Mixed-Precision Programming with CUDA 8

Originally published at: https://developer.nvidia.com/blog/mixed-precision-programming-cuda-8/

Update, March 25, 2019: The latest Volta and Turing GPUs now incoporate Tensor Cores, which accelerate certain types of FP16 matrix math. This enables faster and easier mixed-precision computation within popular AI frameworks. Making use of Tensor Cores requires using CUDA 9 or later. NVIDIA has also added automatic mixed precision capabilities to TensorFlow, PyTorch, and MXNet. Interested in learning more…

1. cuDNN 6 will add support for INT8 inference convolutions. Does the INT8 converlution will use DP4A?
2.When cudnn 6 can be download?
3.Can I get some exsamples in github about the "INT8 convolution"?