NVIDIA Launches RTX A2000 Low-Profile Ampere Graphics Card For SFF Workstations
When we typically think of NVIDIA's RTX A-Series professional-oriented graphics cards, we imagine beefy cards that are thick and/or long. The RTX A2000, on the other hand, is a relatively small card that can fit into small-form-factor (SFF) workstations. As a result, NVIDIA equipped the GA106 with a low-profile cooler while adhering to a 2-slot card layout.
Given its entry-level status, the RTX A2000 is down considerably on power compared to the RTX A4000, the next step up the ladder in NVIDIA's Ampere workstation family. It features 3,328 CUDA cores and 104 Tensor cores with a maximum GPU boost clock of 1,200MHz. NVIDIA quotes single-precision performance of 8 TFLOPS (63.9 Tensor TFLOPS) compared to 19.2 TFLOPS (153.4 Tensor TFLOPS) for the RTX A4000.
The RTX A4000 also sports 6GB of 12Gbps GDDR6 memory operating on a 192-bit memory bus. This compares to 16GB 14Gbps GDDR6 (256-bit memory bus) for the RTX A4000. The RTX A2000's cut-down specs also mean that it sips power compared to its more performant sibling. It features a TDP of just 70 watts, which is half that of the RTX A4000.
NVIDIA switched to four Mini DisplayPort 1.4a ports (with latching mechanism) on the I/O bracket on the connectivity front. According to NVIDIA, the card can drive up to four 5K displays at 60Hz or four 4K displays at 120Hz.
"Introducing RT Cores into the NVIDIA RTX A2000 has resulted in impressive rendering speedups for photorealistic visualization compared to the previous generation GPUs," said Steven Blevins, director of Digital Practice at Cuhaci & Peterson. "The small form factor and low power usage of the NVIDIA RTX A2000 is extraordinary and ensures fitment in just about any existing workstation chassis," added Ken Grothman, virtual design and construction manager at Gilbane Building Company.
NVIDIA says that the RTX A2000 will ship in October priced at $450 and will ship in pre-built systems from ASUS, Lenovo, Hewlett-Packard, and Dell.