p2.16xlarge vs trn1n.32xlarge
| p2.16xlarge | trn1n.32xlarge |
| GPU instance | Machine Learning ASIC |
| GPU intensive tasks like machine/deep learning, high-performance computing p – Performance / GPU (Graphics Processing Unit) accelerated 2 – Generation 16xlarge – Size | Machine Learning (ML) training trn – AWS Trainium 1 – Generation n – Network and EBS optimized 32xlarge – Size |
| 64 | 128 |
| 732 GiB | 512 GiB |
| n/a | x86_64 |
| Intel Xeon E5-2686 v4 (Broadwell) | Intel Xeon Scalable (Icelake) |
| 3.5 GHz | |
| 0 | 2 |
| no | no |
| 16 | 16 |
| no | no |
| 512 GiB | |
| no | no |
| p2.16xlarge, p2.8xlarge, p2.xlarge | trn1n.32xlarge |
| p2.16xlarge, p3.16xlarge | |
| trn1.32xlarge | |
| ml.trn1n.32xlarge | |
| no | no |
| no | no |
| no | no |
| 201 | |
| EBS only | 4 x 1900 NVMe SSD |
| default | |
| supported | |
| required | |
| 10000 MBps | 81920 MBps |
| 80000 Mbps | |
| 10000 Mbps | |
| 260000 IOPS | |
| 260000 IOPS | |
| 80000 Mbps | |
| 10000 Mbps | |
| 20 Gigabit | 1600 Gigabit |
| yes | yes |
| 80 | |
| 16 | |
| 50 | |
| 50 | |
| no | yes |
| required | |
| no | yes |
| 16 | |
| no | yes |
| no | no |
| 1.00x | 1.21x |
Regional Prices
Geography | Region | p2.16xlarge | trn1n.32xlarge |
|---|---|---|---|
| Hidden | Hidden (hidden-1) | 15.5520 | n/a |
| Hidden | Hidden (hidden-2) | 17.2800 | n/a |
| Hidden | Hidden (hidden-3) | 21.2160 | n/a |
| Hidden | Hidden (hidden-4) | 23.4400 | n/a |
| Hidden | Hidden (hidden-5) | 24.6720 | n/a |
| Hidden | Hidden (hidden-6) | 24.6720 | n/a |
| Hidden | Hidden (hidden-7) | 27.4880 | n/a |
| Hidden | Hidden (hidden-8) | 27.4880 | n/a |
| US | US East (Ohio) (us-east-2) | 14.4000 | 24.7800 |
| US | US East (Virginia) (us-east-1) | 14.4000 | 24.7800 |
| US | US West (Oregon) (us-west-2) | 14.4000 | 24.7800 |
Activate subscription to see all regions and unlock all features
Related Comparisons
Other instances worth weighing against these.