h1.2xlarge vs trn1.2xlarge

Instance Typeh1.2xlargetrn1.2xlarge
Instance FamilyStorage optimizedMachine Learning ASIC
DetailsAmazon EMR workloads, distributed file systems (HDFS, MapR-FS), network file systems, log/data processing apps (like Apache Kafka), and big data clusters.

hHDD / hard disk throughput optimized
1 – Generation
2xlarge – Size
Machine Learning (ML) training

trn – AWS Trainium
1 – Generation
2xlarge – Size
vCPUs88
Memory32 GiB32 GiB
CPU Architecturex86_64x86_64
ProcessorIntel Xeon E5-2686 v4 (Broadwell)Intel Xeon Scalable (Icelake)
Clock Speed2.3 GHz3.5 GHz
Threads Per Core22
Accelerators
GPUnono
GPUs1
Inference Acceleratornono
Accelerator Memory32 GiB
FPGAnono
Instances
Sizesh1.16xlarge, h1.2xlarge, h1.4xlarge, h1.8xlargetrn1.2xlarge, trn1.32xlarge
SageMakerml.trn1.2xlarge
Free Tiernono
Burstablenono
Hibernationnono
EC2 Compute Units31
Instance Storage1 x 2000 HDD1 x 475 NVMe SSD
EBS (Elastic Block Store)
EBS Optimizeddefaultdefault
Encryptionsupportedsupported
NVMeunsupportedrequired
Dedicated Throughput1750 MBpsup to 20480 MBps
Optimized Baseline Bandwidth1750 Mbps5000 Mbps
Optimized Baseline Throughput218.75 Mbps625 Mbps
Optimized Baseline Performance12000 IOPS16250 IOPS
Optimized Max Performance12000 IOPS65000 IOPS
Optimized Max Bandwidth1750 Mbps20000 Mbps
Optimized Max Throughput218.75 Mbps2500 Mbps
Networking
Networking PerformanceUp to 10 Gigabit12500 Megabit
Enhanced Networkingyesyes
Max Interfaces44
Max Cards11
IPv4 Addresses per Interface1515
IPv6 Addresses per Interface1515
IPv6 Supportyesyes
ENA Supportsupportedrequired
EFA Supportnono
Encryption in Transit Supportnoyes
ENA SRD Supportnono
Avg Price Diff1.00x2.80x

Regional Prices

US Dollar ($)
Per Hour
Linux
Shared Tenancy
On Demand
Geography
Region
h1.2xlarge
trn1.2xlarge
HiddenHidden (hidden-1)0.5190n/a
USUS East (Ohio) (us-east-2)0.46801.3438
USUS East (Virginia) (us-east-1)0.46801.3438
USUS West (Oregon) (us-west-2)0.46801.3438
Activate subscription to see all regions and unlock all features