NVIDIA
NVIDIA RTX PRO 6000 Blackwell 96 GB GDDR7 900-5G153-2200-000 workstation graphics card
Delivers 3511 AI TOPS with 96 GB GDDR7 ECC memory on a 512-bit bus for multi-GPU workstation scaling
- InterfacePCI Express 5.0 x16
- Chipset ManufacturerNVIDIA
- GPURTX PRO 6000
- CUDA Cores24064
- Memory Size96GB
- Memory Interface512-bit
- 96 GB GDDR7 ECC memory enables large-model training and high-resolution rendering
- 24,064 CUDA cores deliver 110 TFLOPS single-precision compute for simulation workloads
- 5th-generation Tensor cores provide 3,511 AI TOPS with FP4 precision for inference acceleration
- PCI Express 5.0 x16 interface and 512-bit bus maximise data throughput for multi-GPU scaling
- 600 W power draw in a dual-slot Max-Q design fits dense workstation configurations
Workstation GPU with 96 GB GDDR7
The NVIDIA RTX PRO 6000 Blackwell is a workstation graphics card built for AI, data science and high-end creative workloads. It provides 96 GB of GDDR7 memory with ECC and a 512-bit interface. The card connects via PCI Express 5.0 x16 and occupies two expansion slots.
Thermals and 600 W power draw
Rated at 600 W, the card requires a single 16-pin auxiliary power connector. The Max-Q design targets density-optimised systems where up to four GPUs can be installed together. Active cooling handles the thermal load in normal operation.
Dual-slot fit and multi-GPU scaling
The dual-slot form factor fits standard workstation chassis without extra width. Four DisplayPort outputs drive multiple displays for visualisation tasks. Scalable architecture allows adding up to three more cards to increase compute capacity.
Points forts
- 96 GB GDDR7 ECC memory enables large-model training and high-resolution rendering
- 24,064 CUDA cores deliver 110 TFLOPS single-precision compute for simulation workloads
- 5th-generation Tensor cores provide 3,511 AI TOPS with FP4 precision for inference acceleration
- PCI Express 5.0 x16 interface and 512-bit bus maximise data throughput for multi-GPU scaling
- 600 W power draw in a dual-slot Max-Q design fits dense workstation configurations
Caractéristiques
| Brand | NVIDIA |
|---|---|
| Series | Blackwell |
| Model | 900-5G153-2200-000 |
| Part Number | 900-5G153-2200-000 |
| Interface | PCI Express 5.0 x16 |
| Chipset Manufacturer | NVIDIA |
| GPU | RTX PRO 6000 |
| CUDA Cores | 24064 |
| Memory Size | 96GB |
| Memory Interface | 512-bit |
| Memory Type | GDDR7 with ECC |
| DirectX | DirectX 12 |
| OpenGL | OpenGL 4.6 |
| DisplayPort | 4 |
| AI Features | AI TOPS: 3511 Tensor Cores: 5th Gen Ray Tracing Cores: 4th Gen NVIDIA Encoder (NVENC): 4x 9th Gen NVIDIA Decoder (NVDEC): 4x 6th Gen CUDA 12.8 OpenCL 3.0 DirectCompute |
| Operating Systems Supported | Windows, Linux |
| Auxiliary Power Connector | 1x 16-pin |
| Power Consumption | 600W |
| Slot Width | Dual Slot |
| Poids d’expédition | 1.36 kg |
| Dimensions du colis | 324 × 163 × 60 mm |
Questions sur cet article
How do I install the 16-pin power connector on this dual-slot RTX PRO 6000 and what is the common mistake?
Connect the single 16-pin auxiliary power cable fully until it clicks; the common mistake is not seating it completely, which can cause instability under the 600 W power draw.
What maintenance will the 96 GB GDDR7 ECC memory and active cooling need over time?
Clean dust from the dual-slot heatsink and fans periodically; the GDDR7 memory with ECC requires no user service, and the card has no user-replaceable parts.
What limit appears first when the 24064 CUDA cores run sustained AI workloads at 3511 TOPS?
The 600 W power ceiling and dual-slot thermal capacity are the first limits reached during prolonged maximum compute loads.
Également dans ce rayon
Les clients ont aussi comparé
Même rayon, même caisse — huit monnaies et un taux bloqué 30 minutes.
Silverstone Decathlon R Gold 750W ATX 3.0 power supply
Silverstone Technology FA311-B FARA 311 Micro-ATX PC case
- Micro ATX



