Tesla T4 and GeForce RTX 4070 AD103 - video card comparison

Tesla T4

GeForce RTX 4070 AD103

Architecture

Turing (2018−2021)

Ada Lovelace

GPU code name

TU104

AD103

Market segment

Workstation

Desktop

Release date

September 2018

March 2024 (recently)

Interface

PCIe 3.0 x16

PCIe 4.0 x16

Supplementary power connectors

None

1x 16-pin

Boost clock speed

1590 MHz

2475 MHz

Core clock speed

585 MHz

Manufacturing process technology

12 nm

5 nm

Number of transistors

13,600 million

45,900 million

Pipelines / CUDA cores

2560

5888

Power consumption (TDP)

70 Watt

200 Watt

Texture fill rate

254.4

455.4

Memory type

GDDR6

GDDR6X

Maximum RAM amount

16 GB

12 GB

Memory bus width

256 Bit

192 Bit

Memory clock speed

10000 MHz

21 GB/s

Memory bandwidth

320.0 GB/s

504.2 GB/s

Display Connectors

No outputs

1x HDMI 2.1, 3x DisplayPort 1.4a

HDMI

DirectX

12 Ultimate (12_1)

12 Ultimate (12_2)

CUDA

7.5

8.9

OpenCL

1.2

3.0

OpenGL

4.6

Shader Model

6.5

6.7

Vulkan

1.2.131

1.3