Compute Comparison
vs
All comparisons →

CDNA 3.5

AMD Instinct MI325X 256GB

wins

4

The AMD Instinct MI325X is an upgraded MI300X with 256 GB HBM3e — the highest memory capacity of any single GPU.

Massive context LLM inferenceMemory-bound workloadsLarge open-source models

Ampere

NVIDIA A10G

wins

1

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

AWS inferenceGraphics workloadsMulti-tenant AI

Performance comparison

Visual bars — winner highlighted

AMD Instinct MI325X 256GB

Metric

NVIDIA A10G

FP16 Tensor TFLOPS (dense)

1.5k T
125 T

Bandwidth

6.0k GB/s
600 GB/s

VRAM

256 GB
24 GB

TDP (lower=better)

750 W
150 W

TFLOPS/watt

1.96
0.83

GB/s per watt

8.00
4.00

Full specification table

SpecAMD Instinct MI325X 256GBNVIDIA A10G
ArchitectureCDNA 3.5Ampere
VRAM256 GB HBM3e24 GB GDDR6
VRAM typeHBM3eGDDR6
Memory bandwidth6,000 GB/s600 GB/s
FP16 Tensor TFLOPS (dense)1,470 TFLOPS125 TFLOPS
TDP750 W150 W
TFLOPS/watt1.96 T/W0.83 T/W
GB/s per watt8.00 GB/s/W4.00 GB/s/W
NVLinkNoNo
Release year20242021

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

AMD Instinct MI325X 256GB

FP16 TFLOPS1,470 TFLOPS
TDP750 W
TFLOPS/watt1.960 T/W
GB/s per watt8.00 GB/s/W

NVIDIA A10G

FP16 TFLOPS125 TFLOPS
TDP150 W
TFLOPS/watt0.833 T/W
GB/s per watt4.00 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 AMD Instinct MI325X 256GB for:

  • Massive context LLM inference
  • Memory-bound workloads
  • Large open-source models

Weaknesses / Limitations

  • ROCm software ecosystem lags CUDA maturity
  • Very limited cloud availability vs NVIDIA equivalents
  • No FP8 hardware support in current ROCm stack

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