p2.8xlarge vs p3.16xlarge

Instance Typep2.8xlargep3.16xlarge
Instance FamilyGPU instanceGPU instance
DetailsGPU intensive tasks like machine/deep learning, high-performance computing

p – Performance / GPU (Graphics Processing Unit) accelerated
2 – Generation
8xlarge – Size
GPU intensive tasks like machine/deep learning, high-performance computing

p – Performance / GPU (Graphics Processing Unit) accelerated
3 – Generation
16xlarge – Size
vCPUs3264
Memory488 GiB488 GiB
CPU Architecturen/an/a
ProcessorIntel Xeon E5-2686 v4 (Broadwell)Intel Xeon E5-2686 v4 (Broadwell)
Threads Per Core00
Accelerators
GPUnono
GPUs88
Inference Acceleratornono
Accelerator Memory128 GiB
FPGAnono
Instances
Sizesp2.16xlarge, p2.8xlarge, p2.xlargep3.16xlarge, p3.2xlarge, p3.8xlarge
Generationsp2.8xlarge, p3.8xlargep2.16xlarge, p3.16xlarge
Free Tiernono
Burstablenono
Hibernationnono
EC2 Compute Units97201
Instance StorageEBS onlyEBS only
EBS (Elastic Block Store)
Dedicated Throughput5000 MBps14000 MBps
Networking
Networking Performance10 Gigabit25 Gigabit
Enhanced Networkingyesyes
IPv6 Supportnono
EFA Supportnono
Encryption in Transit Supportnono
ENA SRD Supportnono
Avg Price Diff1.00x2.86x

Regional Prices

US Dollar ($)
Per Hour
Linux
Shared Tenancy
On Demand
Geography
Region
p2.8xlarge
p3.16xlarge
HiddenHidden (hidden-1)7.776026.4400
HiddenHidden (hidden-2)8.640026.9280
HiddenHidden (hidden-3)10.608028.7120
HiddenHidden (hidden-4)11.720029.3760
HiddenHidden (hidden-5)12.336030.5840
HiddenHidden (hidden-6)12.336033.5520
HiddenHidden (hidden-7)13.744033.8720
HiddenHidden (hidden-8)13.744033.8720
HiddenHidden (hidden-9)n/a33.8720
USUS East (Ohio) (us-east-2)7.200024.4800
USUS East (Virginia) (us-east-1)7.200024.4800
USUS West (Oregon) (us-west-2)7.200024.4800
Activate subscription to see all regions and unlock all features  

Related Comparisons

Other instances worth weighing against these.