Graphic Processing Unit or GPU as we know it for several years now, has gained a new importance in the light of Artificial Intelligence. GPU is a specialized circuit designed to manipulate memory rapidly to create faster images.
GPUs are highly efficient in manipulating computer graphics and image processing. However they have now gained importance due to their efficiency in fast computing in a parallel fashion. GPU has parallel architecture consisting of several thousand smaller yet efficient cores designed to handle multiple tasks simultaneously. This is where they differ from CPU, which has only a few cores designed for sequential processing.
GPU in Artificial Intelligence
GPUs have been found to be tremendously powerful as compared to CPUs. In one of the project, 12 NVIDIA GPUs delivered deep-learning performance of 2000 CPUs. That is phenomenal! NVIDIA GPUs are speeding up the DNNs (Deep neural Networks) by 10-20x, resulting in reduction in the training times for the Artificial Intelligence. NVIDIA has also provided rich platform for developers (CUDA) which improved developers’ productivity helping them innovate quickly.
Other Uses of GPU
We had known GPU long only for their graphics related use such as gaming. Several gaming consoles were powered by GPUs. However, as explained above GPUs are now very popular in the field of Artificial Intelligence. They have also been extremely useful and popular in several other areas such as:
- Self Driving cars – to train the algorithm to detect the vehicles even in difficult conditions
- Healthcare and Life Sciences – deep genomics studies
It is evident that the parallel processing that GPUs offer are going to dominate the near future and can be seen from the investor interest in this field. In last year itself there have been several investments from key VCs in the area of hardware.