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	NVIDIA A100 SXM	NVIDIA A40	AMD MI250X	Google TPU v5e
Architecture	Ampere	Ampere	CDNA 2	TPU v5
Form Factor	SXM	PCIe	OAM	Mezzanine
VRAM	80 GB	48 GB	128 GB	—
Memory Bandwidth	2,039 GB/s	696 GB/s	3,277 GB/s	—
TFLOPs (FP32)	19.5	37.4	47.9	—
TFLOPs (FP16)	312	150	383	—
TFLOPs	312	150	383	197
TFLOPs (FP8)	—	—	—	—
TDP	400 W	300 W	560 W	250 W
Launch Date	May 2020	Oct 2020	Nov 2021	Aug 2023

Efficiency Metrics

Metric	A100 SXM	A40	MI250X	TPU v5e
TFLOPs per Watt (FP32-eq)	0.39	0.25	0.34	0.39
Memory Bandwidth per GB	25.5 GB/s	14.5 GB/s	25.6 GB/s	—

Performance Equivalence

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

To match 1x NVIDIA A100 SXM

NVIDIA A40

Compute (FP32-eq)

2.08x

Need 2.08x A40

FP32 Compute

0.52x

A40 is 1.92x faster

VRAM

1.67x

Need 1.67x A40

Memory Bandwidth

2.93x

Need 2.93x A40

AMD MI250X

Compute (FP32-eq)

0.81x

MI250X is 1.23x faster

FP32 Compute

0.41x

MI250X is 2.46x faster

VRAM

0.63x

MI250X has 1.60x more

Memory Bandwidth

0.62x

MI250X has 1.61x more

Google TPU v5e

Compute (FP32-eq)

1.58x

Need 1.58x TPU v5e

To match 1x NVIDIA A40

NVIDIA A100 SXM

Compute (FP32-eq)

0.48x

A100 SXM is 2.08x faster

FP32 Compute

1.92x

Need 1.92x A100 SXM

VRAM

0.60x

A100 SXM has 1.67x more

Memory Bandwidth

0.34x

A100 SXM has 2.93x more

AMD MI250X

Compute (FP32-eq)

0.39x

MI250X is 2.55x faster

FP32 Compute

0.78x

MI250X is 1.28x faster

VRAM

0.38x

MI250X has 2.67x more

Memory Bandwidth

0.21x

MI250X has 4.71x more

Google TPU v5e

Compute (FP32-eq)

0.76x

TPU v5e is 1.31x faster

To match 1x AMD MI250X

NVIDIA A100 SXM

Compute (FP32-eq)

1.23x

Need 1.23x A100 SXM

FP32 Compute

2.46x

Need 2.46x A100 SXM

VRAM

1.60x

Need 1.60x A100 SXM

Memory Bandwidth

1.61x

Need 1.61x A100 SXM

NVIDIA A40

Compute (FP32-eq)

2.55x

Need 2.55x A40

FP32 Compute

1.28x

Need 1.28x A40

VRAM

2.67x

Need 2.67x A40

Memory Bandwidth

4.71x

Need 4.71x A40

Google TPU v5e

Compute (FP32-eq)

1.94x

Need 1.94x TPU v5e

To match 1x Google TPU v5e

NVIDIA A100 SXM

Compute (FP32-eq)

0.63x

A100 SXM is 1.58x faster

NVIDIA A40

Compute (FP32-eq)

1.31x

Need 1.31x A40

AMD MI250X

Compute (FP32-eq)

0.51x

MI250X is 1.94x faster

Pricing

Price Type	A100 SXM	A40	MI250X	TPU v5e
CAPEX (Street Price)	$15,000	—	$12,000	—
OPEX (per hour)	$4.05/hr	—	$2.00/hr	$1.80/hr
Price per TFLOPs (FP32-eq)	$96	—	$63	—