Lenovo and NVIDIA Partner on New Hybrid AI Solutions

[SPONSORED GUEST ARTICLE] Lenovo unveiled the expansion of its ThinkSystem AI portfolio designed for the most demanding AI, data analytics and HPC workloads, featuring two powerful 8-way NVIDIA GPU systems combining massive computational capabilities with power efficiency. Engineered for generative AI, the Lenovo systems incorporate support for the NVIDIA HGX AI supercomputing platform, including NVIDIA H100 and H200 Tensor Core GPUs and the new Grace Blackwell….

Eviden Delivers 104 Pflops Nvidia-Powered EXA1-HE Supercomputer for CEA

Paris – April 17, 2024 – CEA (The French Alternative Energies and Atomic Energy Commission) and Eviden, the Atos advanced computing business unit, today announce the delivery of the EXA1 HE supercomputer, based on Eviden’s BullSequana….

HPC News Bytes 20240415: Intel Gaudi 3, Meta’s MTIA Chip, Easing GPU Shortages, AI Category Theory, China’s Growth Strategy

Happy Tax Day to you! Here’s a quck (6:23) romp through recent news from the world of HPC-AI, including: Intel’s Gaudi 3 GPU and Xeon-6 CPU, Meta’s new accelerator chip, GPU shortage easing….

TSMC in the USA: Up to $6.6B in CHIPS Act Funding for 3rd (2nm or More) Arizona Fab

TSMC announced it has signed a preliminary agreement with the U.S. Dept. of Commerce for up to $6.6 billion under the CHIPS and Science Act. TSMC also announced plans to build a third fab at its TSMC Arizona site in  Phoenix using 2nm or more processes….

HPC News Bytes 20240401: A $100B AI Data Center, Eviden Says It’s Healthy, Alibaba’s RISC-V Chip, New Optical Interconnect Group, Nvidia Fights CUDA Translation

Happy April Fool’s Day! It was as always an interesting week in the world of HPC-AI, this edition of HPC News Bytes includes commentary on: Microsoft and….

@HPCpodcast: Matt Sieger of OLCF-6 on the Post-Exascale ‘Discovery’ Vision

What does a supercomputer center do when it’s operating two systems among the TOP-10 most powerful in the world — one of them the first system to cross the exascale milestone? It starts planning its successor.  The center is Oak Ridge National Lab, a U.S. Department….

Achronix FPGAs Add Support for Bluespec’s RISC-V Chips

March 26, 2024, SANTA CLARA, Calif. & FRAMINGHAM, Mass.– FPGA technology company Achronix Semiconductor Corporation and Bluespec, Inc., a RISC-V tools and silicon IP company, today announced a family of Linux-capable RISC-V soft processors that are available for the Achronix Speedster7t FPGA family. Marking an industry first, Bluespec’s RISC-V processors now integrate into the Achronix 2D […]

At GTC 2024: Liqid UltraStack Servers with up to 20 NVIDIA GPUs

At GTC 2024 we caught up with Sumit Puri, co-founder of composable computing company Liqid, to discuss the company’s GPU-power UltraStack, which he said transforms 2U servers into high-density GPU systems powered by NVIDIA L40S GPUs.

HPC News Bytes 20240325: Final GTC Thoughts, Intel and CHIPS Act Largesse, Ultra Ethernet Consortium Expands, Samsung’s GPU/HBM Chip

A happy end-of-March morning to you! Here’s a rapid (5:57) run-though of the latest news in HPC-AI, including: final reflections on Nvidia GTC 2024, Intel to receive $8.5 billion via US CHIPS Act, the expanding Ultra Ethernet Consortium, Samsung’s upcoming GPU/HBM blend.

@HPCpodcast at GTC 2024: Taking in the Scene at Nvidia’s Week at the Center of the Tech World

Live from Nvidia’s GTC 2024 AI-everywhere show, Doug and Shahin get together in the usual coffee shop chat of the @HPCpodcast.  We cover a wide range of announcements from Nvidia headlined by their new Blackwell GPU architecture, discuss the contrast between the Nvidia model and thoe of other players, and share some observations (including celebrity sitings and a stare-down between a dog and a robot….