.
The installation will sit at the Swiss National Supercomputing Centre, putting dedicated quantum hardware alongside one of Europe's better known high performance computing environments. IBM will operate the system, while ETH Zurich will provide access, expertise, technical resources and links into Swiss academia and industry. [...]
📅 - BQP Raises $8M to Target GPU Efficiency and Quantum Workloads - IBM Ventures has invested in BQP, bringing the physics software company's total funding to $8 million as it expands a platform designed to extract more useful work from existing CPU and GPU infrastructure while preparing for future quantum acceleration. For infrastructure operators, the near-term story is less quantum computing than better utilization of costly compute already installed.
BQP is addressing an awkward problem inside high performance computing. The company argues that conventional physics solvers can leave roughly 85% of GPU floating point capacity unused because many simulation workloads are constrained by memory access rather than raw arithmetic capability. Its BQPhy [...]
📅 - IBM Nighthawk r2 Raises Quantum Throughput Without More Qubits - IBM has released Quantum Nighthawk r2, a 120-qubit processor that raises circuit throughput to more than 100,000 circuits per second while preserving Quantum Heron-class gate fidelity. The machine adds reset hardware, cuts initialization errors sharply, and targets deeper workloads, giving enterprises and researchers more quantum compute without increasing programmable qubit count or changing the architecture.
The number to watch is not 120. IBM has held the programmable qubit count steady from the first Nighthawk generation and instead attacked a less glamorous bottleneck – the dead time between repeated circuit executions. Earlier systems could spend hundreds of microseconds waiting for [...]
📅 - Aranya Raises $11M to Automate Bare Metal GPU Infrastructure - Aranya has raised $11 million to automate the conversion of bare-metal GPU infrastructure into production-ready AI clusters, claiming deployments in under 48 hours. The startup is targeting inference providers, AI companies, neoclouds, and data centers that own expensive hardware but still need networking, orchestration, observability, and operational tooling before those servers can reliably run workloads at scale.
The financing consists of a $9 million seed round led by First Round Capital and an earlier $2 million pre-seed round led by Asylum Ventures. Box Group, Vermilion Cliffs, Founder Collective, Parable VC, Uncommon Ventures, and other investors participated. Aranya says it is [...]