TU102 and GeForce RTX 3070 - video card comparison

TU102

GeForce RTX 3070

Architecture

Turing (2018−2021)

Ampere (2020−2022)

GPU code name

TU102

Ampere GA104

Market segment

Desktop

Desktop

Release date

-

September 2020

Interface

-

PCIe 4.0 x16

Supplementary power connectors

-

1x 12-pin

Boost clock speed

-

1725 MHz

Core clock speed

-

1500 MHz

Manufacturing process technology

12 nm

8 nm

Number of transistors

18,600 million

17,400 million

Pipelines / CUDA cores

4608

5888

Power consumption (TDP)

-

220 Watt

Texture fill rate

-

317.4

Memory type

-

GDDR6

Maximum RAM amount

-

8 GB

Memory bus width

-

256 Bit

Memory clock speed

-

14000 MHz

Memory bandwidth

-

448.0 GB/s

Shared memory

-

-

Display Connectors

-

1x HDMI, 3x DisplayPort

HDMI

-

+

DirectX

12 Ultimate (12_2)

12 Ultimate (12_2)

CUDA

7.5

8.5

OpenCL

3.0

2.0

OpenGL

4.6

4.6

Shader Model

6.7

6.5

Vulkan

1.3

1.2

NVENC

7th Gen

-

PureVideo HD

VP10

-