I’ve written about the power and value of GPUs in the past, and I have written posts to launch many generations of GPU-equipped EC2 instances including the CG1, G2, G3, P2, P3, and P3dn instance types.
Today I would like to give you a sneak peek at our newest GPU-equipped instance, the G4. Designed for machine learning training & inferencing, video transcoding, and other demanding applications, G4 instances will be available in multiple sizes and also in bare metal form. We are still fine-tuning the specs, but you can look forward to:
- AWS-custom Intel CPUs (4 to 96 vCPUs)
- 1 to 8 NVIDIA T4 Tensor Core GPUs
- Up to 384 GiB of memory
- Up to 1.8 TB of fast, local NVMe storage
- Up to 100 Gbps networking
The brand-new NVIDIA T4 GPUs feature 320 Turing Tensor cores, 2,560 CUDA cores, and 16 GB of memory. In addition to support for machine learning inferencing and video processing, the T4 includes RT Cores for real-time ray tracing and can provide up to 2x the graphics performance of the NVIDIA M60 (watch Ray Tracing in Games with NVIDIA RTX to learn more).
I’ll have a lot more to say about these powerful, high-end instances very soon, so stay tuned!
PS – If you are interested in joining a private preview, sign up now.
Originally posted on AWS News Blog
Author: Jeff Barr