With 4x mini-DisplayPort 1.4a, the NVIDIA RTX 4000 SFF Ada Generation can power 4x 4K monitors at 120Hz, 4x 5K monitors at 60Hz, or 2x 8K monitors at 60Hz. The GPU is VR ready for real-time rendering and features ISV certification for a number of complex software applications.
This card features 20GB of GDDR6 ECC memory. Memory bandwidth for the NVIDIA RTX 4000 SFF Ada Generation is listed at 280GB/s with a 160-bit memory interface. For the larger NVIDIA RTX 4000 Ada, the memory interface is still 160-bit but memory bandwidth is at 360GB/s, 80GB/s faster than on the SFF version. Single-precision performance is listed at 19.2 TFLOPS, RT Cores performance at 44.3 TFLOPS, and Tensor Core performance at 306.8 TFLOPS, just slightly less than the full-sized version.
The NVIDIA 4000 SFF Ada has active cooling with an integrated fan. A 70W power draw is provided directly to the card through the PCIe 4.0 x8 slot on the system board and requires no additional power cables. A minimal host power supply unit of 250W is recommended.
| GPU Memory | 20GB GDDR6 |
| Memory Interface | 160 bit |
| Memory Bandwidth | 280GB/s |
| Error Correcting Code (ECC) | Yes |
| Ada Lovelace Based CUDA Cores | 6,144 |
| 4th Gen Tensor Core | 192 |
| 3rd Gen RT Cores | 48 |
| Single-Precision Performance | 19.2 TFLOPS |
| RT Core Performance | 44.3 TFLOPS |
| Tensor Performance | 306.8 TFLOPS |
| System Interface | PCIe 4.0 x16 |
| Power Consumption | Total board power: 70W |
| Thermal Solution | Active |
| Form Factor | 2.7” H x 6.6” L, dual slot |
| Display Connectors | 4x Mini DisplayPort 1.4a |
| Max Simultaneous Displays | 4x 4096 x 2160 @ 120Hz 4x 5120 x 2880 @ 60Hz 2x 7680 x 4320 @ 60Hz |
| Encode/Decode Engines | 2x encode, 2x decode (+AV1 encode and decode) |
| VR Ready | Yes |
| vGPU Software Support | No |
| Graphics APIs | DirectX 12, Shader Model 6.6, OpenGL 4.6, Vulkan 1.3 |
| Compute APIs | CUDA 11.6, OpenCL 3.0, DirectCompute |
| NVIDIA NVLink | No |