Fri, Oct 18, 2019 - Page 2 News List

NARL cloud platform starts

By Chien Hui-ju and Sherry Hsiao  /  Staff reporter, with staff writer

Taiwan Mobile Co president Jamie Lin speaks at an event hosted by the Ministry of Science and Technology yesterday in Taipei.

Photo: CNA

The Taiwan Computing Cloud (TWCC) platform officially began commercial operations this month after a soft launch in May, and its efficiency is estimated to up to 8 times better than existing platforms on the market, the National Applied Research Laboratories (NARL) said yesterday.

The platform was developed by a team that included members of NARL’s National Center for High-performance Computing (NCHC) — an affiliate of the Ministry of Science and Technology — and local firms Quanta Computer, ASUS and Taiwan Mobile.

The platform runs on the supercomputer Taiwania 2, which was also built by the team and was named the 23rd most powerful supercomputer by the Top500 Supercomputer List in June.

Half of TWCC’s computational resources would be reserved for academic use, while the other half would be available for use by businesses for innovation, the ministry said yesterday in Taipei.

Businesses would pay NT$82 per hour for each graphics processing unit (GPU), the center said, calling it a roughly “15 percent discount” on the market price.

Academic institutions would be charged based on a separate pricing structure, it added.

More than 150 projects have participated in the testing of the platform since the trial began in May, NCHC Director-General Shepherd Shi (史曉斌) said.

The platform has operated for more than 8 million GPU core-hours, with a system usage rate of more than 90 percent, he said.

Through continued improvements, TWCC now has a 40 percent to 800 percent lead in several “test items” over other platforms in terms of its big data parallel processing performance, he added.

TWCC’s parallel processing performance is 40 percent higher than that of other platforms, while its artificial intelligence (AI) object detection training is eight times faster, the NCHC said.

Giving examples of how TWCC has helped businesses and startup teams, Shi said the platform has helped shorten Bovia Co’s AI model training time by 90 percent, reduced Taipei Veterans General Hospital’s medical imaging processing time to one sixth of what it was originall.

It has also increased aetherAI’s deep learning efficiency by nearly 500 times, he said.

Quanta Computer senior vice president Mike Yang (楊麒令) said a service like the TWCC could help overseas Taiwanese businesses returning to Taiwan adjust to having a smaller staff, as well as improve the quality of their AI.

Comments will be moderated. Keep comments relevant to the article. Remarks containing abusive and obscene language, personal attacks of any kind or promotion will be removed and the user banned. Final decision will be at the discretion of the Taipei Times.

TOP top