Amazon had only recently introduced EC2 instance that could compute high capacity workloads. Now, almost within a week’s time, Amazon has released its P3, the latest GPU optimized instance. This new technology is designed for compute-intensive and advanced workloads. It is available in 3 different configurations and is used in financial analytics, cryptography, machine learning and anything else you can put your finger on.
How good is the new P3?
The processing power is 14 times faster than the earlier instance P2. P3 runs on Nvidia’s new Tesla V100 GPUs, which are based on the chip maker’s Volta technology. According to Nvidia, each Volta GPU has the performance capability of 100 CPUs, making it especially suited for complex AI and deep learning workloads. Jeff Barr has quoted that “Each of the NVIDIA GPUs is packed with 5,120 CUDA cores and another 640 Tensor cores and can deliver up to 125 TFLOPS of mixed-precision floating point, 15.7 TFLOPS of single-precision floating point, and 7.8 TFLOPS of double-precision floating point. On the two larger sizes, the GPUs are connected together via NVIDIA NVLink 2.0 running at a total data rate of up to 300 GBps. This allows the GPUs to exchange intermediate results and other data at high speed, without having to move it through the CPU or the PCI-Express fabric.”
Powerful performance & scalability are the two aspects that are making P3 so special. Scaling up & down with Amazon EC2 ENA- based enhanced networking supports up to 25gbps network bandwidth, it turns out to be very helpful in distributed deep learning frameworks such as MXNet, it scales up with near perfect efficiency. The P3 instances also feature up to 64 vCPUs based on custom Intel Xeon E5 (Broadwell) processors and 488GB of DRAM.
The benefits of P3 instance is simple and straightforward. The three main benefits you’ll encounter are –
- Low cost
The NVIDIA Tesla V100 GPUs have tremendously impacted the processing time. Machine learning, HPC workloads & other floating-point sensitive workloads are done in a matter of minutes or let’s say it’s done in a jiffy!
With P3 instances you can now deploy GPUs in matter of minutes. The Pay-as-you-go model is what is bringing the agility in P3. Engineers are able to introduce new innovations to the market faster while optimizing their total operational costs.
The EC2 pricing models such as on-demand, spot, and reserved instances are more cost-effective when compared to building out on premise GPU compute clusters. P3 instances are fundamentally disrupting how organizations typically consume hardware for AI, HPC, machine learning and high performance computing.
Customers can launch P3 instances using the AWS Management Console, AWS Command Line Interface (CLI), and AWS SDKs. Amazon EC2 P3 Instances are generally available in the US East (N. Virginia), US West (Oregon), EU West (Ireland), and Asia Pacific (Tokyo) regions with support for additional regions coming soon. They are available in three sizes, with one, four, and eight GPUs, and can be purchased On-demand, Reserved or Spot instances.
Source: For more details, please visit the below URLs