Radeon Pro 575 and GeForce RTX 3090 - video card comparison

Radeon Pro 575

GeForce RTX 3090

Architecture

GCN 4.0 (2016−2020)

Ampere (2020−2022)

GPU code name

Polaris 20

Ampere GA102

Market segment

Mobile workstation

Desktop

Release date

June 2017

September 2020

Interface

PCIe 3.0 x16

PCIe 4.0 x16

Supplementary power connectors

None

1x 12-pin

Boost clock speed

-

1700 MHz

Core clock speed

1096 MHz

1400 MHz

Floating-point performance

4,506 gflops

-

Manufacturing process technology

14 nm

8 nm

Number of transistors

5,700 million

28,300 million

Pipelines / CUDA cores

2048

10496

Power consumption (TDP)

120 Watt

350 Watt

Texture fill rate

140.3

556.0

Memory type

GDDR5

GDDR6X

Maximum RAM amount

4 GB

24 GB

Memory bus width

256 Bit

384 Bit

Memory clock speed

6780 MHz

19500 MHz

Memory bandwidth

217.0 GB/s

936.2 GB/s

Shared memory

-

-

Display Connectors

No outputs

1x HDMI, 3x DisplayPort

HDMI

-

+

DirectX

12 (12_0)

12 Ultimate (12_2)

CUDA

-

8.5

OpenCL

2.0

2.0

OpenGL

4.6

4.6

Shader Model

6.4

6.5

Vulkan

1.2.131

1.2