The insatiable demand for AI processing is giving rise to new alternatives. New companies are jumping in. And Intel, not one to be left behind, has unveiled a selection of new creations.
Starting with Ponte Vecchio, which is something the chip giant has been working on for quite some time.
And it has finally made it official now.
Intel unveiled its latest effort at SC19, with the company’s GPU chief taking to the stage to talk about its plans for GPU-driven HPC and AI initiatives. The existence of the long-rumored Ponto Vecchio GPU was also confirmed at the event.
This chip is based on the Intel Xᵉ architecture, a 7nm affair that will be at the heart of all GPU solutions from the company.
Named after Ponte Vecchio, a stone bridge that overlooks the Arno River in Florence, Italy, this one will not be a gaming graphics card upon release, but instead a solution designed for high-performance computing and artificial intelligence.
Not unlike the NVIDIA Tesla V100 unit.
The company is going all in for running AI workloads on a range of different hardware, largely in the cloud. And to that end, it has also demonstrated the power that the oneAPI technology brings to the table, in how it integrates all this hardware seamlessly into the software ecosystem.
The idea being to overcome the complexities associated with working across multiple architecture.
Developers will be able to write their AI code once, and run it anywhere. They will just need to target the basic hardware type, for example, CPUs, GPUs, or FPGAs. Beyond that, oneAPI will keep the code optimized for any supported hardware variant.
The same code will run on either the NVIDIA Tesla V100 or Ponte Vecchio.
This, the company hopes, will address one of the largest pain points of artificial intelligence and machine learning development, and allow coders to focus on their actual project and re-use the same code on different hardware.
The oneAPI toolkit is already available in beta form, and developers can check it out here.