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	SambaNova SN40L	NVIDIA H100 SXM	Qualcomm Cloud AI 100
Architecture	CDNA 2	Reconfigurable Dataflow Unit (RDU)	Hopper	Qualcomm AI
Form Factor	OAM	—	SXM	PCIe
VRAM	128 GB	640 GB	80 GB	16 GB
Memory Bandwidth	3,277 GB/s	—	3,350 GB/s	134 GB/s
TFLOPs (FP32)	47.9	—	67	50
TFLOPs (FP16)	383	—	1,979	400
TFLOPs	383	—	1,979	50
TFLOPs (FP8)	—	—	3,958	—
TDP	560 W	700 W	700 W	75 W
Launch Date	Nov 2021	Jan 2024	Sep 2022	Sep 2020

Efficiency Metrics

Metric	MI250X	SN40L	H100 SXM	Cloud AI 100
TFLOPs per Watt (FP32-eq)	0.34	—	1.41	0.67
Memory Bandwidth per GB	25.6 GB/s	—	41.9 GB/s	8.4 GB/s

Performance Equivalence

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

To match 1x AMD MI250X

SambaNova SN40L

VRAM

0.20x

SN40L has 5.00x more

NVIDIA H100 SXM

Compute (FP32-eq)

0.19x

H100 SXM is 5.17x faster

FP32 Compute

0.71x

H100 SXM is 1.40x faster

VRAM

1.60x

Need 1.60x H100 SXM

Memory Bandwidth

0.98x

H100 SXM has 1.02x more

Qualcomm Cloud AI 100

Compute (FP32-eq)

3.83x

Need 3.83x Cloud AI 100

FP32 Compute

0.96x

Cloud AI 100 is 1.04x faster

VRAM

8.00x

Need 8.00x Cloud AI 100

Memory Bandwidth

24.46x

Need 24.46x Cloud AI 100

To match 1x SambaNova SN40L

AMD MI250X

VRAM

5.00x

Need 5.00x MI250X

NVIDIA H100 SXM

VRAM

8.00x

Need 8.00x H100 SXM

Qualcomm Cloud AI 100

VRAM

40.00x

Need 40.00x Cloud AI 100

To match 1x NVIDIA H100 SXM

AMD MI250X

Compute (FP32-eq)

5.17x

Need 5.17x MI250X

FP32 Compute

1.40x

Need 1.40x MI250X

VRAM

0.63x

MI250X has 1.60x more

Memory Bandwidth

1.02x

Need 1.02x MI250X

SambaNova SN40L

VRAM

0.13x

SN40L has 8.00x more

Qualcomm Cloud AI 100

Compute (FP32-eq)

19.79x

Need 19.79x Cloud AI 100

FP32 Compute

1.34x

Need 1.34x Cloud AI 100

VRAM

5.00x

Need 5.00x Cloud AI 100

Memory Bandwidth

25.00x

Need 25.00x Cloud AI 100

To match 1x Qualcomm Cloud AI 100

AMD MI250X

Compute (FP32-eq)

0.26x

MI250X is 3.83x faster

FP32 Compute

1.04x

Need 1.04x MI250X

VRAM

0.13x

MI250X has 8.00x more

Memory Bandwidth

0.04x

MI250X has 24.46x more

SambaNova SN40L

VRAM

0.03x

SN40L has 40.00x more

NVIDIA H100 SXM

Compute (FP32-eq)

0.05x

H100 SXM is 19.79x faster

FP32 Compute

0.75x

H100 SXM is 1.34x faster

VRAM

0.20x

H100 SXM has 5.00x more

Memory Bandwidth

0.04x

H100 SXM has 25.00x more

Pricing

Price Type	MI250X	SN40L	H100 SXM	Cloud AI 100
CAPEX (Street Price)	$12,000	—	$30,000	—
OPEX (per hour)	$2.00/hr	—	$3.50/hr	—
Price per TFLOPs (FP32-eq)	$63	—	$30	—