Compare XPUs

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

Select XPUs to Compare

Clear all (5)Jump to results

Filter by Vendor

Showing 72 XPUs • 5 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 L40S	NVIDIA B200	Cerebras WSE-3	AWS Inferentia2	Huawei Ascend 910B
Architecture	Ada Lovelace	Blackwell	Wafer Scale Engine	Inferentia Gen2	Da Vinci
Form Factor	PCIe	SXM	—	—	—
VRAM	48 GB	192 GB	44 GB	32 GB	64 GB
Memory Bandwidth	864 GB/s	8,000 GB/s	—	—	—
TFLOPs (FP32)	91.6	90	—	—	—
TFLOPs (FP16)	733	2,250	—	—	640
TFLOPs	733	2,250	—	190	—
TFLOPs (FP8)	1,466	4,500	—	—	—
TDP	350 W	1000 W	23000 W	150 W	400 W
Launch Date	Oct 2023	Mar 2024	Mar 2024	Nov 2022	Aug 2023

Efficiency Metrics

Metric	L40S	B200	WSE-3	Inferentia2	Ascend 910B
TFLOPs per Watt (FP32-eq)	1.05	1.13	—	0.63	—
Memory Bandwidth per GB	18.0 GB/s	41.7 GB/s	—	—	—

Performance Equivalence

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

To match 1x NVIDIA L40S

NVIDIA B200

Compute (FP32-eq)

0.33x

B200 is 3.07x faster

FP32 Compute

1.02x

Need 1.02x B200

VRAM

0.25x

B200 has 4.00x more

Memory Bandwidth

0.11x

B200 has 9.26x more

Cerebras WSE-3

VRAM

1.09x

Need 1.09x WSE-3

AWS Inferentia2

Compute (FP32-eq)

3.86x

Need 3.86x Inferentia2

VRAM

1.50x

Need 1.50x Inferentia2

Huawei Ascend 910B

VRAM

0.75x

Ascend 910B has 1.33x more

To match 1x NVIDIA B200

NVIDIA L40S

Compute (FP32-eq)

3.07x

Need 3.07x L40S

FP32 Compute

0.98x

L40S is 1.02x faster

VRAM

4.00x

Need 4.00x L40S

Memory Bandwidth

9.26x

Need 9.26x L40S

Cerebras WSE-3

VRAM

4.36x

Need 4.36x WSE-3

AWS Inferentia2

Compute (FP32-eq)

11.84x

Need 11.84x Inferentia2

VRAM

6.00x

Need 6.00x Inferentia2

Huawei Ascend 910B

VRAM

3.00x

Need 3.00x Ascend 910B

To match 1x Cerebras WSE-3

NVIDIA L40S

VRAM

0.92x

L40S has 1.09x more

NVIDIA B200

VRAM

0.23x

B200 has 4.36x more

AWS Inferentia2

VRAM

1.38x

Need 1.38x Inferentia2

Huawei Ascend 910B

VRAM

0.69x

Ascend 910B has 1.45x more

To match 1x AWS Inferentia2

NVIDIA L40S

Compute (FP32-eq)

0.26x

L40S is 3.86x faster

VRAM

0.67x

L40S has 1.50x more

NVIDIA B200

Compute (FP32-eq)

0.08x

B200 is 11.84x faster

VRAM

0.17x

B200 has 6.00x more

Cerebras WSE-3

VRAM

0.73x

WSE-3 has 1.38x more

Huawei Ascend 910B

VRAM

0.50x

Ascend 910B has 2.00x more

To match 1x Huawei Ascend 910B

NVIDIA L40S

VRAM

1.33x

Need 1.33x L40S

NVIDIA B200

VRAM

0.33x

B200 has 3.00x more

Cerebras WSE-3

VRAM

1.45x

Need 1.45x WSE-3

AWS Inferentia2

VRAM

2.00x

Need 2.00x Inferentia2

Pricing

Price Type	L40S	B200	WSE-3	Inferentia2	Ascend 910B
CAPEX (Street Price)	$10,000	$70,000	—	—	—
OPEX (per hour)	$1.50/hr	—	—	—	—
Price per TFLOPs (FP32-eq)	$27	$62	—	—	—