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	Cambricon MLU370	AWS Trainium	NVIDIA GeForce RTX 4080	AMD Radeon PRO W7900
Architecture	MLU Core	Inferentia/Trainium	Ada Lovelace	RDNA 3
Form Factor	—	—	PCIe	PCIe
VRAM	48 GB	32 GB	16 GB	48 GB
Memory Bandwidth	—	—	717 GB/s	864 GB/s
TFLOPs (FP32)	—	—	48.7	61
TFLOPs (FP16)	—	—	97.5	122
TFLOPs	256	190	97.5	122
TFLOPs (FP8)	—	—	—	—
TDP	300 W	200 W	320 W	295 W
Launch Date	Nov 2022	Nov 2021	Nov 2022	Apr 2023

Efficiency Metrics

Metric	MLU370	Trainium	GeForce RTX 4080	Radeon PRO W7900
TFLOPs per Watt (FP32-eq)	0.43	0.47	0.15	0.21
Memory Bandwidth per GB	—	—	44.8 GB/s	18.0 GB/s

Performance Equivalence

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

To match 1x Cambricon MLU370

AWS Trainium

Compute (FP32-eq)

1.35x

Need 1.35x Trainium

VRAM

1.50x

Need 1.50x Trainium

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

2.63x

Need 2.63x GeForce RTX 4080

VRAM

3.00x

Need 3.00x GeForce RTX 4080

AMD Radeon PRO W7900

Compute (FP32-eq)

2.10x

Need 2.10x Radeon PRO W7900

VRAM

1.00x

Radeon PRO W7900 has 1.00x more

To match 1x AWS Trainium

Cambricon MLU370

Compute (FP32-eq)

0.74x

MLU370 is 1.35x faster

VRAM

0.67x

MLU370 has 1.50x more

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

1.95x

Need 1.95x GeForce RTX 4080

VRAM

2.00x

Need 2.00x GeForce RTX 4080

AMD Radeon PRO W7900

Compute (FP32-eq)

1.56x

Need 1.56x Radeon PRO W7900

VRAM

0.67x

Radeon PRO W7900 has 1.50x more

To match 1x NVIDIA GeForce RTX 4080

Cambricon MLU370

Compute (FP32-eq)

0.38x

MLU370 is 2.63x faster

VRAM

0.33x

MLU370 has 3.00x more

AWS Trainium

Compute (FP32-eq)

0.51x

Trainium is 1.95x faster

VRAM

0.50x

Trainium has 2.00x more

AMD Radeon PRO W7900

Compute (FP32-eq)

0.80x

Radeon PRO W7900 is 1.25x faster

FP32 Compute

0.80x

Radeon PRO W7900 is 1.25x faster

VRAM

0.33x

Radeon PRO W7900 has 3.00x more

Memory Bandwidth

0.83x

Radeon PRO W7900 has 1.21x more

To match 1x AMD Radeon PRO W7900

Cambricon MLU370

Compute (FP32-eq)

0.48x

MLU370 is 2.10x faster

VRAM

1.00x

MLU370 has 1.00x more

AWS Trainium

Compute (FP32-eq)

0.64x

Trainium is 1.56x faster

VRAM

1.50x

Need 1.50x Trainium

NVIDIA GeForce RTX 4080

Compute (FP32-eq)

1.25x

Need 1.25x GeForce RTX 4080

FP32 Compute

1.25x

Need 1.25x GeForce RTX 4080

VRAM

3.00x

Need 3.00x GeForce RTX 4080

Memory Bandwidth

1.21x

Need 1.21x GeForce RTX 4080

Pricing

Price Type	MLU370	Trainium	GeForce RTX 4080	Radeon PRO W7900
CAPEX (Street Price)	—	—	—	—
OPEX (per hour)	—	—	—	—
Price per TFLOPs (FP32-eq)	—	—	—	—