Compare XPUs

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

Select XPUs to Compare

Clear all (5)Jump to results

Filter by Vendor

Showing 128 XPUs • 5 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 Habana Gaudi 3	Tenstorrent Grayskull	Biren Technology BR100	NVIDIA RTX A2000	NVIDIA GeForce RTX 3090
Architecture	Gaudi Gen3	Tensix Core	Biren GPU	Ampere	Ampere
Form Factor	OAM	—	—	PCIe	PCIe
VRAM	128 GB	8 GB	64 GB	12 GB	24 GB
Memory Bandwidth	3,700 GB/s	—	—	288 GB/s	936 GB/s
TFLOPs (FP32)	—	—	—	8	35.6
TFLOPs (FP16)	—	—	1,000	—	—
TFLOPs	1,835	200	—	16	71
TFLOPs (FP8)	3,670	—	—	—	—
TDP	900 W	65 W	550 W	70 W	350 W
Launch Date	Apr 2024	Mar 2021	Aug 2022	Oct 2021	Sep 2020

Efficiency Metrics

Metric	Gaudi 3	Grayskull	BR100	RTX A2000	GeForce RTX 3090
TFLOPs per Watt (FP32-eq)	1.02	1.54	—	0.11	0.10
Memory Bandwidth per GB	28.9 GB/s	—	—	24.0 GB/s	39.0 GB/s

Performance Equivalence

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

To match 1x Intel Habana Gaudi 3

Tenstorrent Grayskull

Compute (FP32-eq)

9.18x

Need 9.18x Grayskull

VRAM

16.00x

Need 16.00x Grayskull

Biren Technology BR100

VRAM

2.00x

Need 2.00x BR100

NVIDIA RTX A2000

Compute (FP32-eq)

114.69x

Need 114.69x RTX A2000

VRAM

10.67x

Need 10.67x RTX A2000

Memory Bandwidth

12.85x

Need 12.85x RTX A2000

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

25.85x

Need 25.85x GeForce RTX 3090

VRAM

5.33x

Need 5.33x GeForce RTX 3090

Memory Bandwidth

3.95x

Need 3.95x GeForce RTX 3090

To match 1x Tenstorrent Grayskull

Intel Habana Gaudi 3

Compute (FP32-eq)

0.11x

Gaudi 3 is 9.18x faster

VRAM

0.06x

Gaudi 3 has 16.00x more

Biren Technology BR100

VRAM

0.13x

BR100 has 8.00x more

NVIDIA RTX A2000

Compute (FP32-eq)

12.50x

Need 12.50x RTX A2000

VRAM

0.67x

RTX A2000 has 1.50x more

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

2.82x

Need 2.82x GeForce RTX 3090

VRAM

0.33x

GeForce RTX 3090 has 3.00x more

To match 1x Biren Technology BR100

Intel Habana Gaudi 3

VRAM

0.50x

Gaudi 3 has 2.00x more

Tenstorrent Grayskull

VRAM

8.00x

Need 8.00x Grayskull

NVIDIA RTX A2000

VRAM

5.33x

Need 5.33x RTX A2000

NVIDIA GeForce RTX 3090

VRAM

2.67x

Need 2.67x GeForce RTX 3090

To match 1x NVIDIA RTX A2000

Intel Habana Gaudi 3

Compute (FP32-eq)

0.01x

Gaudi 3 is 114.69x faster

VRAM

0.09x

Gaudi 3 has 10.67x more

Memory Bandwidth

0.08x

Gaudi 3 has 12.85x more

Tenstorrent Grayskull

Compute (FP32-eq)

0.08x

Grayskull is 12.50x faster

VRAM

1.50x

Need 1.50x Grayskull

Biren Technology BR100

VRAM

0.19x

BR100 has 5.33x more

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

0.23x

GeForce RTX 3090 is 4.44x faster

FP32 Compute

0.22x

GeForce RTX 3090 is 4.45x faster

VRAM

0.50x

GeForce RTX 3090 has 2.00x more

Memory Bandwidth

0.31x

GeForce RTX 3090 has 3.25x more

To match 1x NVIDIA GeForce RTX 3090

Intel Habana Gaudi 3

Compute (FP32-eq)

0.04x

Gaudi 3 is 25.85x faster

VRAM

0.19x

Gaudi 3 has 5.33x more

Memory Bandwidth

0.25x

Gaudi 3 has 3.95x more

Tenstorrent Grayskull

Compute (FP32-eq)

0.35x

Grayskull is 2.82x faster

VRAM

3.00x

Need 3.00x Grayskull

Biren Technology BR100

VRAM

0.38x

BR100 has 2.67x more

NVIDIA RTX A2000

Compute (FP32-eq)

4.44x

Need 4.44x RTX A2000

FP32 Compute

4.45x

Need 4.45x RTX A2000

VRAM

2.00x

Need 2.00x RTX A2000

Memory Bandwidth

3.25x

Need 3.25x RTX A2000

Pricing

Price Type	Gaudi 3	Grayskull	BR100	RTX A2000	GeForce RTX 3090
CAPEX (Street Price)	$15,000	—	—	—	—
OPEX (per hour)	$1.20/hr	—	—	$0.04/hr	$0.11/hr
Price per TFLOPs (FP32-eq)	$16	—	—	—	—