GPU memory 12GB GDDR6
Memory interface 192-bit
Memory bandwidth 288 GB/s
Error-correcting code (ECC) Yes
NVIDIA Ampere architecturebased CUDA Cores 3,328
NVIDIA third-generation Tensor Cores 104
NVIDIA second-generation RT Cores 26
Single-precision performance 8.0 TFLOPS
RT Core performance 15.6 TFLOPS
Tensor performance 63.9 TFLOPS
System interface PCI Express 4.0 x16
Power consumption Total board power: 70 W
Thermal solution Active
Form factor 2.7” H x 6.6” L, dual slot
Display connectors 4x mDP 1.4a with latching mechanism
Max simultaneous displays: 4x 4096 x 2160 @ 120Hz, 4x 5120 x 2880 @ 60Hz, 2x 7680 x 4320 @ 60Hz
Encode/decode engines 1x encode, 1x decode, (+AV1 decode)
VR ready Yes
Graphics APIs DirectX 12.07, Shader Model 5.17, OpenGL 4.685, Vulkan 1.2
Compute APIs CUDA, DirectCompute, OpenCL