Architecture |
Ampere |
GPU memory |
4 * 16GB GDDR6 |
Memory Bandwidth |
4 * 200GB/s |
Error-correcting code (ECC) |
Yes |
Ampere architecture-based CUDA Cores |
4 * 1280 |
NVIDIA 3rd Gen Tensor Cores |
4 * 40 |
NVIDIA 2nd Gen RT Cores |
4 * 10 |
FP32 || TF32 || Sparsity (TFLOPS) |
4x 4.5 || 4x 9 || 4x 18 |
FP16 || Sparsity (TFLOPS) |
4x 17.9 || 4x 35.9 |
INT8 || Sparsity (TOPS) |
4x 35.9 || 4x 71.8 |
System Interface |
PCIe Gen4 (x16) |
Maximum Power Consumption |
250W |
Thermal Solution |
Passive |
Form Factor |
Dual-slot, FH, FL |
Power Connector |
8-pin CPU |
Encode/Decode Engines |
4NVENC / 8NVDEC |
|
Includes AV1 decode |
Secure and Measured Boot with Hardware Root of Trust for GPU |
Yes (optional) |
vGPU Support |
NVIDIA Virtual PC (vPC) |
|
NVIDIA Virtual Applications (vApps) |
|
NVIDIA RTX Virtual Workstations (vWS) |
|
NVIDIA AI Enterprise |
|
NVIDIA Virtual Compute Server (vCS) |
Graphics APIs |
DirectX 12.07 (Hardware Feature Level 12 + 1) |
|
Shader Model 5.17 (Hardware Feature Level 12 + 1) |
|
OpenGL 4.68 (Based on published Khronos specification. Visit www.krhonos.org/conformance for further details.) |
|
Vulkan 1.18 ((Based on published Khronos specification. Visit www.krhonos.org/conformance for further details.) |
Compute APIs |
CUDA |
|
DirectCompute |
|
OpenCL™ |
|
OpenACC® |
MIG Support |
No |