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	Groq LPU Inference Engine	Microsoft Maia 100	NVIDIA H200 PCIe
Architecture	Ada Lovelace	TSP (Tensor Streaming Processor)	Maia	Hopper
Form Factor	PCIe	—	—	PCIe
VRAM	48 GB	230 GB	64 GB	141 GB
Memory Bandwidth	864 GB/s	—	—	4,800 GB/s
TFLOPs (FP32)	91.6	—	—	51
TFLOPs (FP16)	733	—	—	1,513
TFLOPs	733	—	700	1,513
TFLOPs (FP8)	1,466	—	—	—
TDP	350 W	300 W	500 W	350 W
Launch Date	Oct 2023	Feb 2024	Nov 2023	Mar 2024

Efficiency Metrics

Metric	L40S	LPU Inference Engine	Maia 100	H200 PCIe
TFLOPs per Watt (FP32-eq)	1.05	—	0.70	2.16
Memory Bandwidth per GB	18.0 GB/s	—	—	34.0 GB/s

Performance Equivalence

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

To match 1x NVIDIA L40S

Groq LPU Inference Engine

VRAM

0.21x

LPU Inference Engine has 4.79x more

Microsoft Maia 100

Compute (FP32-eq)

1.05x

Need 1.05x Maia 100

VRAM

0.75x

Maia 100 has 1.33x more

NVIDIA H200 PCIe

Compute (FP32-eq)

0.48x

H200 PCIe is 2.06x faster

FP32 Compute

1.80x

Need 1.80x H200 PCIe

VRAM

0.34x

H200 PCIe has 2.94x more

Memory Bandwidth

0.18x

H200 PCIe has 5.56x more

To match 1x Groq LPU Inference Engine

NVIDIA L40S

VRAM

4.79x

Need 4.79x L40S

Microsoft Maia 100

VRAM

3.59x

Need 3.59x Maia 100

NVIDIA H200 PCIe

VRAM

1.63x

Need 1.63x H200 PCIe

To match 1x Microsoft Maia 100

NVIDIA L40S

Compute (FP32-eq)

0.95x

L40S is 1.05x faster

VRAM

1.33x

Need 1.33x L40S

Groq LPU Inference Engine

VRAM

0.28x

LPU Inference Engine has 3.59x more

NVIDIA H200 PCIe

Compute (FP32-eq)

0.46x

H200 PCIe is 2.16x faster

VRAM

0.45x

H200 PCIe has 2.20x more

To match 1x NVIDIA H200 PCIe

NVIDIA L40S

Compute (FP32-eq)

2.06x

Need 2.06x L40S

FP32 Compute

0.56x

L40S is 1.80x faster

VRAM

2.94x

Need 2.94x L40S

Memory Bandwidth

5.56x

Need 5.56x L40S

Groq LPU Inference Engine

VRAM

0.61x

LPU Inference Engine has 1.63x more

Microsoft Maia 100

Compute (FP32-eq)

2.16x

Need 2.16x Maia 100

VRAM

2.20x

Need 2.20x Maia 100

Pricing

Price Type	L40S	LPU Inference Engine	Maia 100	H200 PCIe
CAPEX (Street Price)	$10,000	—	—	—
OPEX (per hour)	$1.50/hr	—	—	—
Price per TFLOPs (FP32-eq)	$27	—	—	—