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 Habana Gaudi 3	AMD MI325X	NVIDIA RTX 4000 SFF Ada Generation	NVIDIA Quadro M60
Architecture	Gaudi Gen3	CDNA 3.5	Ada Lovelace	Maxwell
Form Factor	OAM	OAM	PCIe	PCIe
VRAM	128 GB	256 GB	20 GB	16 GB
Memory Bandwidth	3,700 GB/s	6,000 GB/s	360 GB/s	320 GB/s
TFLOPs (FP32)	—	207	26.7	4.825
TFLOPs (FP16)	—	1,400	—	—
TFLOPs	1,835	1,400	52	4.825
TFLOPs (FP8)	3,670	2,800	—	—
TDP	900 W	800 W	70 W	300 W
Launch Date	Apr 2024	Oct 2024	Apr 2023	Mar 2016

Efficiency Metrics

Metric	Gaudi 3	MI325X	RTX 4000 SFF Ada Generation	Quadro M60
TFLOPs per Watt (FP32-eq)	1.02	0.88	0.37	0.02
Memory Bandwidth per GB	28.9 GB/s	23.4 GB/s	18.0 GB/s	20.0 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

AMD MI325X

Compute (FP32-eq)

1.31x

Need 1.31x MI325X

VRAM

0.50x

MI325X has 2.00x more

Memory Bandwidth

0.62x

MI325X has 1.62x more

NVIDIA RTX 4000 SFF Ada Generation

Compute (FP32-eq)

35.29x

Need 35.29x RTX 4000 SFF Ada Generation

VRAM

6.40x

Need 6.40x RTX 4000 SFF Ada Generation

Memory Bandwidth

10.28x

Need 10.28x RTX 4000 SFF Ada Generation

NVIDIA Quadro M60

Compute (FP32-eq)

190.16x

Need 190.16x Quadro M60

VRAM

8.00x

Need 8.00x Quadro M60

Memory Bandwidth

11.56x

Need 11.56x Quadro M60

To match 1x AMD MI325X

Intel Habana Gaudi 3

Compute (FP32-eq)

0.76x

Gaudi 3 is 1.31x faster

VRAM

2.00x

Need 2.00x Gaudi 3

Memory Bandwidth

1.62x

Need 1.62x Gaudi 3

NVIDIA RTX 4000 SFF Ada Generation

Compute (FP32-eq)

26.92x

Need 26.92x RTX 4000 SFF Ada Generation

FP32 Compute

7.75x

Need 7.75x RTX 4000 SFF Ada Generation

VRAM

12.80x

Need 12.80x RTX 4000 SFF Ada Generation

Memory Bandwidth

16.67x

Need 16.67x RTX 4000 SFF Ada Generation

NVIDIA Quadro M60

Compute (FP32-eq)

145.08x

Need 145.08x Quadro M60

FP32 Compute

42.90x

Need 42.90x Quadro M60

VRAM

16.00x

Need 16.00x Quadro M60

Memory Bandwidth

18.75x

Need 18.75x Quadro M60

To match 1x NVIDIA RTX 4000 SFF Ada Generation

Intel Habana Gaudi 3

Compute (FP32-eq)

0.03x

Gaudi 3 is 35.29x faster

VRAM

0.16x

Gaudi 3 has 6.40x more

Memory Bandwidth

0.10x

Gaudi 3 has 10.28x more

AMD MI325X

Compute (FP32-eq)

0.04x

MI325X is 26.92x faster

FP32 Compute

0.13x

MI325X is 7.75x faster

VRAM

0.08x

MI325X has 12.80x more

Memory Bandwidth

0.06x

MI325X has 16.67x more

NVIDIA Quadro M60

Compute (FP32-eq)

5.39x

Need 5.39x Quadro M60

FP32 Compute

5.53x

Need 5.53x Quadro M60

VRAM

1.25x

Need 1.25x Quadro M60

Memory Bandwidth

1.13x

Need 1.13x Quadro M60

To match 1x NVIDIA Quadro M60

Intel Habana Gaudi 3

Compute (FP32-eq)

0.01x

Gaudi 3 is 190.16x faster

VRAM

0.13x

Gaudi 3 has 8.00x more

Memory Bandwidth

0.09x

Gaudi 3 has 11.56x more

AMD MI325X

Compute (FP32-eq)

0.01x

MI325X is 145.08x faster

FP32 Compute

0.02x

MI325X is 42.90x faster

VRAM

0.06x

MI325X has 16.00x more

Memory Bandwidth

0.05x

MI325X has 18.75x more

NVIDIA RTX 4000 SFF Ada Generation

Compute (FP32-eq)

0.19x

RTX 4000 SFF Ada Generation is 5.39x faster

FP32 Compute

0.18x

RTX 4000 SFF Ada Generation is 5.53x faster

VRAM

0.80x

RTX 4000 SFF Ada Generation has 1.25x more

Memory Bandwidth

0.89x

RTX 4000 SFF Ada Generation has 1.13x more

Pricing

Price Type	Gaudi 3	MI325X	RTX 4000 SFF Ada Generation	Quadro M60
CAPEX (Street Price)	$15,000	$45,000	—	—
OPEX (per hour)	$1.20/hr	$36.92/hr	$0.15/hr	$0.75/hr
Price per TFLOPs (FP32-eq)	$16	$64	—	—