The NVIDIA RTX PRO 6000 Blackwell Server Edition is the top-end model in this generation of professional video cards from NVIDIA. As the name suggests, it uses their new Blackwell architecture - the same tech underlying the GeForce RTX 50 Series. However, it has configurable power consumption which can be tailored for the available cooling and power supply capacity in a given system. At the top end, it can rival the GeForce RTX 5090's performance but with a focus on support for professional applications and three times the onboard VRAM (96GB vs 32 on the 5090). At reduced power, it would have lower clock speeds and overall performance - but will still be an extremely powerful GPU!
Because it is designed for use exclusively in a server environment, this variant of the RTX PRO 6000 Blackwell has a passive heatsink (no fan) and the DisplayPort connectors that are present are disabled (so there is no video output). This means that it is wholly dependent on cooling provided by the fans in the rackmount chassis in which it is installed. For systems where sufficient cooling is not available, there are other models with built-in fans: the single-card Workstation Edition and the multi-GPU friendly Max-Q Workstation Edition.
Because it is designed for use exclusively in a server environment, this variant of the RTX PRO 6000 Blackwell has a passive heatsink (no fan) and the DisplayPort connectors that are present are disabled (so there is no video output). This means that it is wholly dependent on cooling provided by the fans in the rackmount chassis in which it is installed. For systems where sufficient cooling is not available, there are other models with built-in fans: the single-card Workstation Edition and the multi-GPU friendly Max-Q Workstation Edition.
Note: This GPU does have four DisplayPort outputs, but they are not enabled from the factory as Nvidia considers this a compute/data center-class card.
Model: NVIDIA NVRTXPRO6000TCGPU-KIT
Specifications
| Chipset Manufacturer | NVIDIA |
| Product Category | Compute |
| Motherboard Connection | PCI Express 5.0 x16 |
| Cooling Method | Passive Heatsink |
Core Specifications | |
| Processors | 24,064 CUDA Cores |
Memory Specifications | |
| Onboard Memory | 96GB |
| Memory Type | GDDR7 ECC |
| Memory Bus Width | 512-bit |
| Bandwidth | 1597 GB/s |
Performance | |
| Single Precision Floating Point (Peak) | 120 Tflops |
Power Connectors | |
| Plug 1 | 12V-2x6 |
Dimensions | |
| Length | 267 mm (10.5 in) |
| Height | 112 mm (4.4 in) |
| Width | 40 mm (1.6 in) |
| Net Weight | 1.219 kg (2.7 lbs) |
