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 A40	AMD MI210	Etched Sohu	Huawei Ascend 910B
Architecture	Ampere	CDNA 2	Transformer ASIC	Da Vinci
Form Factor	PCIe	PCIe	PCIe	—
VRAM	48 GB	64 GB	144 GB	64 GB
Memory Bandwidth	696 GB/s	1,638 GB/s	4,800 GB/s	—
TFLOPs (FP32)	37.4	45.3	10,000	—
TFLOPs (FP16)	150	181	—	640
TFLOPs	150	181	10,000	—
TFLOPs (FP8)	—	—	—	—
TDP	300 W	300 W	700 W	400 W
Launch Date	Oct 2020	Jan 2022	Jun 2024	Aug 2023

Efficiency Metrics

Metric	A40	MI210	Sohu	Ascend 910B
TFLOPs per Watt (FP32-eq)	0.25	0.30	14.29	—
Memory Bandwidth per GB	14.5 GB/s	25.6 GB/s	33.3 GB/s	—

Performance Equivalence

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

To match 1x NVIDIA A40

AMD MI210

Compute (FP32-eq)

0.83x

MI210 is 1.21x faster

FP32 Compute

0.83x

MI210 is 1.21x faster

VRAM

0.75x

MI210 has 1.33x more

Memory Bandwidth

0.42x

MI210 has 2.35x more

Etched Sohu

Compute (FP32-eq)

0.01x

Sohu is 133.33x faster

FP32 Compute

0.00x

Sohu is 267.38x faster

VRAM

0.33x

Sohu has 3.00x more

Memory Bandwidth

0.14x

Sohu has 6.90x more

Huawei Ascend 910B

VRAM

0.75x

Ascend 910B has 1.33x more

To match 1x AMD MI210

NVIDIA A40

Compute (FP32-eq)

1.21x

Need 1.21x A40

FP32 Compute

1.21x

Need 1.21x A40

VRAM

1.33x

Need 1.33x A40

Memory Bandwidth

2.35x

Need 2.35x A40

Etched Sohu

Compute (FP32-eq)

0.01x

Sohu is 110.50x faster

FP32 Compute

0.00x

Sohu is 220.75x faster

VRAM

0.44x

Sohu has 2.25x more

Memory Bandwidth

0.34x

Sohu has 2.93x more

Huawei Ascend 910B

VRAM

1.00x

Ascend 910B has 1.00x more

To match 1x Etched Sohu

NVIDIA A40

Compute (FP32-eq)

133.33x

Need 133.33x A40

FP32 Compute

267.38x

Need 267.38x A40

VRAM

3.00x

Need 3.00x A40

Memory Bandwidth

6.90x

Need 6.90x A40

AMD MI210

Compute (FP32-eq)

110.50x

Need 110.50x MI210

FP32 Compute

220.75x

Need 220.75x MI210

VRAM

2.25x

Need 2.25x MI210

Memory Bandwidth

2.93x

Need 2.93x MI210

Huawei Ascend 910B

VRAM

2.25x

Need 2.25x Ascend 910B

To match 1x Huawei Ascend 910B

NVIDIA A40

VRAM

1.33x

Need 1.33x A40

AMD MI210

VRAM

1.00x

MI210 has 1.00x more

Etched Sohu

VRAM

0.44x

Sohu has 2.25x more

Pricing

Price Type	A40	MI210	Sohu	Ascend 910B
CAPEX (Street Price)	—	$6,000	—	—
OPEX (per hour)	—	—	—	—
Price per TFLOPs (FP32-eq)	—	$66	—	—