Nvidia professional users rejoice; the Quadro RTX 6000 has gone up for pre-order on Nvidia's website (opens in new tab) for $6,300 a piece. But you better hurry up, as stock is limited to five per customer.
Introduced to the hardware world last August, the Quadro RTX 6000 is Nvidia's mid-level Turing-based Quadro model for tackling professional workflows, such as content creation in the video and film industry, automotive and architectural design and scientific visualization. The Quadro RTX 6000 is built around the full-fledged TU102 silicon, taking up 754-square-millimeters of space. Its large die is produced inside TSMC’s 12nm FinFET oven and contains 18.6 billion transistors. Additionally, the Quadro RTX 6000 comes equipped with 4,608 CUDA cores, 576 Tensor cores for AI and 72 RT cores for ray tracing activities. The graphics card also has 24GB of GDDR6 memory across a 384-bit memory interface. Unfortunately, Nvidia didn't reveal details on the Quadro RTX 6000's core or boost clocks.
The Quadro RTX 6000 posts strong performance on paper. According to Nvidia, the graphics card pumps out 16.3 TFLOPS of single precision (FP32) and 32.6 TFLOPS of half precision (FP16) performance. It is also capable of delivering up to 10 GigaRays/sec of ray tracing performance, 84 Tera RTX-OPS and 130.5 Tensor TFLOPS.
On the aesthetics side, the Quadro RTX 6000 features a traditional blower design. It measures 11.176 x 26.67cm and occupies two PCI slots. The graphics card has a maximum power consumption of 295W and draws its power from a PCIe 3.0 x16 slot and a pair of 8-pin PCIe power connectors. The Quadro RTX 6000 sports Nvidia's NVLink connector. Therefore, consumers can pair two of these bad boys together with a Quadro RTX NVLink HB Bridge to scale memory capacity to 48GB and drive bandwidth up to 100GB/s.
As for display connectors, the Quadro RTX 6000 is equipped with four DisplayPort 1.4 outputs and a USB Type-C port for Nvidia VirtualLink. The graphics card is priced at $6,300, and interested consumers can pre-order it directly from Nvidia.
