Intel’s Aurora supercomputer is expected to exceed 2 ExaFLOPS performance

As part of Intel’s 2021 Innovation Event, the company offered a brief update on the Aurora supercomputer, which Intel is building for the Argonne National Laboratory. The first of two U.S. exascale supercomputers under construction, Aurora and its critical processors are finally merging, allowing Intel to finally narrow down its performance projections. As it turns out, the 1-and-change exaFLOPS system will look more like a 2-exaFLOPS system – Aurora’s performance is high enough that Intel now expects the system to exceed 2 exaFLOPS double-precision performance.

Planned to be the first of two public exascale systems in the U.S., the Aurora supercomputer has gone through a turbulent development process. The contract was originally awarded to Intel and Cray back in 2015 for a pre-exascale system based on Intel’s Xeon Phi accelerators, a plan that failed when Intel stopped developing Xeon Phi. Instead, the Aurora contract was renegotiated to become an exascale system based on a combination of Intel’s Xeon CPUs and what became their Ponte Vecchio Xe-HPC GPU. Since then, Intel has been working to the end to make the necessary silicon to make the delivery time already shifted from 2020 to 2021 to 2022 (ish), going as far as making Ponte Vecchia parts on TSMC’s competing 5nm technology. process.

But it seems the light is finally at the end of the tunnel. As Intel strives to complete the system, its performance is ahead of expectations. According to the chipmaking company, they now expect the assembled supercomputer to be able to deliver over 2 exaFLOPS dual precision performance (FP64). The system previously did not have a specific performance figure attached to it, other than the fact that it would be over 1 exaFLOPS in FP64 bandwidth.

This higher performance figure for Aurora comes thanks to Ponte Vecchio, who, according to CEO Pat Gelsinger, is exaggerating in terms of performance. Gelsinger did not go into further detail how The Ponte Vecchio exaggerates, but given that the IPC and overall efficiency are relatively easy to determine during simulations, the most likely candidate is that the Ponte Vecchio has a higher clock speed than Intel’s previous projections. The Ponte Vecchio was one of the first HPC chips (and the first Intel GPU) built on TSMC’s N5 process, so there were a lot of unknowns coming into this project.

For Intel, this is without a doubt a welcome fortune for a project that has encountered many obstacles. Repeated delays have already allowed rival AMD to get the honor of delivering the first exascale system with Frontier, which is currently being installed and is expected to offer 1.5 exaFLOPS performance. So while Intel will no longer be the first, once Aurora hits the network next year, it will be faster than these two systems.

Source link

Naveen Kumar

Friendly communicator. Music maven. Explorer. Pop culture trailblazer. Social media practitioner.

Related Articles

Leave a Reply

Your email address will not be published.

Back to top button