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	Alibaba Hanguang 800	AMD MI300X	AWS Trainium
Architecture	Ampere	XuanTie	CDNA 3	Inferentia/Trainium
Form Factor	SXM	—	OAM	—
VRAM	80 GB	—	192 GB	32 GB
Memory Bandwidth	2,039 GB/s	—	5,300 GB/s	—
TFLOPs (FP32)	19.5	—	163	—
TFLOPs (FP16)	312	—	1,307	—
TFLOPs	312	—	1,307	190
TFLOPs (FP8)	—	—	2,614	—
TDP	400 W	150 W	750 W	200 W
Launch Date	May 2020	Sep 2019	Dec 2023	Nov 2021

Efficiency Metrics

Metric	A100 SXM	Hanguang 800	MI300X	Trainium
TFLOPs per Watt (FP32-eq)	0.39	—	0.87	0.47
Memory Bandwidth per GB	25.5 GB/s	—	27.6 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

AMD MI300X

Compute (FP32-eq)

0.24x

MI300X is 4.19x faster

FP32 Compute

0.12x

MI300X is 8.36x faster

VRAM

0.42x

MI300X has 2.40x more

Memory Bandwidth

0.38x

MI300X has 2.60x more

AWS Trainium

Compute (FP32-eq)

1.64x

Need 1.64x Trainium

VRAM

2.50x

Need 2.50x Trainium

To match 1x AMD MI300X

NVIDIA A100 SXM

Compute (FP32-eq)

4.19x

Need 4.19x A100 SXM

FP32 Compute

8.36x

Need 8.36x A100 SXM

VRAM

2.40x

Need 2.40x A100 SXM

Memory Bandwidth

2.60x

Need 2.60x A100 SXM

AWS Trainium

Compute (FP32-eq)

6.88x

Need 6.88x Trainium

VRAM

6.00x

Need 6.00x Trainium

To match 1x AWS Trainium

NVIDIA A100 SXM

Compute (FP32-eq)

0.61x

A100 SXM is 1.64x faster

VRAM

0.40x

A100 SXM has 2.50x more

AMD MI300X

Compute (FP32-eq)

0.15x

MI300X is 6.88x faster

VRAM

0.17x

MI300X has 6.00x more

Pricing

Price Type	A100 SXM	Hanguang 800	MI300X	Trainium
CAPEX (Street Price)	$15,000	—	$35,000	—
OPEX (per hour)	$4.05/hr	—	$10.40/hr	—
Price per TFLOPs (FP32-eq)	$96	—	$54	—