Amazon EC2 P3 Instances

The next generation of EC2 compute-optimized GPU instances became available on October 25, 2017. P3 instances are powered by up to 8 of the latest-generation NVIDIA Tesla V100 GPUs and are ideal for computationally advanced workloads such as machine learning (ML), high performance computing (HPC), data compression, and cryptography. They are also ideal for specific industry applications for scientific computing and simulations, financial analytics, and image and video processing.

P3 instances provide a powerful platform for ML and HPC by also leveraging 64 vCPUs using the custom Intel Xeon E5 processors, 488 GB of RAM, and up to 25 Gbps of aggregate network bandwidth leveraging Elastic Network Adapter technology.

Based on NVIDIA’s latest Volta architecture, each Tesla V100 GPUs provide 125 TFLOPS of mixed-precision performance, 15.7 TFLOPS of single precision (FP32) performance and 7.8 TFLOPS of double precision (FP64) performance. This is possible because each Tesla V100 GPUs contains 5,120 CUDA Cores and 640 Tensor Cores. A 300 GB/s NVLink hyper-mesh interconnect allows GPU-to-GPU communication at high speed and low latency.

For more on information on Tensor Cores check out:
http://www.tomshardware.com/news/nvidia-tensor-core-tesla-v100,34384.html

For ML applications, P3 instances offer up to 14x performance improvement over the existing P2 instances, allowing developers to train their machine learning models in hours instead of days, and bring their innovations to market faster. Check out the table below for the sizes:

They are available in US East (N. Virginia), US West (Oregon), EU West (Ireland) and Asia Pacific (Tokyo) regions. Customers can purchase P3 instances as On-Demand Instances, Reserved Instances, Spot Instances, and Dedicated Hosts.