GPU memory 4GB GDDR6
Memory interface 64-bit
Memory bandwidth 96GB/s
NVIDIA Ampere architecture-based
CUDA Cores
768
NVIDIA third-generation Tensor Cores 24
NVIDIA second-generation RT Cores 6
Single-precision performance 2.7 TFLOPS2
RT Core performance 5.4 TFLOPS2
FP16 Tensor performance 21.7 TFLOPS3
Peak INT8 Tensor performance 43.3 TOPS4
System Interface PCIe 4.0 x85
Power consumption Total board power: 50W
Thermal solution Active
Form factor 2.7” H x 6.4” L, single slot
Display connectors 4x Mini DisplayPort 1.4a
Max simultaneous displays 4x 4096 x 2160 @ 120 Hz
4x 5120 x 2880 @ 60 Hz
Encode/decode engines 1x encode, 1x decode (+AV1 decode)
Graphics APIs DirectX 12, Shader Model 6.6,
OpenGL 4.66, Vulkan 1.36
Compute APIs CUDA 11.6, OpenCL 3.0, DirectCompute