trn2.48xlarge vs g7e.48xlarge
| trn2.48xlarge | g7e.48xlarge |
| Machine Learning ASIC | GPU instance |
| Machine Learning (ML) training trn – AWS Trainium 2 – Generation 48xlarge – Size | Machine Learning (ML), 3D visualizations, graphics-heavy remote workstations, 3D rendering, application streaming, video encoding, and server-side graphics tasks. g – GPU (Graphics Processing Unit) 7 – Generation e – Extra storage or memory 48xlarge – Size |
| 192 | 192 |
| 2048 GiB | 2048 GiB |
| x86_64 | x86_64 |
| Intel Xeon Scalable (Sapphire Rapids) | Intel Xeon Scalable (Emerald Rapids) |
| 3.5 GHz | 2.4 GHz |
| 2 | 2 |
| no | yes |
| RTX PRO Server 6000 | |
| NVIDIA | |
| 16 | 8 |
| 96 GiB | |
| 768 GiB | |
| no | no |
| 15363 GiB | 768 GiB |
| no | no |
| trn2.48xlarge | g7e.12xlarge, g7e.24xlarge, g7e.2xlarge, g7e.48xlarge, g7e.4xlarge, g7e.8xlarge |
| g6e.48xlarge, g7e.48xlarge | |
| ml.trn2.48xlarge | ml.g7e.48xlarge |
| no | no |
| no | no |
| no | no |
| EBS only | 4 x 3800 NVMe SSD |
| default | default |
| supported | supported |
| required | required |
| 81920 MBps | 100000 MBps |
| 80000 Mbps | 100000 Mbps |
| 10000 Mbps | 12500 Mbps |
| 260000 IOPS | 400000 IOPS |
| 260000 IOPS | 400000 IOPS |
| 80000 Mbps | 100000 Mbps |
| 10000 Mbps | 12500 Mbps |
| 3200 Gigabit | 1600 Gigabit |
| yes | yes |
| 32 | 40 |
| 16 | 4 |
| 50 | 64 |
| 50 | 64 |
| yes | yes |
| required | required |
| yes | yes |
| 16 | 4 |
| yes | yes |
| no | yes |
| Infinityx |
Regional Prices
Geography | Region | trn2.48xlarge | g7e.48xlarge |
|---|---|---|---|
| Hidden | Hidden (hidden-1) | n/a | 48.0691 |
| US | US East (Ohio) (us-east-2) | n/a | 33.1443 |
| US | US East (Virginia) (us-east-1) | n/a | 33.1443 |
| US | US West (Oregon) (us-west-2) | n/a | 33.1443 |