NVIDIA made quite the splash with its Turing GPU architecture a couple of months ago now, unveiling a family of Quadro RTX graphics cards with Turing GPUs as well as a new range of GeForce RTX graphics cards, with both families continuing to grow.
The company has just announced during the annual Autodesk University Conference its new Quadro RTX 4000 graphicws card that will join the ranks of the Quadro RTX 5000, RTX 6000, and flagship RTX 8000. The new Quadro RTX 4000 has quite a few highlights: there's 40% more memory bandwidth than the previous-gen Quadro P4000 card, with 36 RT cores for some real-time ray tracing content.
Inside, the Quadro RTX 4000 features Turing TU106 which has 2304 CUDA cores, 288 Tensor Cores, 36 RT cores, and 7.1 TFLOPs of FP32 performance. NVIDIA using 8GB of GDDR6 on the Quadro RTX 4000, pricing it at $900. The TDP is just 160W, 25W lower than the GeForce RTX 2070 which also packs the TU106 and 8GB GDDR6. This is because the GPU clocks on the Quadro RTX 4000 are lower than the GeForce RTX 2070, saving precious power consumption (and thus, less heat).
The big deal here is that the Quadro RTX 4000 is a single-slot card that has 3 x DP1.4 ports and 1 x VirtualLink port for great connectivity in a small footprint. NVIDIA will begin shipping its new Quadro RTX 4000 in December.