AsianScientist (Sep. 22, 2021) – It goes with out saying that your morning commute in Hanoi, Vietnam, can be vastly totally different from what it might be in Houston, Texas. There have been over 276 million autos on US roads in 2019, a determine that works out to 842 vehicles per 1,000 individuals. In distinction, motorbikes dominate the roads in Vietnam, which has simply 23 vehicles per 1,000 individuals—practically 37 occasions much less.
But in terms of coaching autonomous autos (AV), the overwhelming majority of information comes from the US and Europe. VinFast, a Vietnamese automotive startup with its sights set on the worldwide market, is hoping to make use of supercomputers to bridge the distinction.
“Once I got here again to Vietnam, I needed to relearn learn how to drive right here—the site visitors situations are very totally different from the US,” mentioned Dr. Hung Bui, a synthetic intelligence (AI) researcher previously with Google DeepMind and Adobe Analysis, and at the moment the director of VinAI, the AI analysis arm of VinFast’s guardian firm, Vingroup.
“After some time I bought the cling of it, however it bought me considering a machine in all probability will do a fair higher job—Vietnam’s driving situations present the last word problem for methods making an attempt to achieve Degree 5 autonomy,” he mentioned, referring to completely autonomous vehicles that won’t even require steering wheels.
Serving to VinFast alongside is the nation’s strongest AI supercomputer: VinAI’s NVIDIA DGX SuperPOD. Branded because the world’s first cloud-native supercomputer, the DGX SuperPOD shall be used to retrain VinAI’s driving notion system as new knowledge arrives each 24 hours, Hung mentioned.
Supercharging R&D
World-leading AV methods constructed by non-public firms in Southeast Asia are simply one of many many ways in which excessive efficiency computing might influence the area. In gentle of the continuing COVID-19 pandemic, for instance, probably the most urgent wants is in drug improvement and healthcare. In Japan, the Fugaku supercomputer has been used to mannequin the conduct of virus particles spreading amongst individuals.
“Excessive efficiency computing gives researchers with great energy to deal with many grand scientific challenges,” mentioned Dr. Rossen Apostolov, government director of BioExcel, a Centre of Excellence for Computational Biomolecular Analysis on the PDC Centre for Excessive Efficiency Computing, KTH Royal Institute of Expertise in Stockholm, Sweden.
“Supercomputers permit screening of hundreds of biomolecules in a matter of days for growing a remedy or a vaccine towards the virus. Equally, modeling the propagation of a viral an infection by means of a inhabitants permits for well timed and efficient measures to decreasing its unfold,” he added.
Utilizing supercomputers to higher perceive the coronavirus comes amid rising HPC exercise within the hemisphere. June 2021 will mark a 12 months since Asia returned to the main place within the TOP500 supercomputer rankings, with Fugaku—constructed by Fujitsu and analysis middle RIKEN—bringing Japan again to No. 1 for the primary time since 2012.
Except for US machines, the intervening decade has additionally seen Chinese language methods—Sunway TaihuLight and Tianhe-2A—main the pack, leading to a higher Asian presence within the discipline of excessive efficiency computing.
The Affiliation of Southeast Asian Nations (ASEAN) is represented within the rating with Singapore’s 4 HPC methods within the November 2020 checklist. Whereas the TOP500 is just one method of taking the heartbeat of developments in HPC, Asia’s rising function is changing into manifest in quite a lot of nationwide and worldwide initiatives.
[ad_2]
Source link