SKT doubles supercomputer capacity to speed up launch of Korean ChatGPT

“Titan,” the supercomputer [Photo provided by SK telecom]

SK telecom Co. (SKT) has doubled the capacity of its supercomputer, which serves as the brain of its artificial intelligence model AIDAT, as competition mounts to develop and launch a chatbot that relies on generative AI after the release of Open AI’s ChatGPT.

SKT announced Sunday that “Titan,” the supercomputer that is the basis for its super-giant AI advancement since 2021, expanded its capacity to 1,040 NVIDIA A100 GPUs.

Titan supports performance of more than 17.1 petaflops. One petaflop can calculate speed of a computer equal to one 1,000 trillion times per second, and 17.1 petaflop means the computer is capable of computing 17.1 trillion times per second.

SKT expects the capacity expansion to enable more sophisticated learning of AIDAT than before.

SKT commercialized the country’s first commercial version of its self-developed super-large Korean-language model GPT-3 with a beta launch of AIDAT in May of last year. The company plans to strengthen knowledge-based conversation services by applying the ChatGPT model within this year.

The expanded supercomputer capacity will play a major role in improving AIDAT’s conversation flow and answer to a human level, said Kim Young-joon, in charge of SKT’s AIDAT Team.

SKT’s supercomputer Titan ranked 92nd in the world’s top 500 supercomputer rankings announced at SC22 in Dallas, Texas last year.