Unlock the power of professional graphics with the NVIDIA RTX PRO 6000 Blackwell, 96GB, GDDR7, 512 Bits, PCIe Express Graphics Card. This workstation‑grade GPU delivers unmatched speed for demanding workloads, letting you finish projects faster and stay ahead of the competition.
NVIDIA RTX PRO 6000 Blackwell, 96GB, GDDR7, 512 Bits, PCIe Express Graphics Card
Equipped with the new Blackwell architecture, the card features 24,064 CUDA cores, 752 Tensor cores, and 188 fourth‑generation RT cores. Its 96 GB of ECC‑protected GDDR7 memory runs on a 512‑bit bus, providing up to 1.79 TB/s of bandwidth. The PCIe 5.0 ×16 interface ensures low latency and maximum throughput for data‑intensive tasks.
Boost clocks reach 2.617 GHz, while the base clock sits at 1.590 GHz, delivering smooth performance in 3D rendering, AI training, and high‑resolution video editing. Four DisplayPort 2.1b outputs support up to four simultaneous 8K displays, perfect for multi‑screen workstations.
Designed for reliability, the card includes 5th‑generation Tensor cores, 9th‑generation NVENC, 6th‑generation NVDEC, and full ECC error correction. Its dual‑slot, full‑height form factor fits standard workstations, and the 600 W TGP is powered by a single 16‑pin PCIe Gen5 connector.
Benefits
- Extreme memory capacity (96 GB) handles massive datasets without slowdown.
- High‑speed GDDR7 memory and 1.79 TB/s bandwidth accelerate rendering and simulation.
- ECC support guarantees data integrity for critical professional workloads.
- Four DisplayPort 2.1b outputs enable seamless multi‑monitor setups.
Uses of the product
- Professional 3D rendering and visual effects pipelines.
- AI model training and inference with large neural networks.
- Complex CAD/CAE simulations and scientific computing.
- High‑resolution video production and multi‑stream encoding.
Features
- Blackwell GPU architecture with 24,064 CUDA cores for unmatched compute power.
- 96 GB GDDR7 memory (ECC‑enabled) on a 512‑bit bus delivering 1.79 TB/s bandwidth.
- PCIe 5.0 ×16 interface and 16‑pin 12VHPWR connector for efficient power delivery.
- Workstation‑grade double‑flow cooling system to maintain optimal temperatures under load.
Specifications
| Product / Model |
NVIDIA RTX PRO 6000 Blackwell (Workstation Edition) |
| GPU Architecture |
NVIDIA Blackwell |
| GPU Processor / Chip |
GB202 (fully enabled) |
| CUDA Cores (Shading Units) |
24,064 |
| Texture Units (TMUs) |
752 |
| ROPs (Render Output Units) |
192 |
| Tensor Cores |
752 (5th-Generation) |
| Ray-Tracing Cores (RT Cores) |
188 (4th-Generation) |
| Base Clock |
1590 MHz |
| Boost Clock |
2617 MHz |
| Memory Size |
96 GB |
| Memory Type |
GDDR7 (with ECC) |
| Memory Bus Width |
512-bit |
| Transfer Rate |
28 Gbps (typical GDDR7 spec) |
| Memory Bandwidth (Theoretical) |
1.79 TB/s (1792 GB/s) |
| PCIe Interface |
PCI Express 5.0 ×16 |
| Display Ports |
4 × DisplayPort 2.1b |
| Max Simultaneous Displays |
Up to 4 displays |
| Video & Multimedia Engines |
4 × NVENC (9th generation), 4 × NVDEC (6th generation) |
| MIG |
Yes — supports partitioning (e.g. up to multiple 24 GB / 48 GB / full 96 GB instances) |
| Max Power Consumption (Board Power / TGP) |
~ 600 W |
| Power Connector |
1 × 16-pin PCIe Gen5 (12VHPWR / CEM5) |
| Form Factor |
Dual-slot design, full-height; ~ 12 in / 304 mm length, ~5.4 in height |
| Thermal Design |
Workstation-grade cooling, “double-flow through” (air/blower / passive-flow depending on variant) |
| Supported Graphics |
DirectX 12 Ultimate, OpenGL 4.x, Vulkan; CUDA, OpenCL, Vulkan Compute, etc. |
| Intended Use Cases |
Professional / workstation workloads: 3D rendering, CAD/CAE, AI training/inference, large-scale simulation, video production, multi-display / high-resolution workflows, multi-task heavy workloads |
| VRAM Error Correction |
ECC support (for reliability in professional workloads) |
