The Nvidia supercomputer

Dean Takahashi for VentureBeat writes:

Nvidia is collaborating with AstraZeneca, maker of one of the COVID-19 vaccines, to fuel faster drug discoveries and creating a transformer-based generative AI model for chemical structures. Transformer-based neural network architectures, which have become available only in the last several years, allow researchers to leverage massive datasets using self-supervised training methods, avoiding the need for manually labeled examples during pre-training.  

Cambridge-1 uses an Nvidia DGX SuperPod supercomputing cluster. Nvidia hopes it could have a global impact on health care around the world and contribute to the continuing efforts to fight COVID-19.

(emphasis mine)

Nvidia is best known for GPU business and now in the process for merger with ARM Holdings. I am very excited about the future of ARM based chips- especially as they catch up with their x86 counterparts with significantly less power resources. It doesn’t mean that I am advocating Apple Mac with the M1- it is the first generation with significant flaws to make it generally usable for my needs- but the underlying technology will push the evelope for the consumers.

The reason ARM and GPU based systems are exciting is because traditionally they were “separate” with each requiring its inputs to give a combined output. Motherboards, therefore represented a significant barrier to achieve higher “output”. As the systems are being fused into one-called “system on a chip” or SoC, higher throughput is now feasible than previous hardware builds. These designs are getting better with state of the art manufacturing- so called 3 and 5 nm “plants”. Is x86 system doomed? Far from it. However, it represents the first salvo of a ARM based system in the consumer space. There have been several instances where PLayStations were combined to form super computing clusters. The linked article from VentureBeat is rehash of a press release but is significant foray of this architecture in the healthcare.

If you ignore the references to COVID19 (it’s a SEO bait), it becomes imperative that super computing pods become cheaper to own and rent. NVidia will extract all the mileage from this standalone facility to milk the enterprises by attempting to create a solution in search of a problem. Most enterprises are being seduced by the idea of 5G and “AI-revolution”. Yes, it is true that technology is advancing. But the reader must be aware of the hype cycles. I have only linked the article for its hardware chops that GPUs are now mainstream. I am closely watching this space because it becomes cheaper (and more efficient) to run AI algorithms at scale using these instruction sets.