Access to the Super Computer
Starting from 2018, LightOn’s innovation has been used effectively by a lot of scientists. However, over the following few months, its optical coprocessor will be made accessible to a small group of Jean Zay researchers who will use it to do investigation on machine learning principles, differential privacy, satellite imaging evaluation, and natural language processing (NLP) activities. Also Read: Chinese Companies Play a Significant Role in Reducing Mobile Phone Imports
LightOn’s Co-Processor
Image Credits: HPCwire The Optical Processing Unit (OPU) from LightOn utilizes photonics to accelerate arbitrary algorithms on a massive scale. It does, however, function with regular silicon CPUs, including Nvidia’s A100 GPU hardware. The Aurora 2 OPU is used to operate the firm’s Appliance combined computing unit, which is housed in a 2U form size and can be swiftly and easily incorporated within data centers or supercomputers. LightOn claims that their Appliance can achieve maximum performance of 1.5 PetaOPS @ 30W TDP, which is 8 to 40 times faster than GPU-only accelerator. In a press statement, the CEO of LightOn, Igor Carror said, “This pilot initiative combining a new computing technology inside one of the planet’s Supercomputer would not be conceivable without the specific dedication of innovative agencies including GENCI and IDRIS/CNRS. Our world premiere, combined with the rise of Quantum Computing, confirms our belief that the next phase beyond exascale high-performance computing will be mixed computing.” You may be also interested in: Geforce Now and Google Stadia to arrive on Samsung’s new TVs
