GPU architecture
Turing
CUDA parallel processing core
4608
NVIDIA Tensor core
576
NVIDIA RT CORE
72
memory
24GB GDDR6
RTX-OPS
80t
ray projection
10 Giga Rays/Sec
highest single precision (FP32) performance
14.9 TFLOPS
maximum half precision (FP16) performance
29.9 TFLOPS
highest integer arithmetic (INT8) performance
238.9 TOPS
deep learning TeraFLOPS1
119.4 Tensor TFLOPS
memory Interface
384-bit
memory Bandwidth
624 GB/s
maximum power consumption
250W
bus
PCI Express 3.0 x16
plate type
4.4 "high x 10.5" long double slot
weight
1200g
heat dissipation scheme
passive cooling seat
NVLink interconnection technology
100 GB/s
