![New – EC2 P3dn GPU Instances with 100 Gbps Networking & Local NVMe Storage for Faster Machine Learning + P3 Price Reduction | AWS News Blog New – EC2 P3dn GPU Instances with 100 Gbps Networking & Local NVMe Storage for Faster Machine Learning + P3 Price Reduction | AWS News Blog](https://d2908q01vomqb2.cloudfront.net/da4b9237bacccdf19c0760cab7aec4a8359010b0/2018/11/11/p3dn_featured_1.png)
New – EC2 P3dn GPU Instances with 100 Gbps Networking & Local NVMe Storage for Faster Machine Learning + P3 Price Reduction | AWS News Blog
![Price-Performance Analysis of Amazon EC2 GPU Instance Types using NVIDIA's GPU optimized seismic code | AWS HPC Blog Price-Performance Analysis of Amazon EC2 GPU Instance Types using NVIDIA's GPU optimized seismic code | AWS HPC Blog](https://d2908q01vomqb2.cloudfront.net/e6c3dd630428fd54834172b8fd2735fed9416da4/2021/08/09/nvidia-esdk-fig3.png)
Price-Performance Analysis of Amazon EC2 GPU Instance Types using NVIDIA's GPU optimized seismic code | AWS HPC Blog
![Shashank Prasanna on Twitter: "You want the best perf/cost, single-GPU instance for training/inference: g5.xlarge Based on latest Ampere architecture, NVIDIA A10G GPU has 24GB memory. This option is for the most of Shashank Prasanna on Twitter: "You want the best perf/cost, single-GPU instance for training/inference: g5.xlarge Based on latest Ampere architecture, NVIDIA A10G GPU has 24GB memory. This option is for the most of](https://pbs.twimg.com/media/FL1G4X8VQAQ4bm3.jpg:large)