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 128 XPUs • 3 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	NVIDIA RTX 4000 SFF Ada Generation	NVIDIA A100 SXM	NVIDIA GeForce RTX 3070 Ti
Architecture	Ada Lovelace	Ampere	Ampere
Form Factor	PCIe	SXM	PCIe
VRAM	20 GB	80 GB	8 GB
Memory Bandwidth	360 GB/s	2,039 GB/s	608 GB/s
TFLOPs (FP32)	26.7	19.5	21.8
TFLOPs (FP16)	—	312	—
TFLOPs	52	312	43.2
TFLOPs (FP8)	—	—	—
TDP	70 W	400 W	290 W
Launch Date	Apr 2023	May 2020	Jun 2021

Efficiency Metrics

Metric	RTX 4000 SFF Ada Generation	A100 SXM	GeForce RTX 3070 Ti
TFLOPs per Watt (FP32-eq)	0.37	0.39	0.07
Memory Bandwidth per GB	18.0 GB/s	25.5 GB/s	76.0 GB/s

Performance Equivalence

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

To match 1x NVIDIA RTX 4000 SFF Ada Generation

NVIDIA A100 SXM

Compute (FP32-eq)

0.17x

A100 SXM is 6.00x faster

FP32 Compute

1.37x

Need 1.37x A100 SXM

VRAM

0.25x

A100 SXM has 4.00x more

Memory Bandwidth

0.18x

A100 SXM has 5.66x more

NVIDIA GeForce RTX 3070 Ti

Compute (FP32-eq)

1.20x

Need 1.20x GeForce RTX 3070 Ti

FP32 Compute

1.22x

Need 1.22x GeForce RTX 3070 Ti

VRAM

2.50x

Need 2.50x GeForce RTX 3070 Ti

Memory Bandwidth

0.59x

GeForce RTX 3070 Ti has 1.69x more

To match 1x NVIDIA A100 SXM

NVIDIA RTX 4000 SFF Ada Generation

Compute (FP32-eq)

6.00x

Need 6.00x RTX 4000 SFF Ada Generation

FP32 Compute

0.73x

RTX 4000 SFF Ada Generation is 1.37x faster

VRAM

4.00x

Need 4.00x RTX 4000 SFF Ada Generation

Memory Bandwidth

5.66x

Need 5.66x RTX 4000 SFF Ada Generation

NVIDIA GeForce RTX 3070 Ti

Compute (FP32-eq)

7.22x

Need 7.22x GeForce RTX 3070 Ti

FP32 Compute

0.89x

GeForce RTX 3070 Ti is 1.12x faster

VRAM

10.00x

Need 10.00x GeForce RTX 3070 Ti

Memory Bandwidth

3.35x

Need 3.35x GeForce RTX 3070 Ti

To match 1x NVIDIA GeForce RTX 3070 Ti

NVIDIA RTX 4000 SFF Ada Generation

Compute (FP32-eq)

0.83x

RTX 4000 SFF Ada Generation is 1.20x faster

FP32 Compute

0.82x

RTX 4000 SFF Ada Generation is 1.22x faster

VRAM

0.40x

RTX 4000 SFF Ada Generation has 2.50x more

Memory Bandwidth

1.69x

Need 1.69x RTX 4000 SFF Ada Generation

NVIDIA A100 SXM

Compute (FP32-eq)

0.14x

A100 SXM is 7.22x faster

FP32 Compute

1.12x

Need 1.12x A100 SXM

VRAM

0.10x

A100 SXM has 10.00x more

Memory Bandwidth

0.30x

A100 SXM has 3.35x more

Pricing

Price Type	RTX 4000 SFF Ada Generation	A100 SXM	GeForce RTX 3070 Ti
CAPEX (Street Price)	—	$15,000	—
OPEX (per hour)	$0.15/hr	$4.05/hr	$0.08/hr
Price per TFLOPs (FP32-eq)	—	$96	—