NVIDIA unveils pro-level Ampere GPUs for workstations

NVIDIA isn't letting gamers have all the fun. Today at its GTC conference, the company announced its next round of professional-focused graphics cards, based on the Ampere architecture first scene with its RTX 30-series GPUs. For desktops, there's the RTX A5000 and A4000 GPUs, featuring 8,192 and 6,144 CUDA cores, respectively. Those GPUs top out with 24GB of GDDR6 memory, though you can also connect two cards together over NVLINK for up to 48GB of addressable memory. NVIDIA's RTX 3090 GPU technically offers even faster speeds, but its professional hardware is typically built for stability over lengthy workloads.

On the pro laptop front, there's the A2000 to A5000 range, all of which are equipped with the company's latest Max-Q optimization technology. It'll be interesting to see how well NVIDIA's highest end professional GPUs translate to notebooks; on the consumer front, the RTX 3080 isn't nearly as speedy as its desktop counterpart. NVIDIA is also releasing T1200 and T600 GPUs based on its previous Turing hardware, which are more focused on multitasking than heavy-duty graphics performance. NVIDIA also announced new A10 and A16 GPUs for datacenters, which should pair nicely with its recently announced Grace Arm-based CPU for servers and supercomputers.

You can expect to see NVIDIA's new pro-grade desktop GPUs in systems later this month, while laptop models should appear in the second quarter.