Ponte vecchio, intel confirms its new gpu for the hpc sector 'exascale'

Intel Ponte Vecchio is confirmed and is the first GPU of the company focused on the HPC segment 'Exascale'

During Intel's SC19 event, Raja Koduri was onstage, showcasing the company's strategy for high-performance computing by 2021. Intel revealed that its new hardware is codenamed "Ponte Vecchio" and will be built on a 7nm process, as well as some other cool little bits.

Koduri has explained that Intel will have a single "Xe" architecture but multiple sub-architectures / micro-architectures (or as you want to characterize in a GPU) in order to target different parts of the market. Some for the ultra-portable market and others for AI acceleration and large-scale workloads. This clearly marks the versatility of the Intel Xe graphics architecture.

For high-performance computing, the presentation highlighted three key areas that the Xe architecture will address. The first is a flexible parallel vector array matrix engine, which plays into the hands of AI acceleration and AI training to a large extent. The second is Dual High Precision Performance (FP64), which has been dwindling a bit lately due to reduced precision AI workloads, but still an important requirement on traditional HPC workloads, like meteorology, oil and gas, and astronomy. (We should note that the diagram shows a block of 15 × 7 units, and Intel's Gen architecture uses 7 threads per execution unit.) The third of this trident is that Intel's HPC efforts will have a high cache and width. of memory band, which the slides suggest will be coupled directly to the individual compute chiplets to ensure rapid interconnection.

Ponte Vecchio, named after the bridge across the Arno River in Florence, Italy, will be Intel's first “exascale class” graphics solution and clearly uses both chiplet technology (based on 7nm) and Foveros / die methods. stacking. Ponte Vecchio will also use Embedded Multi-Die Interconnect Bridge (EMIB) technology to join the chiplets.

Visit our guide on the best graphics cards in the market

GPU-to-GPU communication will occur through a Compute eXpress Link (CXL) interface, layered on top of PCIe 5.0.

Ponte Vecchio will be the accelerator for the Aurora supercomputer, which will be installed at the Argonne National Laboratory in 2021. Intel has announced that it will have two Sapphire Rapids CPUs (the one that follows Ice Lake), along with six Ponte Vecchio GPUs, in a single node. Last but not least, in one of the slides (above) you can also see the 'Gaming' segment among Intel's plans, so it is confirmed that we will have graphics cards for video games based on this architecture. We will keep you informed.

Anandtechtechpowerup font

Intel reinforces its position in the asic sector with the acquisition of the company easic

Intel intends to continue expanding its business model, for this it has announced the acquisition of eASIC, a leading provider of ASIC based Intel wants to continue expanding its business model, for this it has announced the acquisition of eASIC, a leading provider of ASIC based in Santa Clara.