Science & Tech · AI · 6 hrs ago
Nvidia unveils Rubin HPC GPU with more than 50,000 CUDA cores
Nvidia has unveiled Rubin HPC, a new graphics processor designed for high-performance computing.
The company announced it at the SC26 supercomputing conference as part of its Vera Rubin platform.
The chip has more than 50,000 CUDA cores, nearly 2.5 times the number in Nvidia’s RTX 5090.
It is designed for scientific simulations and data analysis, including work in climate modeling, genomic research and engineering.
Faster simulations could help researchers test more ideas and identify promising materials or designs sooner.
Nvidia has not disclosed key specifications, pricing or which systems will use the GPU.
Availability is expected in 2028, about two years after the standard Rubin generation is introduced.
Nvidia has announced Rubin HPC, a GPU with more than 50,000 CUDA cores on a single chip.
The GPU is designed for high-performance computing, scientific simulations and data analytics.
It adds seventh-generation NVLink connectivity and sparse data acceleration.
Nvidia expects Rubin HPC to be available in 2028.
The company has not disclosed its full specifications, pricing or the systems that will use it.
- Who
- Nvidia announced the Rubin HPC GPU.
- What
- It has more than 50,000 CUDA cores on a single chip and seventh-generation NVLink connectivity.
- When
- Nvidia announced it at the SC26 supercomputing conference. Availability is expected in 2028.
- Where
- The announcement was made at the SC26 supercomputing conference; the location is not stated.
- Why
- It is designed to accelerate high-performance computing, scientific simulations and data analytics.
This story does not have two clearly opposing sides.
Notebookcheck reported Nvidia’s announcement of Rubin HPC at the SC26 supercomputing conference.
Nvidia expects Rubin HPC to become available.
- CUDA cores
- More than 50,000 on a single chip
- Connectivity
- Seventh-generation NVLink
- Expected availability
- 2028
- Comparison
- Nearly 2.5 times the 21,760 CUDA cores in the RTX 5090
- Source of announcement
- SC26 supercomputing conference
What people are saying
For agents tackling science and engineering, faster simulation means faster discovery