m9g.2xlarge vs trn1.2xlarge
| m9g.2xlarge | trn1.2xlarge |
| General purpose | Machine Learning ASIC |
| Small to medium size databases, data processing, caching, backends. m – Multi purpose 9 – Generation g – AWS Graviton processor 2xlarge – Size | Machine Learning (ML) training trn – AWS Trainium 1 – Generation 2xlarge – Size |
| 8 | 8 |
| 32 GiB | 32 GiB |
| arm64 | x86_64 |
| AWS Graviton5 Processor | Intel Xeon Scalable (Icelake) |
| 3.3 GHz | 3.5 GHz |
| 1 | 2 |
| no | no |
| 1 | |
| no | no |
| 32 GiB | |
| no | no |
| m9g.12xlarge, m9g.16xlarge, m9g.24xlarge, m9g.2xlarge, m9g.48xlarge, m9g.4xlarge, m9g.8xlarge, m9g.large, m9g.medium, m9g.metal-48xl, m9g.xlarge | trn1.2xlarge, trn1.32xlarge |
| m6g.2xlarge, m7g.2xlarge, m8g.2xlarge, m9g.2xlarge | |
| m9gd.2xlarge | |
| ml.trn1.2xlarge | |
| db.m9g.2xlarge | |
| no | no |
| no | no |
| yes | no |
| EBS only | 1 x 475 NVMe SSD |
| default | default |
| supported | supported |
| required | required |
| up to 12000 MBps | up to 20480 MBps |
| 3000 Mbps | 5000 Mbps |
| 375 Mbps | 625 Mbps |
| 12000 IOPS | 16250 IOPS |
| 48000 IOPS | 65000 IOPS |
| 12000 Mbps | 20000 Mbps |
| 1500 Mbps | 2500 Mbps |
| Up to 17000 Megabit | 12500 Megabit |
| yes | yes |
| 4 | 4 |
| 1 | 1 |
| 30 | 15 |
| 30 | 15 |
| yes | yes |
| required | required |
| no | no |
| yes | yes |
| no | no |
| 1.00x | 3.27x |
Regional Prices
Geography | Region | m9g.2xlarge | trn1.2xlarge |
|---|---|---|---|
| Hidden | Hidden (hidden-1) | 0.4689 | n/a |
| US | US East (Ohio) (us-east-2) | 0.3914 | 1.3438 |
| US | US East (Virginia) (us-east-1) | 0.3914 | 1.3438 |
| US | US West (Oregon) (us-west-2) | 0.3914 | 1.3438 |
Related Comparisons
Other instances worth weighing against these.