NVIDIA announces workstation-class Ampere architecture GPU


NVIDIA Ampere workstation GPU

NVIDIA Ampere workstation GPU

NVIDIA today’s GTC is not only ARM-related news, but also released a series of professional workstation-level graphics cards, following the RTX A6000 released in October last year, using the same Ampere architecture as the RTX 30 series of consumer graphics cards. On the desktop side, there are 8,192 CUDA core RTX A5000 and 6,144 CUDA core RTX A4000 graphics cards, both of which can be equipped with up to 24GB GDDR6 display, but you can also connect a pair of graphics cards through NVLINK to connect it with 48GB Available memory. In the notebook part, there are four GPUs from RTX A2000 to A5000, and they are all equipped with Max-Q technology, but this also makes their actual performance more difficult to compare. Influence it.

In addition, NVIDIA has also launched two notebook GPUs, T1200 and T600, based on the Turing architecture. They are mainly focused on multi-tasking, and NVIDIA claims that they are a significant performance upgrade compared to the internal GPU. Finally, NVIDIA also introduced new A10 and A16 GPUs for the data center. This is the display chip mentioned in Grace’s article and will be brought to Amazon’s AWS.

The RTX graphics card for desktop computers will be available earlier this month, while the notebook version will have to wait until the second season.

Posted on