Intel HPC and supercomputers: realignment and reaching for Zettascale

0
293

Intel is cleaning up another department: High Performance Computing (HPC) and Supercomputing (SC). The previous boss Trish Damkroger has been dismissed, the department will be split up from now on, and new and at the same time huge goals for the future will be defined.

Exascale was the previous goal of Intel, which should be achieved with Aurora in the form of a first supercomputer with more than one ExaFLOP performance. But this project is stuck, and huge. Delays in almost all areas, but above all the CPU Sapphire Rapids and GPU Ponte Vecchio have effectively turned it into a product for the year 2022/2023. This means that the previously targeted goal of being the first computer of this kind is even in danger, because the all-in-AMD-powered El Capitan solution seems to be closer to the schedule and will now be finished almost at a similar point in time. In the meantime, the US government even had to order another new AMD Nvidia system, while Intel paid the first fines for services not provided.

It was foreseeable that this would not remain without consequences for Intel in the end. Above all, the new CEO Pat Gelsinger has been tidying up every department since he started working in the spring, as this recently less laudable figurehead was an easy target for renovations. The fact that the previous boss Trish Damkroger does not have to leave the company is probably due to the fact that she had not even held the position for two years and only inherited it from her predecessor Rajeeb Hazra after his retirement at Intel and current employer Micron. By then the child had long since fallen into the well.

From now on the division is divided into two departments: The Super Compute Group is led by Jeff McVeigh, the Super Compute Platform Engineering Group from Brijesh Tripathi. They are all part of the Accelerated Computing Systems and Graphics Business Unit (AXG), which is led by Raja Koduri. Ultimately, his power in the group is steadily increasing. The goals, which are mentioned by the head of the Super Compute Platform Engineering Group, among others, are once again huge: Access to exascale computing for everyone and within this decade to reach the next level, zettascale.