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 128 XPUs • 4 selected

Alibaba

Hanguang 800

AMD

MI100

23.1 TFLOPs

AMD

MI210

181 TFLOPs

AMD

MI250X

383 TFLOPs

AMD

MI300X

1,307 TFLOPs

AMD

MI325X

1,400 TFLOPs

AMD

MI350X

2,100 TFLOPs

AMD

MI355X

5,300 TFLOPs

AMD

Radeon Pro V520

23.04 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

FuriosaAI

RNGD (Renegade)

256 TFLOPs

FuriosaAI

Warboy

Google

TPU v4

275 TFLOPs

Google

TPU v5e

197 TFLOPs

Google

TPU v5p

459 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	Intel Data Center GPU Max 1550	NVIDIA GeForce RTX 4070 Ti	NVIDIA GeForce RTX 3090	NVIDIA GH200
Architecture	Ponte Vecchio	Ada Lovelace	Ampere	Hopper
Form Factor	OAM	PCIe	PCIe	SXM
VRAM	128 GB	12 GB	24 GB	96 GB
Memory Bandwidth	3,277 GB/s	504 GB/s	936 GB/s	4,000 GB/s
TFLOPs (FP32)	52	40.1	35.6	67
TFLOPs (FP16)	419	—	—	—
TFLOPs	419	80.16	71	989
TFLOPs (FP8)	—	—	—	—
TDP	600 W	285 W	350 W	1000 W
Launch Date	Jan 2023	Jan 2023	Sep 2020	May 2023

Efficiency Metrics

Metric	Data Center GPU Max 1550	GeForce RTX 4070 Ti	GeForce RTX 3090	GH200
TFLOPs per Watt (FP32-eq)	0.35	0.14	0.10	0.49
Memory Bandwidth per GB	25.6 GB/s	42.0 GB/s	39.0 GB/s	41.7 GB/s

Performance Equivalence

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

To match 1x Intel Data Center GPU Max 1550

NVIDIA GeForce RTX 4070 Ti

Compute (FP32-eq)

5.23x

Need 5.23x GeForce RTX 4070 Ti

FP32 Compute

1.30x

Need 1.30x GeForce RTX 4070 Ti

VRAM

10.67x

Need 10.67x GeForce RTX 4070 Ti

Memory Bandwidth

6.50x

Need 6.50x GeForce RTX 4070 Ti

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

5.90x

Need 5.90x GeForce RTX 3090

FP32 Compute

1.46x

Need 1.46x GeForce RTX 3090

VRAM

5.33x

Need 5.33x GeForce RTX 3090

Memory Bandwidth

3.50x

Need 3.50x GeForce RTX 3090

NVIDIA GH200

Compute (FP32-eq)

0.42x

GH200 is 2.36x faster

FP32 Compute

0.78x

GH200 is 1.29x faster

VRAM

1.33x

Need 1.33x GH200

Memory Bandwidth

0.82x

GH200 has 1.22x more

To match 1x NVIDIA GeForce RTX 4070 Ti

Intel Data Center GPU Max 1550

Compute (FP32-eq)

0.19x

Data Center GPU Max 1550 is 5.23x faster

FP32 Compute

0.77x

Data Center GPU Max 1550 is 1.30x faster

VRAM

0.09x

Data Center GPU Max 1550 has 10.67x more

Memory Bandwidth

0.15x

Data Center GPU Max 1550 has 6.50x more

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

1.13x

Need 1.13x GeForce RTX 3090

FP32 Compute

1.13x

Need 1.13x GeForce RTX 3090

VRAM

0.50x

GeForce RTX 3090 has 2.00x more

Memory Bandwidth

0.54x

GeForce RTX 3090 has 1.86x more

NVIDIA GH200

Compute (FP32-eq)

0.08x

GH200 is 12.34x faster

FP32 Compute

0.60x

GH200 is 1.67x faster

VRAM

0.13x

GH200 has 8.00x more

Memory Bandwidth

0.13x

GH200 has 7.94x more

To match 1x NVIDIA GeForce RTX 3090

Intel Data Center GPU Max 1550

Compute (FP32-eq)

0.17x

Data Center GPU Max 1550 is 5.90x faster

FP32 Compute

0.68x

Data Center GPU Max 1550 is 1.46x faster

VRAM

0.19x

Data Center GPU Max 1550 has 5.33x more

Memory Bandwidth

0.29x

Data Center GPU Max 1550 has 3.50x more

NVIDIA GeForce RTX 4070 Ti

Compute (FP32-eq)

0.89x

GeForce RTX 4070 Ti is 1.13x faster

FP32 Compute

0.89x

GeForce RTX 4070 Ti is 1.13x faster

VRAM

2.00x

Need 2.00x GeForce RTX 4070 Ti

Memory Bandwidth

1.86x

Need 1.86x GeForce RTX 4070 Ti

NVIDIA GH200

Compute (FP32-eq)

0.07x

GH200 is 13.93x faster

FP32 Compute

0.53x

GH200 is 1.88x faster

VRAM

0.25x

GH200 has 4.00x more

Memory Bandwidth

0.23x

GH200 has 4.27x more

To match 1x NVIDIA GH200

Intel Data Center GPU Max 1550

Compute (FP32-eq)

2.36x

Need 2.36x Data Center GPU Max 1550

FP32 Compute

1.29x

Need 1.29x Data Center GPU Max 1550

VRAM

0.75x

Data Center GPU Max 1550 has 1.33x more

Memory Bandwidth

1.22x

Need 1.22x Data Center GPU Max 1550

NVIDIA GeForce RTX 4070 Ti

Compute (FP32-eq)

12.34x

Need 12.34x GeForce RTX 4070 Ti

FP32 Compute

1.67x

Need 1.67x GeForce RTX 4070 Ti

VRAM

8.00x

Need 8.00x GeForce RTX 4070 Ti

Memory Bandwidth

7.94x

Need 7.94x GeForce RTX 4070 Ti

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

13.93x

Need 13.93x GeForce RTX 3090

FP32 Compute

1.88x

Need 1.88x GeForce RTX 3090

VRAM

4.00x

Need 4.00x GeForce RTX 3090

Memory Bandwidth

4.27x

Need 4.27x GeForce RTX 3090

Pricing

Price Type	Data Center GPU Max 1550	GeForce RTX 4070 Ti	GeForce RTX 3090	GH200
CAPEX (Street Price)	$10,000	—	—	—
OPEX (per hour)	—	$0.10/hr	$0.11/hr	$1.49/hr
Price per TFLOPs (FP32-eq)	$48	—	—	—