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	Meta MTIA v1	NVIDIA GeForce RTX 5080	NVIDIA GB200 NVL72	Etched Sohu
Architecture	Meta Training & Inference Accelerator	Blackwell	Blackwell	Transformer ASIC
Form Factor	—	PCIe	Custom	PCIe
VRAM	128 GB	16 GB	13824 GB	144 GB
Memory Bandwidth	—	960 GB/s	130,000 GB/s	4,800 GB/s
TFLOPs (FP32)	—	56.28	6,480	10,000
TFLOPs (FP16)	—	112.6	360,000	—
TFLOPs	—	112.6	360,000	10,000
TFLOPs (FP8)	—	—	—	—
TDP	400 W	360 W	120000 W	700 W
Launch Date	May 2023	Jan 2025	Mar 2024	Jun 2024

Efficiency Metrics

Metric	MTIA v1	GeForce RTX 5080	GB200 NVL72	Sohu
TFLOPs per Watt (FP32-eq)	—	0.16	1.50	14.29
Memory Bandwidth per GB	—	60.0 GB/s	9.4 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 Meta MTIA v1

NVIDIA GeForce RTX 5080

VRAM

8.00x

Need 8.00x GeForce RTX 5080

NVIDIA GB200 NVL72

VRAM

0.01x

GB200 NVL72 has 108.00x more

Etched Sohu

VRAM

0.89x

Sohu has 1.13x more

To match 1x NVIDIA GeForce RTX 5080

Meta MTIA v1

VRAM

0.13x

MTIA v1 has 8.00x more

NVIDIA GB200 NVL72

Compute (FP32-eq)

0.00x

GB200 NVL72 is 3197.16x faster

FP32 Compute

0.01x

GB200 NVL72 is 115.14x faster

VRAM

0.00x

GB200 NVL72 has 864.00x more

Memory Bandwidth

0.01x

GB200 NVL72 has 135.42x more

Etched Sohu

Compute (FP32-eq)

0.01x

Sohu is 177.62x faster

FP32 Compute

0.01x

Sohu is 177.68x faster

VRAM

0.11x

Sohu has 9.00x more

Memory Bandwidth

0.20x

Sohu has 5.00x more

To match 1x NVIDIA GB200 NVL72

Meta MTIA v1

VRAM

108.00x

Need 108.00x MTIA v1

NVIDIA GeForce RTX 5080

Compute (FP32-eq)

3197.16x

Need 3197.16x GeForce RTX 5080

FP32 Compute

115.14x

Need 115.14x GeForce RTX 5080

VRAM

864.00x

Need 864.00x GeForce RTX 5080

Memory Bandwidth

135.42x

Need 135.42x GeForce RTX 5080

Etched Sohu

Compute (FP32-eq)

18.00x

Need 18.00x Sohu

FP32 Compute

0.65x

Sohu is 1.54x faster

VRAM

96.00x

Need 96.00x Sohu

Memory Bandwidth

27.08x

Need 27.08x Sohu

To match 1x Etched Sohu

Meta MTIA v1

VRAM

1.13x

Need 1.13x MTIA v1

NVIDIA GeForce RTX 5080

Compute (FP32-eq)

177.62x

Need 177.62x GeForce RTX 5080

FP32 Compute

177.68x

Need 177.68x GeForce RTX 5080

VRAM

9.00x

Need 9.00x GeForce RTX 5080

Memory Bandwidth

5.00x

Need 5.00x GeForce RTX 5080

NVIDIA GB200 NVL72

Compute (FP32-eq)

0.06x

GB200 NVL72 is 18.00x faster

FP32 Compute

1.54x

Need 1.54x GB200 NVL72

VRAM

0.01x

GB200 NVL72 has 96.00x more

Memory Bandwidth

0.04x

GB200 NVL72 has 27.08x more

Pricing

Price Type	MTIA v1	GeForce RTX 5080	GB200 NVL72	Sohu
CAPEX (Street Price)	—	—	—	—
OPEX (per hour)	—	—	—	—
Price per TFLOPs (FP32-eq)	—	—	—	—