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 128 XPUs • 4 selected

Alibaba

Hanguang 800

AMD

MI100

23.1 TFLOPs

AMD

MI210

181 TFLOPs

AMD

MI250X

383 TFLOPs

AMD

MI300X

1,307 TFLOPs

AMD

MI325X

1,400 TFLOPs

AMD

MI350X

2,100 TFLOPs

AMD

MI355X

5,300 TFLOPs

AMD

Radeon Pro V520

23.04 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

FuriosaAI

RNGD (Renegade)

256 TFLOPs

FuriosaAI

Warboy

Google

TPU v4

275 TFLOPs

Google

TPU v5e

197 TFLOPs

Google

TPU v5p

459 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 RTX 6000 Blackwell	NVIDIA GeForce RTX 5070 Ti	NVIDIA GeForce RTX 3090	NVIDIA RTX 4000
Architecture	Blackwell	Blackwell	Ampere	Turing
Form Factor	PCIe	PCIe	PCIe	PCIe
VRAM	96 GB	16 GB	24 GB	8 GB
Memory Bandwidth	1,536 GB/s	896 GB/s	936 GB/s	416 GB/s
TFLOPs (FP32)	181	47	35.6	7.1
TFLOPs (FP16)	—	—	—	—
TFLOPs	362	133	71	57.6
TFLOPs (FP8)	—	—	—	—
TDP	300 W	300 W	350 W	160 W
Launch Date	Jan 2025	Feb 2025	Sep 2020	Nov 2018

Efficiency Metrics

Metric	RTX 6000 Blackwell	GeForce RTX 5070 Ti	GeForce RTX 3090	RTX 4000
TFLOPs per Watt (FP32-eq)	0.60	0.22	0.10	0.18
Memory Bandwidth per GB	16.0 GB/s	56.0 GB/s	39.0 GB/s	52.0 GB/s

Performance Equivalence

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

To match 1x NVIDIA RTX 6000 Blackwell

NVIDIA GeForce RTX 5070 Ti

Compute (FP32-eq)

2.72x

Need 2.72x GeForce RTX 5070 Ti

FP32 Compute

3.85x

Need 3.85x GeForce RTX 5070 Ti

VRAM

6.00x

Need 6.00x GeForce RTX 5070 Ti

Memory Bandwidth

1.71x

Need 1.71x GeForce RTX 5070 Ti

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

5.10x

Need 5.10x GeForce RTX 3090

FP32 Compute

5.08x

Need 5.08x GeForce RTX 3090

VRAM

4.00x

Need 4.00x GeForce RTX 3090

Memory Bandwidth

1.64x

Need 1.64x GeForce RTX 3090

NVIDIA RTX 4000

Compute (FP32-eq)

6.28x

Need 6.28x RTX 4000

FP32 Compute

25.49x

Need 25.49x RTX 4000

VRAM

12.00x

Need 12.00x RTX 4000

Memory Bandwidth

3.69x

Need 3.69x RTX 4000

To match 1x NVIDIA GeForce RTX 5070 Ti

NVIDIA RTX 6000 Blackwell

Compute (FP32-eq)

0.37x

RTX 6000 Blackwell is 2.72x faster

FP32 Compute

0.26x

RTX 6000 Blackwell is 3.85x faster

VRAM

0.17x

RTX 6000 Blackwell has 6.00x more

Memory Bandwidth

0.58x

RTX 6000 Blackwell has 1.71x more

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

1.87x

Need 1.87x GeForce RTX 3090

FP32 Compute

1.32x

Need 1.32x GeForce RTX 3090

VRAM

0.67x

GeForce RTX 3090 has 1.50x more

Memory Bandwidth

0.96x

GeForce RTX 3090 has 1.04x more

NVIDIA RTX 4000

Compute (FP32-eq)

2.31x

Need 2.31x RTX 4000

FP32 Compute

6.62x

Need 6.62x RTX 4000

VRAM

2.00x

Need 2.00x RTX 4000

Memory Bandwidth

2.15x

Need 2.15x RTX 4000

To match 1x NVIDIA GeForce RTX 3090

NVIDIA RTX 6000 Blackwell

Compute (FP32-eq)

0.20x

RTX 6000 Blackwell is 5.10x faster

FP32 Compute

0.20x

RTX 6000 Blackwell is 5.08x faster

VRAM

0.25x

RTX 6000 Blackwell has 4.00x more

Memory Bandwidth

0.61x

RTX 6000 Blackwell has 1.64x more

NVIDIA GeForce RTX 5070 Ti

Compute (FP32-eq)

0.53x

GeForce RTX 5070 Ti is 1.87x faster

FP32 Compute

0.76x

GeForce RTX 5070 Ti is 1.32x faster

VRAM

1.50x

Need 1.50x GeForce RTX 5070 Ti

Memory Bandwidth

1.04x

Need 1.04x GeForce RTX 5070 Ti

NVIDIA RTX 4000

Compute (FP32-eq)

1.23x

Need 1.23x RTX 4000

FP32 Compute

5.01x

Need 5.01x RTX 4000

VRAM

3.00x

Need 3.00x RTX 4000

Memory Bandwidth

2.25x

Need 2.25x RTX 4000

To match 1x NVIDIA RTX 4000

NVIDIA RTX 6000 Blackwell

Compute (FP32-eq)

0.16x

RTX 6000 Blackwell is 6.28x faster

FP32 Compute

0.04x

RTX 6000 Blackwell is 25.49x faster

VRAM

0.08x

RTX 6000 Blackwell has 12.00x more

Memory Bandwidth

0.27x

RTX 6000 Blackwell has 3.69x more

NVIDIA GeForce RTX 5070 Ti

Compute (FP32-eq)

0.43x

GeForce RTX 5070 Ti is 2.31x faster

FP32 Compute

0.15x

GeForce RTX 5070 Ti is 6.62x faster

VRAM

0.50x

GeForce RTX 5070 Ti has 2.00x more

Memory Bandwidth

0.46x

GeForce RTX 5070 Ti has 2.15x more

NVIDIA GeForce RTX 3090

Compute (FP32-eq)

0.81x

GeForce RTX 3090 is 1.23x faster

FP32 Compute

0.20x

GeForce RTX 3090 is 5.01x faster

VRAM

0.33x

GeForce RTX 3090 has 3.00x more

Memory Bandwidth

0.44x

GeForce RTX 3090 has 2.25x more

Pricing

Price Type	RTX 6000 Blackwell	GeForce RTX 5070 Ti	GeForce RTX 3090	RTX 4000
CAPEX (Street Price)	—	—	—	—
OPEX (per hour)	$1.26/hr	$0.12/hr	$0.11/hr	$0.34/hr
Price per TFLOPs (FP32-eq)	—	—	—	—