Supercomputer in Office: Really or Dream?
For a long time, there have been prolonged discussions and a range of opinions about the definition of “supercomputer” and its application benefits to the economy. In narrow sense, this boils down to the question is it possible supercomputers to be practically integrated in enterprise environment reaching out to everyone including SMEs and broader audience.
Whereas in its intrinsic nature, supercomputers will always stay out of reach to the general public due to constant push and therefore shift of the frontier boundaries in the different time periods, we may see a niche for a particular intermediary. A level that stays between the fastest supercomputers and desktop machine. Tempting to say halfing the way, but it may be more reasonable to think of it as dynamic process wiggling around the middle distance in different point of time.
In such way, we may see NVIDIA last efforts to “design a supercomputer for deskside”. According to design envelope, it should “draw power from a normal wall socket, use plug-and-play GPU-accelerated software and operate not much louder than a whisper.”
NVIDIA dubbed it DGX Station with the goal to become “world’s fastest workstation for leading-edge AI development” and “personal AI supercomputer under the desk”. The entire endeavor has its justifications. IDC estimates that for every $43 spent on HPC, users can generate $515 in revenue, making HPC adoption a highly lucrative investment. Furthermore, IDC study finds that “Organizations are looking to the new-generation of HPC systems to provide the performance, reliability, and flexibility that they need.” Definitely, the new system would offer a greater flexibility and “ease-of-use” in comparison with the traditional HPC system. With its 500 teraFLOPS of AI power, equivalent to hundreds of servers, DGX is well suited to for the role of intermediary in the computational hungry value chain. Especially, in segment where the number of active US startups developing AI systems has increased 14x since 2000.
It would be interesting to see whether some of the other vendors would accept the challenge and offer its own vision for this emerging market niche or would provide alternative paradigm to satiate the enterprise computational need in pursuit of their AI endeavors.
L3C is a proud partner of NVIDIA in its efforts to provide AI capacity and capability to enterprise users in UK. To explore more about AI and Machine Learning based on NVIDIA V100, why not take a tour of our AI services?