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 MI100	AMD MI325X	Cerebras WSE-3	NVIDIA GeForce RTX 4090
Architecture	CDNA	CDNA 3.5	Wafer Scale Engine	Ada Lovelace
Form Factor	PCIe	OAM	—	PCIe
VRAM	32 GB	256 GB	44 GB	24 GB
Memory Bandwidth	1,228 GB/s	6,000 GB/s	—	1,008 GB/s
TFLOPs (FP32)	23.1	175	—	82.6
TFLOPs (FP16)	184.6	1,400	—	165.2
TFLOPs	23.1	1,400	—	165.2
TFLOPs (FP8)	—	2,800	—	—
TDP	300 W	800 W	23000 W	450 W
Launch Date	Nov 2020	Oct 2024	Mar 2024	Oct 2022

Efficiency Metrics

Metric	MI100	MI325X	WSE-3	GeForce RTX 4090
TFLOPs per Watt (FP32-eq)	0.08	0.88	—	0.18
Memory Bandwidth per GB	38.4 GB/s	23.4 GB/s	—	42.0 GB/s

Performance Equivalence

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

To match 1x AMD MI100

AMD MI325X

Compute (FP32-eq)

0.03x

MI325X is 30.30x faster

FP32 Compute

0.13x

MI325X is 7.58x faster

VRAM

0.13x

MI325X has 8.00x more

Memory Bandwidth

0.20x

MI325X has 4.89x more

Cerebras WSE-3

VRAM

0.73x

WSE-3 has 1.38x more

NVIDIA GeForce RTX 4090

Compute (FP32-eq)

0.28x

GeForce RTX 4090 is 3.58x faster

FP32 Compute

0.28x

GeForce RTX 4090 is 3.58x faster

VRAM

1.33x

Need 1.33x GeForce RTX 4090

Memory Bandwidth

1.22x

Need 1.22x GeForce RTX 4090

To match 1x AMD MI325X

AMD MI100

Compute (FP32-eq)

30.30x

Need 30.30x MI100

FP32 Compute

7.58x

Need 7.58x MI100

VRAM

8.00x

Need 8.00x MI100

Memory Bandwidth

4.89x

Need 4.89x MI100

Cerebras WSE-3

VRAM

5.82x

Need 5.82x WSE-3

NVIDIA GeForce RTX 4090

Compute (FP32-eq)

8.47x

Need 8.47x GeForce RTX 4090

FP32 Compute

2.12x

Need 2.12x GeForce RTX 4090

VRAM

10.67x

Need 10.67x GeForce RTX 4090

Memory Bandwidth

5.95x

Need 5.95x GeForce RTX 4090

To match 1x Cerebras WSE-3

AMD MI100

VRAM

1.38x

Need 1.38x MI100

AMD MI325X

VRAM

0.17x

MI325X has 5.82x more

NVIDIA GeForce RTX 4090

VRAM

1.83x

Need 1.83x GeForce RTX 4090

To match 1x NVIDIA GeForce RTX 4090

AMD MI100

Compute (FP32-eq)

3.58x

Need 3.58x MI100

FP32 Compute

3.58x

Need 3.58x MI100

VRAM

0.75x

MI100 has 1.33x more

Memory Bandwidth

0.82x

MI100 has 1.22x more

AMD MI325X

Compute (FP32-eq)

0.12x

MI325X is 8.47x faster

FP32 Compute

0.47x

MI325X is 2.12x faster

VRAM

0.09x

MI325X has 10.67x more

Memory Bandwidth

0.17x

MI325X has 5.95x more

Cerebras WSE-3

VRAM

0.55x

WSE-3 has 1.83x more

Pricing

Price Type	MI100	MI325X	WSE-3	GeForce RTX 4090
CAPEX (Street Price)	—	$45,000	—	—
OPEX (per hour)	—	—	—	—
Price per TFLOPs (FP32-eq)	—	$64	—	—