Nvidia has finally launched the RTX A6000, the most powerful graphics card based on the Ampere GPU. Its single-precision performance is 38.7 TFLOPS – 3.1 TFLOPS more than the GeForce RTX 3090. This 3D accelerator is built on the full-featured GA102 GPU with 10,752 CUDA cores. The memory capacity is 48 GB, but this is GDDR6, not GDDR6X. Therefore, in terms of memory bandwidth, the RTX A6000 is even inferior to the RTX 3090 – 768 GB / s versus 936 GB / s.
The RTX A6000 has four DisplayPort 1.4 ports and does not have an HDMI 2.1 connector. The cooling system uses only one active element – centrifugal type. NVLink allows you to combine two video cards.
And now the most interesting thing is the price. Surprisingly, the RTX A6000 is even cheaper than its predecessor, the Quadro RTX 6000, based on the Turing TU102 GPU and provides lower performance. If for the Quadro RTX 6000, they ask $ 6,300, then the RTX A6000 is $ 4,650.
True, the official price may not coincide with the real one. Thus, PNY Technologies NVIDIA RTX A6000 with a $ 5500 is already presented in the catalog of the American retailer B&H. Sales are promised within 2-4 weeks.