Compare XPUs

Select up to 5 XPUs to compare side-by-side

Select XPUs to Compare

Clear all (4)Jump to results

Filter by Vendor

Showing 72 XPUs • 4 selected

Alibaba

Hanguang 800

AMD

MI100

23.1 TFLOPs

AMD

MI210

181 TFLOPs

AMD

MI250X

383 TFLOPs

AMD

MI300A

980.6 TFLOPs

AMD

MI300X

1,307 TFLOPs

AMD

MI325X

1,400 TFLOPs

AMD

MI350X

2,100 TFLOPs

AMD

Radeon PRO W7900

122 TFLOPs

AMD

Radeon RX 7900 XT

104 TFLOPs

AMD

Radeon RX 7900 XTX

122 TFLOPs

AWS

Inferentia2

190 TFLOPs

AWS

Trainium

190 TFLOPs

AWS

Trainium2

680 TFLOPs

Baidu

Kunlun II

Biren Technology

BR100

Cambricon

MLU370

256 TFLOPs

Cerebras

WSE-3

Enflame Technology

CloudBlazer T20

Etched

Sohu

10,000 TFLOPs

FuriosaAI

Warboy

Google

TPU v4

275 TFLOPs

Google

TPU v5e

197 TFLOPs

Google

TPU v5p

459 TFLOPs

Google

TPU v6e (Trillium)

918 TFLOPs

Graphcore

Bow IPU

Graphcore

IPU-M2000

Groq

LPU Inference Engine

Huawei

Ascend 910B

Iluvatar CoreX

BI-V150

300 TFLOPs

Intel

Data Center GPU Max 1100

177 TFLOPs

Intel

Data Center GPU Max 1550

419 TFLOPs

Intel Habana

Gaudi 2

432 TFLOPs

Intel Habana

Gaudi 3

1,835 TFLOPs

Multi-Metric Comparison

Relative performance across 5 key metrics (normalized to 100 = best in comparison)

Compute Performance (BF16)

Memory Capacity

Power Consumption

Power Efficiency

Specifications

Specification	AMD MI250X	AMD MI325X	AWS Trainium2	NVIDIA GeForce RTX 4080
Architecture	CDNA 2	CDNA 3.5	Trainium Gen2	Ada Lovelace
Form Factor	OAM	OAM	—	PCIe
VRAM	128 GB	256 GB	—	16 GB
Memory Bandwidth	3,277 GB/s	6,000 GB/s	—	717 GB/s
TFLOPs (FP32)	47.9	175	—	48.7
TFLOPs (FP16)	383	1,400	—	97.5
TFLOPs	383	1,400	680	97.5
TFLOPs (FP8)	—	2,800	—	—
TDP	560 W	800 W	300 W	320 W
Launch Date	Nov 2021	Oct 2024	Nov 2023	Nov 2022

Efficiency Metrics

Metric	MI250X	MI325X	Trainium2	GeForce RTX 4080
TFLOPs per Watt (FP32-eq)	0.34	0.88	1.13	0.15
Memory Bandwidth per GB	25.6 GB/s	23.4 GB/s	—	44.8 GB/s

Performance Equivalence

How many units of each GPU are needed to match the performance of the others?

To match 1x AMD MI250X

AMD MI325X

Compute (FP32-eq)

0.27x

MI325X is 3.66x faster

FP32 Compute

0.27x

MI325X is 3.65x faster

VRAM

0.50x

MI325X has 2.00x more

Memory Bandwidth

0.55x

MI325X has 1.83x more

AWS Trainium2

Compute (FP32-eq)

0.56x

Trainium2 is 1.78x faster

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

3.93x

Need 3.93x GeForce RTX 4080

FP32 Compute

0.98x

GeForce RTX 4080 is 1.02x faster

VRAM

8.00x

Need 8.00x GeForce RTX 4080

Memory Bandwidth

4.57x

Need 4.57x GeForce RTX 4080

To match 1x AMD MI325X

AMD MI250X

Compute (FP32-eq)

3.66x

Need 3.66x MI250X

FP32 Compute

3.65x

Need 3.65x MI250X

VRAM

2.00x

Need 2.00x MI250X

Memory Bandwidth

1.83x

Need 1.83x MI250X

AWS Trainium2

Compute (FP32-eq)

2.06x

Need 2.06x Trainium2

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

14.36x

Need 14.36x GeForce RTX 4080

FP32 Compute

3.59x

Need 3.59x GeForce RTX 4080

VRAM

16.00x

Need 16.00x GeForce RTX 4080

Memory Bandwidth

8.37x

Need 8.37x GeForce RTX 4080

To match 1x AWS Trainium2

AMD MI250X

Compute (FP32-eq)

1.78x

Need 1.78x MI250X

AMD MI325X

Compute (FP32-eq)

0.49x

MI325X is 2.06x faster

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

6.97x

Need 6.97x GeForce RTX 4080

To match 1x NVIDIA GeForce RTX 4080

AMD MI250X

Compute (FP32-eq)

0.25x

MI250X is 3.93x faster

FP32 Compute

1.02x

Need 1.02x MI250X

VRAM

0.13x

MI250X has 8.00x more

Memory Bandwidth

0.22x

MI250X has 4.57x more

AMD MI325X

Compute (FP32-eq)

0.07x

MI325X is 14.36x faster

FP32 Compute

0.28x

MI325X is 3.59x faster

VRAM

0.06x

MI325X has 16.00x more

Memory Bandwidth

0.12x

MI325X has 8.37x more

AWS Trainium2

Compute (FP32-eq)

0.14x

Trainium2 is 6.97x faster

Pricing

Price Type	MI250X	MI325X	Trainium2	GeForce RTX 4080
CAPEX (Street Price)	$12,000	$45,000	—	—
OPEX (per hour)	$2.00/hr	—	$1.34/hr	—
Price per TFLOPs (FP32-eq)	$63	$64	—	—