https://www.nextplatform.com/2023/03/14/whos-going-to-build-the-uks-homegrown-exascale-supercomputer/ In-depth coverage of high-end computing at large enterprises, supercomputing centers, hyperscale data centers, and public clouds. Tue, 21 Mar 2023 14:59:14 +0000 hourly 1 https://wordpress.org/?v=6.5.5 https://www.nextplatform.com/2023/03/14/whos-going-to-build-the-uks-homegrown-exascale-supercomputer/#comment-205932 Thu, 16 Mar 2023 20:04:07 +0000 https://www.nextplatform.com/?p=142055#comment-205932 Graphcores filed accounts look like they are earning close to nothing given the billion dollars plus of investment they have taken. Hope the government not planning on subsidising poor execution with taxpayers money to cover up their incompetence in coming up with a national semiconductor strategy a decade too late.

]]> https://www.nextplatform.com/2023/03/14/whos-going-to-build-the-uks-homegrown-exascale-supercomputer/#comment-205904 Wed, 15 Mar 2023 23:57:09 +0000 https://www.nextplatform.com/?p=142055#comment-205904 It seems to me (to some extent at least) that the British MP is mixing metaphors, conflating exascale computing with UK industrial development, resulting in a quantum paradox of buzzword bingo — as examplified by this very sentence. The confusion (in my mind) is between means and ends, carts and horses, or pits and pendulums, knotted together by a gordian catch-22. If the program goal is to foster the advancement of UK science and industry (end) by providing engineers and scientists with a public exaflop facility (means), then it would be most logical to “simply” purchase an HPE-Cray system based on MI300A, at $300M Approx., to be installed by X-mas ’23, and operational in January ’24. AMD, HPE-Cray, Supermicro, Intel, NVIDIA, IBM, and Fujitsu-NEC have essentially licked that challenge, and the exascale vanity boat has all but sailed at this juncture, horses left the barn, the pendulum has swung (those of them who aren’t quite there yet, will be, in a matter of months, not years). Accordingly, Exascale computing is now manufacturing reality, not research, and accordingly, a plan to foster the development of UK science and industry (end) by sponsoring the development of a UK-made Exascale computer (means, or end?) is, at best, today, steam-punk! Might as well sponsor development of Babbage’s analytical engine, and then use that as the basis for UK HPC, AI and cloud computing. As discussed last week on this very Next Platform, the contemporary target for research and industry development is now the Zettascale. Neither Robin Hood, nor William Tell, could possibly reach the apple of knowledge rising from Newton’s contributions if they were aiming any lower!

]]> https://www.nextplatform.com/2023/03/14/whos-going-to-build-the-uks-homegrown-exascale-supercomputer/#comment-205898 Wed, 15 Mar 2023 18:06:26 +0000 https://www.nextplatform.com/?p=142055#comment-205898 Seems unlikely, HMG prefers to sponsor our competitors and watch British competence dwindle. Even better I’d to start a project, get cold feet, keep changing the spec then pull out and blame the supplier.. who is left carrying the can

]]>