7nm Intel Xe GPUs Codenamed “Ponte Vecchio”

Intel’s first Xe GPU constructed on the corporate’s 7 nm silicon fabrication course of might be codenamed “Ponte Vecchio,” in line with a VideoCardz report. These aren’t gaming GPUs, however moderately compute accelerators designed for exascale computing, which leverage the corporate’s CXL (Compute Express Link) interconnect that has bandwidth similar to PCIe gen 4.0, however with scalability options slated to come back out with future generations of PCIe. Intel is getting ready its first enterprise compute platform that includes these accelerators codenamed “Project Aurora,” by which the corporate will exert end-to-end management over not simply the {hardware} stack, but additionally the software program.

“Project Aurora” combines as much as six “Ponte Vecchio” Xe accelerators with as much as two Xeon multi-core processors primarily based on the 7 nm “Sapphire Rapids” microarchitecture, and OneAPI, a unifying API that lets a single type of machine code tackle each the CPU and GPU. With Intel proudly owning the x86 machine structure, it is probably that Xe GPUs will function, amongst different issues, the flexibility to course of x86 directions. The API will be capable of push scalar workloads to the CPU, and and the GPU’s scalar items, and vector workloads to the GPU’s vector-optimized SIMD items. Intel’s essential pitch to the compute market might be considerably lowered software program prices from API and machine-code unification between the CPU and GPU.