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	Graphcore Bow IPU	AMD MI300X	NVIDIA Quadro P600	NVIDIA RTX A2000	NVIDIA A100 SXM
Architecture	Bow-2000	CDNA 3	Pascal	Ampere	Ampere
Form Factor	—	OAM	PCIe	PCIe	SXM
VRAM	3.6 GB	192 GB	2 GB	12 GB	80 GB
Memory Bandwidth	—	5,300 GB/s	64 GB/s	288 GB/s	2,039 GB/s
TFLOPs (FP32)	—	163.4	1.117	8	19.5
TFLOPs (FP16)	1,000	1,307	—	—	312
TFLOPs	—	1,307	1.117	16	312
TFLOPs (FP8)	—	2,614	—	—	—
TDP	200 W	750 W	40 W	70 W	400 W
Launch Date	Mar 2022	Dec 2023	Feb 2017	Oct 2021	May 2020

Efficiency Metrics

Metric	Bow IPU	MI300X	Quadro P600	RTX A2000	A100 SXM
TFLOPs per Watt (FP32-eq)	—	0.87	0.03	0.11	0.39
Memory Bandwidth per GB	—	27.6 GB/s	32.0 GB/s	24.0 GB/s	25.5 GB/s

Performance Equivalence

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

To match 1x Graphcore Bow IPU

AMD MI300X

VRAM

0.02x

MI300X has 53.33x more

NVIDIA Quadro P600

VRAM

1.80x

Need 1.80x Quadro P600

NVIDIA RTX A2000

VRAM

0.30x

RTX A2000 has 3.33x more

NVIDIA A100 SXM

VRAM

0.04x

A100 SXM has 22.22x more

To match 1x AMD MI300X

Graphcore Bow IPU

VRAM

53.33x

Need 53.33x Bow IPU

NVIDIA Quadro P600

Compute (FP32-eq)

585.05x

Need 585.05x Quadro P600

FP32 Compute

146.28x

Need 146.28x Quadro P600

VRAM

96.00x

Need 96.00x Quadro P600

Memory Bandwidth

82.81x

Need 82.81x Quadro P600

NVIDIA RTX A2000

Compute (FP32-eq)

81.69x

Need 81.69x RTX A2000

FP32 Compute

20.43x

Need 20.43x RTX A2000

VRAM

16.00x

Need 16.00x RTX A2000

Memory Bandwidth

18.40x

Need 18.40x RTX A2000

NVIDIA A100 SXM

Compute (FP32-eq)

4.19x

Need 4.19x A100 SXM

FP32 Compute

8.38x

Need 8.38x A100 SXM

VRAM

2.40x

Need 2.40x A100 SXM

Memory Bandwidth

2.60x

Need 2.60x A100 SXM

To match 1x NVIDIA Quadro P600

Graphcore Bow IPU

VRAM

0.56x

Bow IPU has 1.80x more

AMD MI300X

Compute (FP32-eq)

0.00x

MI300X is 585.05x faster

FP32 Compute

0.01x

MI300X is 146.28x faster

VRAM

0.01x

MI300X has 96.00x more

Memory Bandwidth

0.01x

MI300X has 82.81x more

NVIDIA RTX A2000

Compute (FP32-eq)

0.14x

RTX A2000 is 7.16x faster

FP32 Compute

0.14x

RTX A2000 is 7.16x faster

VRAM

0.17x

RTX A2000 has 6.00x more

Memory Bandwidth

0.22x

RTX A2000 has 4.50x more

NVIDIA A100 SXM

Compute (FP32-eq)

0.01x

A100 SXM is 139.66x faster

FP32 Compute

0.06x

A100 SXM is 17.46x faster

VRAM

0.03x

A100 SXM has 40.00x more

Memory Bandwidth

0.03x

A100 SXM has 31.86x more

To match 1x NVIDIA RTX A2000

Graphcore Bow IPU

VRAM

3.33x

Need 3.33x Bow IPU

AMD MI300X

Compute (FP32-eq)

0.01x

MI300X is 81.69x faster

FP32 Compute

0.05x

MI300X is 20.43x faster

VRAM

0.06x

MI300X has 16.00x more

Memory Bandwidth

0.05x

MI300X has 18.40x more

NVIDIA Quadro P600

Compute (FP32-eq)

7.16x

Need 7.16x Quadro P600

FP32 Compute

7.16x

Need 7.16x Quadro P600

VRAM

6.00x

Need 6.00x Quadro P600

Memory Bandwidth

4.50x

Need 4.50x Quadro P600

NVIDIA A100 SXM

Compute (FP32-eq)

0.05x

A100 SXM is 19.50x faster

FP32 Compute

0.41x

A100 SXM is 2.44x faster

VRAM

0.15x

A100 SXM has 6.67x more

Memory Bandwidth

0.14x

A100 SXM has 7.08x more

To match 1x NVIDIA A100 SXM

Graphcore Bow IPU

VRAM

22.22x

Need 22.22x Bow IPU

AMD MI300X

Compute (FP32-eq)

0.24x

MI300X is 4.19x faster

FP32 Compute

0.12x

MI300X is 8.38x faster

VRAM

0.42x

MI300X has 2.40x more

Memory Bandwidth

0.38x

MI300X has 2.60x more

NVIDIA Quadro P600

Compute (FP32-eq)

139.66x

Need 139.66x Quadro P600

FP32 Compute

17.46x

Need 17.46x Quadro P600

VRAM

40.00x

Need 40.00x Quadro P600

Memory Bandwidth

31.86x

Need 31.86x Quadro P600

NVIDIA RTX A2000

Compute (FP32-eq)

19.50x

Need 19.50x RTX A2000

FP32 Compute

2.44x

Need 2.44x RTX A2000

VRAM

6.67x

Need 6.67x RTX A2000

Memory Bandwidth

7.08x

Need 7.08x RTX A2000

Pricing

Price Type	Bow IPU	MI300X	Quadro P600	RTX A2000	A100 SXM
CAPEX (Street Price)	—	$35,000	—	—	$15,000
OPEX (per hour)	—	$10.40/hr	$0.05/hr	$0.04/hr	$4.05/hr
Price per TFLOPs (FP32-eq)	—	$54	—	—	$96