◆This graphics card is composed of two Kepler GK210 GPUs, each graphics processing unit can support up to 12GB GDDR5 flash memory (total 24GB), in addition, Tesla K80 also supports 480GB/s bandwidth.

◆The target areas of Tesla K80 include data analysis and scientific computing. Its single-precision computing performance can reach up to 9 trillion floating-point operations per second, while the consumer-grade GeForce GTX 980 can hardly exceed 5 trillion per second.

◆This graphics card is composed of two Kepler GK210 GPUs, and each graphics processing unit can support up to 12GB of GDDR5 flash memory (24GB in total). In addition, Tesla K80 also supports 480GB/s bandwidth, 4992 CUDA parallel processing cores, custom technologies such as Dynamic Nvidia GPU Boost, and dynamic parallelism (Dynamic Parallelism). According to Nvidia, these configurations allow the K80 to significantly surpass the best hardware available. They also claim that Tesla K80 is 10 times faster than the best CPU.

◆Compared with the predecessor K40, Tesla K80 has a significant improvement. The running scores also show that the K80 also leads the way, whether it is chemistry, physics or machine learning.