SuperMUC-NG getting Intel Ponte Vecchio improve

The SuperMUC-NG supercomputer at the moment websites fairly comfortably at quantity 15 within the Top500. It has been well-placed for a number of years however is beginning to slip within the rankings just a little. The present design relies upon getting older however mightily highly effective mixture of (holds breath); 6,336 skinny nodes, every with twin Intel Xeon Skylake CPUs and 96GB of reminiscence, alongside 144 fats nodes, geared up with the identical CPUs bolstered by an additional 672GB of reminiscence per node. It is a water-cooled system based mostly on the Leibniz Supercomputing Centre (LRZ) in München, Germany, and options 70 petabytes of storage, delivers 19.5 Linpack petaflops.

Phase 2 of SuperMUC-NG has been introduced this week, and it’ll contain a lot of shiny new Intel {hardware}. The upgrades are coming not simply to the CPUs, however with a stack of latest GPU know-how to speed up and advance the supercomputer’s AI capabilities.

Key enhancements in Phase 2 are as follows:

  • Intel Xeon ‘Sapphire Rapids’ CPUs,
  • Intel Ponte Vecchio GPUs,
  • Lenovo SD650-I v3 server platform,
  • Intel distributed asynchronous object storage (DAOS) powered by Xeon ‘Ice Lake’ CPUs and Intel Optane Persistent Memory 200-series,
  • The unified oneAPI software program stack.

HPC Wire feedback that the appliance of Ponte Vecchio GPUs in LRZ will solely be the second such announcement following that made by the Argonne National Laboratory’s forthcoming Aurora exascale supercomputer. The DAOS system is predicted to reach this autumn however the majority of the brand new compute system will not be put in till spring 2022. Needless to say, the researchers and scientists are making ready for the upgrades with new code and algorithms.

The change within the structure of the SuperMUC-NG coincides with the change in demand from its customers, in response to Dieter Kranzlmüller, director of LRZ. “Over the last years, we’ve observed our users accessing our systems not only for classical modeling and simulation, but increasingly for data analysis with artificial intelligence methods,” commented Kranzlmüller.

In associated information Intel’s Raja Koduri has been teasing ‘Ponte Vecchio Eye Candy’ on his private Twitter feed. Yesterday the ex-AMD engineer shared an exploding diagram of the superior package deal chiplet-based GPU.

What do you think?

Written by Gideon


Leave a Reply

Your email address will not be published. Required fields are marked *


Ten Years On: Portal 2

Ten Years On: Portal 2

Statistics Canada launches 11 playlists to celebrate the 2021 Census

Statistics Canada launches 11 playlists to rejoice the 2021 Census