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	NVIDIA L40S	Cerebras WSE-3	NVIDIA RTX 6000 Ada Generation	AMD MI300A
Architecture	Ada Lovelace	Wafer Scale Engine	Ada Lovelace	CDNA 3
Form Factor	PCIe	—	PCIe	OAM
VRAM	48 GB	44 GB	48 GB	128 GB
Memory Bandwidth	864 GB/s	—	960 GB/s	5,300 GB/s
TFLOPs (FP32)	91.6	—	91.1	122.6
TFLOPs (FP16)	733	—	182.2	980.6
TFLOPs	733	—	182.2	980.6
TFLOPs (FP8)	1,466	—	—	—
TDP	350 W	23000 W	300 W	760 W
Launch Date	Oct 2023	Mar 2024	Dec 2022	Dec 2023

Efficiency Metrics

Metric	L40S	WSE-3	RTX 6000 Ada Generation	MI300A
TFLOPs per Watt (FP32-eq)	1.05	—	0.30	0.65
Memory Bandwidth per GB	18.0 GB/s	—	20.0 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 NVIDIA L40S

Cerebras WSE-3

VRAM

1.09x

Need 1.09x WSE-3

NVIDIA RTX 6000 Ada Generation

Compute (FP32-eq)

4.02x

Need 4.02x RTX 6000 Ada Generation

FP32 Compute

1.01x

Need 1.01x RTX 6000 Ada Generation

VRAM

1.00x

RTX 6000 Ada Generation has 1.00x more

Memory Bandwidth

0.90x

RTX 6000 Ada Generation has 1.11x more

AMD MI300A

Compute (FP32-eq)

0.75x

MI300A is 1.34x faster

FP32 Compute

0.75x

MI300A is 1.34x faster

VRAM

0.38x

MI300A has 2.67x more

Memory Bandwidth

0.16x

MI300A has 6.13x more

To match 1x Cerebras WSE-3

NVIDIA L40S

VRAM

0.92x

L40S has 1.09x more

NVIDIA RTX 6000 Ada Generation

VRAM

0.92x

RTX 6000 Ada Generation has 1.09x more

AMD MI300A

VRAM

0.34x

MI300A has 2.91x more

To match 1x NVIDIA RTX 6000 Ada Generation

NVIDIA L40S

Compute (FP32-eq)

0.25x

L40S is 4.02x faster

FP32 Compute

0.99x

L40S is 1.01x faster

VRAM

1.00x

L40S has 1.00x more

Memory Bandwidth

1.11x

Need 1.11x L40S

Cerebras WSE-3

VRAM

1.09x

Need 1.09x WSE-3

AMD MI300A

Compute (FP32-eq)

0.19x

MI300A is 5.38x faster

FP32 Compute

0.74x

MI300A is 1.35x faster

VRAM

0.38x

MI300A has 2.67x more

Memory Bandwidth

0.18x

MI300A has 5.52x more

To match 1x AMD MI300A

NVIDIA L40S

Compute (FP32-eq)

1.34x

Need 1.34x L40S

FP32 Compute

1.34x

Need 1.34x L40S

VRAM

2.67x

Need 2.67x L40S

Memory Bandwidth

6.13x

Need 6.13x L40S

Cerebras WSE-3

VRAM

2.91x

Need 2.91x WSE-3

NVIDIA RTX 6000 Ada Generation

Compute (FP32-eq)

5.38x

Need 5.38x RTX 6000 Ada Generation

FP32 Compute

1.35x

Need 1.35x RTX 6000 Ada Generation

VRAM

2.67x

Need 2.67x RTX 6000 Ada Generation

Memory Bandwidth

5.52x

Need 5.52x RTX 6000 Ada Generation

Pricing

Price Type	L40S	WSE-3	RTX 6000 Ada Generation	MI300A
CAPEX (Street Price)	$10,000	—	—	—
OPEX (per hour)	$1.50/hr	—	—	—
Price per TFLOPs (FP32-eq)	$27	—	—	—