Flopper.io
Flopper.io GPU Comparison Sheet

AWS Inferentia vs AWS Trainium3

Side by side datacenter GPU specifications from the Flopper.io database.

Overview

AWS InferentiaAWS Trainium3
FP32 TFLOPS—183.0
Memory8 GB144 GB
Memory Bandwidth50 GB/s4.9 TB/s
TDP——

Performance by Precision

PrecisionAWS Inferentia peak TFLOPS (dense)AWS Trainium3 peak TFLOPS (dense)
FP64 ——
FP32 —183.0
TF32 —671.0
BF16 64.0671.0
FP16 64.0671.0
FP8 —2517.0
FP6 ——
FP4 —2517.0
INT8 128.0—

Specifications

AWS InferentiaAWS Trainium3
ArchitectureNeuronCore-v1NeuronCore-v4
Process Node—3nm
Launch Year20182025
Form Factor——
Memory TypeDDR4HBM3e
Interface Width——
InterconnectNeuronLink-v1NeuronLink-v4
TDP——
FP32 Efficiency——