Compare XPUs

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

Select XPUs to Compare

Clear all (3)Jump to results

Filter by Vendor

Showing 72 XPUs • 3 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 H100 SXM	AMD MI300X	NVIDIA RTX PRO 6000 Blackwell Server Edition
Architecture	Hopper	CDNA 3	Blackwell
Form Factor	SXM	OAM	PCIe
VRAM	80 GB	192 GB	96 GB
Memory Bandwidth	3,350 GB/s	5,300 GB/s	1,792 GB/s
TFLOPs (FP32)	67	163	125
TFLOPs (FP16)	1,979	1,307	250
TFLOPs	1,979	1,307	250
TFLOPs (FP8)	3,958	2,614	—
TDP	700 W	750 W	600 W
Launch Date	Sep 2022	Dec 2023	Mar 2025

Efficiency Metrics

Metric	H100 SXM	MI300X	RTX PRO 6000 Blackwell Server Edition
TFLOPs per Watt (FP32-eq)	1.41	0.87	0.21
Memory Bandwidth per GB	41.9 GB/s	27.6 GB/s	18.7 GB/s

Performance Equivalence

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

To match 1x NVIDIA H100 SXM

AMD MI300X

Compute (FP32-eq)

1.51x

Need 1.51x MI300X

FP32 Compute

0.41x

MI300X is 2.43x faster

VRAM

0.42x

MI300X has 2.40x more

Memory Bandwidth

0.63x

MI300X has 1.58x more

NVIDIA RTX PRO 6000 Blackwell Server Edition

Compute (FP32-eq)

7.92x

Need 7.92x RTX PRO 6000 Blackwell Server Edition

FP32 Compute

0.54x

RTX PRO 6000 Blackwell Server Edition is 1.87x faster

VRAM

0.83x

RTX PRO 6000 Blackwell Server Edition has 1.20x more

Memory Bandwidth

1.87x

Need 1.87x RTX PRO 6000 Blackwell Server Edition

To match 1x AMD MI300X

NVIDIA H100 SXM

Compute (FP32-eq)

0.66x

H100 SXM is 1.51x faster

FP32 Compute

2.43x

Need 2.43x H100 SXM

VRAM

2.40x

Need 2.40x H100 SXM

Memory Bandwidth

1.58x

Need 1.58x H100 SXM

NVIDIA RTX PRO 6000 Blackwell Server Edition

Compute (FP32-eq)

5.23x

Need 5.23x RTX PRO 6000 Blackwell Server Edition

FP32 Compute

1.30x

Need 1.30x RTX PRO 6000 Blackwell Server Edition

VRAM

2.00x

Need 2.00x RTX PRO 6000 Blackwell Server Edition

Memory Bandwidth

2.96x

Need 2.96x RTX PRO 6000 Blackwell Server Edition

To match 1x NVIDIA RTX PRO 6000 Blackwell Server Edition

NVIDIA H100 SXM

Compute (FP32-eq)

0.13x

H100 SXM is 7.92x faster

FP32 Compute

1.87x

Need 1.87x H100 SXM

VRAM

1.20x

Need 1.20x H100 SXM

Memory Bandwidth

0.53x

H100 SXM has 1.87x more

AMD MI300X

Compute (FP32-eq)

0.19x

MI300X is 5.23x faster

FP32 Compute

0.77x

MI300X is 1.30x faster

VRAM

0.50x

MI300X has 2.00x more

Memory Bandwidth

0.34x

MI300X has 2.96x more

Pricing

Price Type	H100 SXM	MI300X	RTX PRO 6000 Blackwell Server Edition
CAPEX (Street Price)	$30,000	$35,000	—
OPEX (per hour)	$3.50/hr	$10.40/hr	—
Price per TFLOPs (FP32-eq)	$30	$54	—