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 A10	Groq LPU Inference Engine	NVIDIA GeForce RTX 5070	Etched Sohu
Architecture	Ampere	TSP (Tensor Streaming Processor)	Blackwell	Transformer ASIC
Form Factor	PCIe	—	PCIe	PCIe
VRAM	24 GB	230 GB	12 GB	144 GB
Memory Bandwidth	600 GB/s	—	672 GB/s	4,800 GB/s
TFLOPs (FP32)	31.2	—	30.8	10,000
TFLOPs (FP16)	125	—	61.6	—
TFLOPs	125	—	61.6	10,000
TFLOPs (FP8)	—	—	—	—
TDP	150 W	300 W	250 W	700 W
Launch Date	Apr 2021	Feb 2024	Feb 2025	Jun 2024

Efficiency Metrics

Metric	A10	LPU Inference Engine	GeForce RTX 5070	Sohu
TFLOPs per Watt (FP32-eq)	0.42	—	0.12	14.29
Memory Bandwidth per GB	25.0 GB/s	—	56.0 GB/s	33.3 GB/s

Performance Equivalence

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

To match 1x NVIDIA A10

Groq LPU Inference Engine

VRAM

0.10x

LPU Inference Engine has 9.58x more

NVIDIA GeForce RTX 5070

Compute (FP32-eq)

2.03x

Need 2.03x GeForce RTX 5070

FP32 Compute

1.01x

Need 1.01x GeForce RTX 5070

VRAM

2.00x

Need 2.00x GeForce RTX 5070

Memory Bandwidth

0.89x

GeForce RTX 5070 has 1.12x more

Etched Sohu

Compute (FP32-eq)

0.01x

Sohu is 160.00x faster

FP32 Compute

0.00x

Sohu is 320.51x faster

VRAM

0.17x

Sohu has 6.00x more

Memory Bandwidth

0.13x

Sohu has 8.00x more

To match 1x Groq LPU Inference Engine

NVIDIA A10

VRAM

9.58x

Need 9.58x A10

NVIDIA GeForce RTX 5070

VRAM

19.17x

Need 19.17x GeForce RTX 5070

Etched Sohu

VRAM

1.60x

Need 1.60x Sohu

To match 1x NVIDIA GeForce RTX 5070

NVIDIA A10

Compute (FP32-eq)

0.49x

A10 is 2.03x faster

FP32 Compute

0.99x

A10 is 1.01x faster

VRAM

0.50x

A10 has 2.00x more

Memory Bandwidth

1.12x

Need 1.12x A10

Groq LPU Inference Engine

VRAM

0.05x

LPU Inference Engine has 19.17x more

Etched Sohu

Compute (FP32-eq)

0.00x

Sohu is 324.68x faster

FP32 Compute

0.00x

Sohu is 324.68x faster

VRAM

0.08x

Sohu has 12.00x more

Memory Bandwidth

0.14x

Sohu has 7.14x more

To match 1x Etched Sohu

NVIDIA A10

Compute (FP32-eq)

160.00x

Need 160.00x A10

FP32 Compute

320.51x

Need 320.51x A10

VRAM

6.00x

Need 6.00x A10

Memory Bandwidth

8.00x

Need 8.00x A10

Groq LPU Inference Engine

VRAM

0.63x

LPU Inference Engine has 1.60x more

NVIDIA GeForce RTX 5070

Compute (FP32-eq)

324.68x

Need 324.68x GeForce RTX 5070

FP32 Compute

324.68x

Need 324.68x GeForce RTX 5070

VRAM

12.00x

Need 12.00x GeForce RTX 5070

Memory Bandwidth

7.14x

Need 7.14x GeForce RTX 5070

Pricing

Price Type	A10	LPU Inference Engine	GeForce RTX 5070	Sohu
CAPEX (Street Price)	—	—	—	—
OPEX (per hour)	—	—	—	—
Price per TFLOPs (FP32-eq)	—	—	—	—