Intel is already testing Ponte Vecchio, its first 7nm Xe GPU

At the end of last year Intel announced that in 2021 it would launch its new high-performance “Ponte Vecchio” GPUs, based on Xe2 at 7nm, and it seems that they are already beginning to test them according to a new registration before the EEC.

According to this record, Intel is testing three “Ponte Vecchio RVP Pre-Alpha” cards, where RVP stands for Reference Validation Platform, or Reference Validation Platform. This means that it is a very early prototype that simply seeks to validate the operation of the GPU, although neither the PCB nor the cooling use the final design.

Unfortunately, the list of these three models does not tell us much, since no one yet knows how to decipher their codes GAPV3KI2TC, GAPV4KI2TC, and GAPV5KE2TC. However this shows us that Intel is not only working on one variant for launch, but at least three.

What we know so far about Ponte Vecchio

These GPUs will have a multi-die design, which will allow connecting up to 16 DIEs per GPU communicated through Xe Link, a technology based on CXL. They will also have what Intel calls "Rambo Caché", which is a huge cache block that all GPUs will be able to access together, for greater performance, since what Intel calls XEMF (Xe Memory Fabric) that uses the same concept but with VRAM.

These GPUs will debut on the Aurora supercomputer, which will feature 6 of these GPUs of up to 16 DIEs, along with two next-generation Xeon "Sapphire Rapids", achieving maximum performance thanks to special interconnections between CPU and GPU.

Are you eager to see the Ponte Vecchio GPUs at 7nm from Intel?

Source: PCGamer