VinAI Research, VinGroup’s AI research arm revealed, has claimed to be the first in Southeast Asia to deploy the newly-launched NVIDIA DGX A100 data center solution.
Vietnam’s first artificial intelligence research lab looks to use the supercomputer to train large AI models for language and videos by enabling more natural human interaction with machines through voices, gestures and biometrics.
VinAI will scale up AI research with NVIDIA’s DGX A100
Dr Bui Hai Hung, the Director of VinAI Research, said: “We have some exciting models and experiments waiting for the new machine.”
Around 70 scientists, residents and engineers will make full use of the new AI system that can reduce the time required to complete an experiment from more than one week to less than 24 hours.
He added: “Our lab’s computing-facility utilisation is always maxed out at 100% so there is no shortage of workload.”
The research lab contributed to the global fight against the COVID-19 pandemic by automatically analysing tweets for COVID-19 events and enabling face recognition of people donning face masks.
VinAI, founded in 2019, is prepared to scale up their research for applied AI projects, enabled by systems like NVIDIA’s DGX A100.
Dennis Ang, NVIDIA’s Director of Enterprise Business for the SEA and ANZ Region, said: “The new NVIDIA DGX A100 will empower VinAI to optimise computing power and resources to accelerate diverse workloads including data analytics, training and inference.”
The investment in AI is part of VinGroup’s long-term strategy of becoming a technology-focused corporation.
The NVIDIA DGX A100 was revealed in May 2020 as the world’s first five-petaflops server that can be divided into as many as 56 applications running independently. The system includes NVIDIA’s new A100 GPU and will cost around US$199,000.
Jensen Huang, NVIDIA’s Founder and CEO, said in a keynote recorded from the kitchen of his California home that the system allows a single server to either “scale up” to race through computationally intensive tasks such as AI training, or “scale out,” for AI deployment.
NVIDIA estimated that a data center powered by five DGX A100s working on AI training and inference can do the work of a typical data center with 50 DGX-1 systems and 600 CPU systems consuming 630 kilowatts and costing over $11 million.