Compute Comparison
vs
All comparisons →

Ampere

NVIDIA A10G

wins

3

The NVIDIA A10G is the AWS-specific variant of the A10, available on g5 instances.

AWS inferenceGraphics workloadsMulti-tenant AI

Ampere

NVIDIA RTX A5000 24GB

wins

1

The RTX A5000 is a professional Ampere GPU with 24 GB GDDR6 and NVLink 3.0 support.

Professional inferenceBudget NVLink multi-GPU13B–30B model serving

Performance comparison

Visual bars — winner highlighted

NVIDIA A10G

Metric

NVIDIA RTX A5000 24GB

FP16 Tensor TFLOPS (dense)

125 T
56 T

Bandwidth

600 GB/s
768 GB/s

VRAM

24 GB
24 GB

TDP (lower=better)

150 W
230 W

TFLOPS/watt

0.83
0.24

GB/s per watt

4.00
3.34

Full specification table

SpecNVIDIA A10GNVIDIA RTX A5000 24GB
ArchitectureAmpereAmpere
VRAM24 GB GDDR624 GB GDDR6
VRAM typeGDDR6GDDR6
Memory bandwidth600 GB/s768 GB/s
FP16 Tensor TFLOPS (dense)125 TFLOPS55.6 TFLOPS
TDP150 W230 W
TFLOPS/watt0.83 T/W0.24 T/W
GB/s per watt4.00 GB/s/W3.34 GB/s/W
NVLinkNoYes
Release year20212021

Live cloud pricing

On-demand pricing across providers — updated in real time

Per hourPer GB VRAM/hr
Loading prices…

Power efficiency analysis

TFLOPS/watt and GB/s/watt — critical for data center TCO

NVIDIA A10G

FP16 TFLOPS125 TFLOPS
TDP150 W
TFLOPS/watt0.833 T/W
GB/s per watt4.00 GB/s/W

NVIDIA RTX A5000 24GB

FP16 TFLOPS55.6 TFLOPS
TDP230 W
TFLOPS/watt0.242 T/W
GB/s per watt3.34 GB/s/W

TFLOPS/watt measures compute efficiency — how much AI throughput you get per watt of power consumed. For data centers with PUE of 1.2–1.5, a 10% improvement in TFLOPS/watt translates directly to lower electricity costs and cooling requirements. GB/s/watt measures memory bandwidth efficiency, which is the binding constraint for memory-bound LLM inference workloads.

When to choose each GPU

Choose NVIDIA A10G for:

  • AWS inference
  • Graphics workloads
  • Multi-tenant AI

Weaknesses / Limitations

  • AWS-specific — not available on other cloud platforms
  • 24 GB VRAM ceiling limits model size
  • Older Ampere architecture — lower TFLOPS than Ada equivalents

Choose NVIDIA RTX A5000 24GB for:

  • Professional inference
  • Budget NVLink multi-GPU
  • 13B–30B model serving

Weaknesses / Limitations

  • GDDR6 memory bandwidth far below HBM alternatives
  • Older Ampere architecture — lower TFLOPS than Ada equivalents
  • Professional but older architecture — no FP8 support