Description
Architecture
NVIDIA Ampere Architecture
Process Size
8nm
Transistors
13.25 Billion
Die Size
276 mm2
CUDA Cores
3328
Tensor Cores
104
RT Cores
26
Peak FP32 Performance
7.99 TFLOPS
RT Core Performance
15.6 TFLOPS
Tensor Core Performance
63.9 TFLOPS
GPU Memory
6 GB GDDR6 ECC
GPU Memory Clock
6001 MHz
Memory Interface
192-bit
Memory Bandwidth
288 GB/sec
Maximum Power Consumption
70 W
