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 RTX 3080 Ti 12GB

wins

1

The RTX 3080 Ti is a near-flagship Ampere consumer GPU with 12 GB GDDR6X and 68.2 TFLOPS FP16.

Budget inferenceSmall model fine-tuningExperimentation

Performance comparison

Visual bars — winner highlighted

AMD Instinct MI325X 256GB

Metric

NVIDIA RTX 3080 Ti 12GB

FP16 Tensor TFLOPS (dense)

1.5k T
68 T

Bandwidth

6.0k GB/s
912 GB/s

VRAM

256 GB
12 GB

TDP (lower=better)

750 W
350 W

TFLOPS/watt

1.96
0.19

GB/s per watt

8.00
2.61

Full specification table

SpecAMD Instinct MI325X 256GBNVIDIA RTX 3080 Ti 12GB
ArchitectureCDNA 3.5Ampere
VRAM256 GB HBM3e12 GB GDDR6X
VRAM typeHBM3eGDDR6X
Memory bandwidth6,000 GB/s912 GB/s
FP16 Tensor TFLOPS (dense)1,470 TFLOPS68.2 TFLOPS
TDP750 W350 W
TFLOPS/watt1.96 T/W0.19 T/W
GB/s per watt8.00 GB/s/W2.61 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 RTX 3080 Ti 12GB

FP16 TFLOPS68.2 TFLOPS
TDP350 W
TFLOPS/watt0.195 T/W
GB/s per watt2.61 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 RTX 3080 Ti 12GB for:

  • Budget inference
  • Small model fine-tuning
  • Experimentation

Weaknesses / Limitations

  • Only 12 GB VRAM — limits to very small quantized models
  • No BF16 hardware acceleration
  • Consumer-grade reliability — no ECC memory