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 72 XPUs • 4 selected

Alibaba

Hanguang 800

AMD

MI100

23.1 TFLOPs

AMD

MI210

181 TFLOPs

AMD

MI250X

383 TFLOPs

AMD

MI300A

980.6 TFLOPs

AMD

MI300X

1,307 TFLOPs

AMD

MI325X

1,400 TFLOPs

AMD

MI350X

2,100 TFLOPs

AMD

Radeon PRO W7900

122 TFLOPs

AMD

Radeon RX 7900 XT

104 TFLOPs

AMD

Radeon RX 7900 XTX

122 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

Etched

Sohu

10,000 TFLOPs

FuriosaAI

Warboy

Google

TPU v4

275 TFLOPs

Google

TPU v5e

197 TFLOPs

Google

TPU v5p

459 TFLOPs

Google

TPU v6e (Trillium)

918 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	AMD MI250X	FuriosaAI Warboy	NVIDIA GeForce RTX 5070	AMD MI300A
Architecture	CDNA 2	FuriosaAI NPU	Blackwell	CDNA 3
Form Factor	OAM	—	PCIe	OAM
VRAM	128 GB	32 GB	12 GB	128 GB
Memory Bandwidth	3,277 GB/s	—	672 GB/s	5,300 GB/s
TFLOPs (FP32)	47.9	—	30.8	122.6
TFLOPs (FP16)	383	—	61.6	980.6
TFLOPs	383	—	61.6	980.6
TFLOPs (FP8)	—	—	—	—
TDP	560 W	80 W	250 W	760 W
Launch Date	Nov 2021	Oct 2021	Feb 2025	Dec 2023

Efficiency Metrics

Metric	MI250X	Warboy	GeForce RTX 5070	MI300A
TFLOPs per Watt (FP32-eq)	0.34	—	0.12	0.65
Memory Bandwidth per GB	25.6 GB/s	—	56.0 GB/s	41.4 GB/s

Performance Equivalence

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

To match 1x AMD MI250X

FuriosaAI Warboy

VRAM

4.00x

Need 4.00x Warboy

NVIDIA GeForce RTX 5070

Compute (FP32-eq)

6.22x

Need 6.22x GeForce RTX 5070

FP32 Compute

1.56x

Need 1.56x GeForce RTX 5070

VRAM

10.67x

Need 10.67x GeForce RTX 5070

Memory Bandwidth

4.88x

Need 4.88x GeForce RTX 5070

AMD MI300A

Compute (FP32-eq)

0.39x

MI300A is 2.56x faster

FP32 Compute

0.39x

MI300A is 2.56x faster

VRAM

1.00x

MI300A has 1.00x more

Memory Bandwidth

0.62x

MI300A has 1.62x more

To match 1x FuriosaAI Warboy

AMD MI250X

VRAM

0.25x

MI250X has 4.00x more

NVIDIA GeForce RTX 5070

VRAM

2.67x

Need 2.67x GeForce RTX 5070

AMD MI300A

VRAM

0.25x

MI300A has 4.00x more

To match 1x NVIDIA GeForce RTX 5070

AMD MI250X

Compute (FP32-eq)

0.16x

MI250X is 6.22x faster

FP32 Compute

0.64x

MI250X is 1.56x faster

VRAM

0.09x

MI250X has 10.67x more

Memory Bandwidth

0.21x

MI250X has 4.88x more

FuriosaAI Warboy

VRAM

0.38x

Warboy has 2.67x more

AMD MI300A

Compute (FP32-eq)

0.06x

MI300A is 15.92x faster

FP32 Compute

0.25x

MI300A is 3.98x faster

VRAM

0.09x

MI300A has 10.67x more

Memory Bandwidth

0.13x

MI300A has 7.89x more

To match 1x AMD MI300A

AMD MI250X

Compute (FP32-eq)

2.56x

Need 2.56x MI250X

FP32 Compute

2.56x

Need 2.56x MI250X

VRAM

1.00x

MI250X has 1.00x more

Memory Bandwidth

1.62x

Need 1.62x MI250X

FuriosaAI Warboy

VRAM

4.00x

Need 4.00x Warboy

NVIDIA GeForce RTX 5070

Compute (FP32-eq)

15.92x

Need 15.92x GeForce RTX 5070

FP32 Compute

3.98x

Need 3.98x GeForce RTX 5070

VRAM

10.67x

Need 10.67x GeForce RTX 5070

Memory Bandwidth

7.89x

Need 7.89x GeForce RTX 5070

Pricing

Price Type	MI250X	Warboy	GeForce RTX 5070	MI300A
CAPEX (Street Price)	$12,000	—	—	—
OPEX (per hour)	$2.00/hr	—	—	—
Price per TFLOPs (FP32-eq)	$63	—	—	—