Nvidia Unveils New Rubin HPC GPU for Scientific Computing
The chip powers the Vera Rubin platform for complex simulations by 2028
Nvidia announced the new Rubin HPC GPU during the SC26 event. The component is part of the Vera Rubin platform and was developed to optimize data processing in engineering and science.
The hardware features more than 50,000 CUDA cores. The technology includes the seventh generation of the NVLink interconnect system to accelerate communication between processing units.
The architecture prioritizes the balance between data movement and calculation capacity. The goal is to prevent cores from remaining idle while waiting for information from extensive datasets.
This design aims to meet the demands of climate models and genomic analyses. The company projects the product will reach the market by 2028.
The advancement in simulation capacity may reduce the time required for industrial discoveries. Efficiency in processing sparse matrices is one of the technical differentiators of the new unit.
Spotted an error? Write to suporte@criptoico.com.br.