HPC luminary Jack Dongarra’s fascinating comments at SC22 on the low efficiency of leadership-class supercomputers highlighted by the latest High Performance Conjugate Gradients (HPCG) benchmark results will, I believe, influence the next generation of supercomputer architectures to optimize for sparse matrix computations. The upcoming technology that will help address this problem is CXL. Next generation architectures will use CXL3.0 switches to connect processing nodes, pooled memory and I/O resources into very large, coherent fabrics within a rack, and use Ethernet between racks. I call this a “Petalith” architecture (explanation below), and I think CXL will play a significant and growing role in shaping this emerging development in the high performance interconnect space.
Cerebras Announces 2 AI Partnerships – for Large Language Model Training and ‘Generative AI’ Content Creation
Cerebras Systems, maker of the “dinner plate sized” AI processor, announced two alliances today, one with Cirrascale Cloud Services, provider of deep learning solutions for AVs, NLP and computer vision, and with Jasper, maker of an AI content platform for AI-based copywriting and content creation. Under the Cirrascale-Cerebras partnership, the two companies announced the availability […]
TOP500: Frontier Maintains Big Lead, Europe at Nos. 3 and 4, China Quiet
The new TOP500 list of the world’s most powerful supercomputers, released today at the SC22 conference in Dallas, while short on surprises underlines several significant HPC trends. First the headline: the HPE-built, AMD-powered Frontier system, which was crowned the world’s first exascale-class system when the previous TOP500 list was released last spring, remains at the top of the list, delivering nearly three times the power of its nearest rival on the list. Frontier remains at 1.102 exaFLOPS….
SiPearl and AMD in Partnership for Exascale Supercomputing in Europe
Maisons-Laffitte (France), 14th November 2022 – SiPearl, the HPC microprocessor designer for European supercomputers, and AMD announced a joint offering for exascale supercomputing in Europe combining SiPearl’s HPC microprocessor, Rhea, with AMD Instinct accelerators. The latter chip, along with AMD EPYC CPUs, power Frontier, the world’s first exascale-class system at Oak Ridge National Laboratory. Initially, […]
It’s Official: Intel Announces Ponte Vecchio and Sapphire Rapids, Hints at ‘Falcon Shores’ XPU in 2024
Intel today officially announced its long awaited, next-generation server chips: the Xeon CPU Max, known to date as “Sapphire Rapids,” and the Data Center GPU Max, which for more than two years has been known – and been a focal point of scrutiny, speculation and frustration – as “Ponte Vecchio.” Both chips have been delayed, […]
@HPCpodcast: Intel Shipping Aurora Blades; Rocky Linux vs. CentOS; Tesla’s ‘Dojo’ AI Supercomputer
In this edition of the @HPCpodcast, Shahin and Doug kick things off on the news from Intel Innovation Day that the company is shipping Sapphire Rapids CPU / Ponte Vecchio GPU-powered blades to the Argonne Leadership Computing Facility for use in the lab’s Aurora exascale supercomputer, due for delivery by late 2022 or early 2023. It’s encouraging news, but still …
Samsung: 1.4 nm Process Technology in Production by 2027
At its Foundry Forum today in San Jose, Samsung Electronics spotlighted its foundry business strategy with emphasis on high performance computing (HPC), artificial intelligence (AI), 5/6G connectivity and automotive applications. A featured announcement: Samsung committed to bringing 1.4-nanometer (nm), for production in 2027. That will be preceded by the introduction of 2nm process in 2025 […]
Supermicro Announces 8U ‘Universal GPU’ Server for NVIDIA H100’s
HPC-AI server maker Supermicro today announced what the company said is its most advanced GPU server incorporating eight NVIDIA H100 Tensor Core GPUs. Supermicro now offers three Universal GPU servers: the 4U, 5U and the new 8U. The Universal GPU platforms also support Intel and AMD CPUs up to 400W, 350W and higher, according to […]