Tesla K40s and Radeon PRO W7900 - video card comparison

Tesla K40s

Radeon PRO W7900

Architecture

Kepler (2012−2018)

RDNA 3.0 (2022)

GPU code name

GK110B

Navi 31

Market segment

Workstation

Workstation

Release date

November 2013

April 2023

Interface

PCIe 3.0 x16

PCIe 4.0 x16

Supplementary power connectors

-

2x 8-pin

Boost clock speed

876 MHz

2495 MHz

Core clock speed

745 MHz

-

Floating-point performance

4,291 gflops

-

Manufacturing process technology

28 nm

5 nm

Number of transistors

7,080 million

57,700 million

Pipelines / CUDA cores

2880

6144

Power consumption (TDP)

245 Watt

295 Watt

Texture fill rate

210.2

958.1

Memory type

GDDR5

GDDR6

Maximum RAM amount

12 GB

48 GB

Memory bus width

384 Bit

384 Bit

Memory clock speed

6008 MHz

18 GB/s

Memory bandwidth

288.4 GB/s

864.0 GB/s

Display Connectors

No outputs

3x DisplayPort 2.1, 1x mini-DisplayPort 2.1

DirectX

12 (11_1)

12 Ultimate (12_2)

CUDA

3.5

-

OpenCL

1.2

2.2

OpenGL

4.6

4.6

Shader Model

5.1

6.7

Vulkan

1.1.126

1.3