NVIDIA DGX A100 AI is the world's largest GPU, priced at USD 199K
NVIDIA recently revealed the DGX A100 AI built using the next-generation Ampere GPU architecture.
 |
| NVIDIA DGX A100 AI |
A massive and ultra-powerful GPU system
This GPU system was built for scientific computing, cloud graphics, data analytics, and more. It is said to offer unprecedented compute density, performance, and flexibility as it is the world's first 5 petaFLOPS AI system featuring the NVIDIA A100 Tensor Core GPU.
 |
| Add caption |
NVIDIA added that it should enable enterprises to consolidate training, inference, and analytics into a unified, easy-to-deploy AI infrastructure that includes direct access to NVIDIA AI experts.
The DGX system consists of 8 A100s combined using NVIDIA's third-generation NVLink. 32GB of GPU memory, and 12.4TB of memory bandwidth. It also comes with 15TB of Gen4 NVMe internal storage.
It has a total of 54 billion transistors with 5 petaFLOPS of performance. It is also powerful enough to split workloads up to 56 instances.
Compared with other systems, NVIDIA boasts that it has the ability of an entire data center on a single rack.
In comparison, NVIDIA said that to handle AI training tasks, you'll need 600 central processing unit (CPU) systems to handle millions of queries for datacenter applications. It'll require a total of 25 racks of servers and 630 kilowatts of power. The total cost is USD 11 million.
With this new NVIDIA technology, you can do the same amount of processing for just USD 1 million in a single rack with 28 kilowatts of power.
In short, it should be more cost-efficient and powerful than today's more expensive systems.
Some of the first applications of the new system include the COVID-19 research at the US Argonne National Laboratory.
The NVIDIA DGX A100 AI starts at USD 199K (around PHP 10M).
Post a Comment