Architecture |
Ada Lovelace |
Memory Interface |
384 bit |
GPU memory |
48GB GDDR6 |
Memory Bandwidth |
960GB/s |
Error Correcting Code (ECC) |
Yes |
NVIDIA Ada Lovelace architecture-based CUDA Cores |
18,176 |
NVIDIA 4th Gen Tensor Cores |
568 |
NVIDIA 3rd RT Cores |
142 |
Single-Precision Performance |
91.1 TFLOPS Peak |
RT Core Performance |
210.6 TFLOPS Peak |
Tensor Perfomance |
1457 TFLOPS with sparsity |
System Interface |
PCIe 4.0 x 16 |
Power Consumption |
Total Board Power: 300W |
Thermal Solution |
Active |
Form Factor |
4.4" (H) x 10.5" (L) |
|
Dual-slot |
|
Full height |
Display Connectors |
4 * Mini DisplayPort 1.4a |
Maximum Simultaneous Displays |
4 * 4096 x 2160 @ 120Hz |
|
4 * 5120 x 2880 @ 60Hz |
|
2 * 7680 x 4320 @ 60Hz |
Power Connector |
1 * PCIe CEM5 16-pin |
Encode/Decode Engines |
3 * Encode |
|
3 * Decode |
|
+AV1 encode and decode |
VR Ready |
Yes |
vGPU Software Support |
NVIDIA vPC / vApps |
|
NVIDIA RTX Virtual Workstation |
vGPU Profiles Supported |
For supported profile information, visit: https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/solutions/resources/documents1/Virtual-GPU-Packaging-and-Licensing-Guide.pdf |
Graphics APIs |
DirectX 12 |
|
Shader Model 6.6 |
|
OpenGL 4.6 (Based on Khronos specification) |
|
Vulkan 1.3 (Based on Khronos specification) |
Compute APIs |
CUDA 11.6 |
|
OpenCL 3.0 |
|
DirectCompute |
NVIDIA NVLink® |
No |