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	AMD MI250X	Google TPU v5e	NVIDIA GB200 NVL72	AMD Radeon PRO W7900
Architecture	CDNA 2	TPU v5	Blackwell	RDNA 3
Form Factor	OAM	Mezzanine	Custom	PCIe
VRAM	128 GB	—	13824 GB	48 GB
Memory Bandwidth	3,277 GB/s	—	130,000 GB/s	864 GB/s
TFLOPs (FP32)	47.9	—	6,480	61
TFLOPs (FP16)	383	—	360,000	122
TFLOPs	383	197	360,000	122
TFLOPs (FP8)	—	—	—	—
TDP	560 W	250 W	120000 W	295 W
Launch Date	Nov 2021	Aug 2023	Mar 2024	Apr 2023

Efficiency Metrics

Metric	MI250X	TPU v5e	GB200 NVL72	Radeon PRO W7900
TFLOPs per Watt (FP32-eq)	0.34	0.39	1.50	0.21
Memory Bandwidth per GB	25.6 GB/s	—	9.4 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 AMD MI250X

Google TPU v5e

Compute (FP32-eq)

1.94x

Need 1.94x TPU v5e

NVIDIA GB200 NVL72

Compute (FP32-eq)

0.00x

GB200 NVL72 is 939.95x faster

FP32 Compute

0.01x

GB200 NVL72 is 135.28x faster

VRAM

0.01x

GB200 NVL72 has 108.00x more

Memory Bandwidth

0.03x

GB200 NVL72 has 39.67x more

AMD Radeon PRO W7900

Compute (FP32-eq)

3.14x

Need 3.14x Radeon PRO W7900

FP32 Compute

0.79x

Radeon PRO W7900 is 1.27x faster

VRAM

2.67x

Need 2.67x Radeon PRO W7900

Memory Bandwidth

3.79x

Need 3.79x Radeon PRO W7900

To match 1x Google TPU v5e

AMD MI250X

Compute (FP32-eq)

0.51x

MI250X is 1.94x faster

NVIDIA GB200 NVL72

Compute (FP32-eq)

0.00x

GB200 NVL72 is 1827.41x faster

AMD Radeon PRO W7900

Compute (FP32-eq)

1.61x

Need 1.61x Radeon PRO W7900

To match 1x NVIDIA GB200 NVL72

AMD MI250X

Compute (FP32-eq)

939.95x

Need 939.95x MI250X

FP32 Compute

135.28x

Need 135.28x MI250X

VRAM

108.00x

Need 108.00x MI250X

Memory Bandwidth

39.67x

Need 39.67x MI250X

Google TPU v5e

Compute (FP32-eq)

1827.41x

Need 1827.41x TPU v5e

AMD Radeon PRO W7900

Compute (FP32-eq)

2950.82x

Need 2950.82x Radeon PRO W7900

FP32 Compute

106.23x

Need 106.23x Radeon PRO W7900

VRAM

288.00x

Need 288.00x Radeon PRO W7900

Memory Bandwidth

150.46x

Need 150.46x Radeon PRO W7900

To match 1x AMD Radeon PRO W7900

AMD MI250X

Compute (FP32-eq)

0.32x

MI250X is 3.14x faster

FP32 Compute

1.27x

Need 1.27x MI250X

VRAM

0.38x

MI250X has 2.67x more

Memory Bandwidth

0.26x

MI250X has 3.79x more

Google TPU v5e

Compute (FP32-eq)

0.62x

TPU v5e is 1.61x faster

NVIDIA GB200 NVL72

Compute (FP32-eq)

0.00x

GB200 NVL72 is 2950.82x faster

FP32 Compute

0.01x

GB200 NVL72 is 106.23x faster

VRAM

0.00x

GB200 NVL72 has 288.00x more

Memory Bandwidth

0.01x

GB200 NVL72 has 150.46x more

Pricing

Price Type	MI250X	TPU v5e	GB200 NVL72	Radeon PRO W7900
CAPEX (Street Price)	$12,000	—	—	—
OPEX (per hour)	$2.00/hr	$1.80/hr	—	—
Price per TFLOPs (FP32-eq)	$63	—	—	—