Tesla S2050 and Tesla K40t - video card comparison

Tesla S2050

Tesla K40t

Architecture

Fermi (2010−2014)

Kepler (2012−2018)

GPU code name

GF100

GK110B

Market segment

Workstation

Workstation

Release date

July 2011

November 2013

Interface

PCIe 2.0 x16

PCIe 3.0 x16

Boost clock speed

-

876 MHz

Core clock speed

574 MHz

745 MHz

Floating-point performance

4x 1,027.7 gflops

-

Manufacturing process technology

40 nm

28 nm

Number of transistors

3,100 million

7,080 million

Pipelines / CUDA cores

448

2880

Power consumption (TDP)

900 Watt

245 Watt

Texture fill rate

32.14

210.2

Memory type

GDDR5

GDDR5

Maximum RAM amount

3 GB

12 GB

Memory bus width

384 Bit

384 Bit

Memory clock speed

3092 MHz

6008 MHz

Memory bandwidth

148.4 GB/s

288.4 GB/s

Display Connectors

No outputs

No outputs

DirectX

12 (11_0)

12 (11_1)

CUDA

2.0

3.5

OpenCL

1.1

1.2

OpenGL

4.6

4.6

Shader Model

5.1

5.1

Vulkan

N/A

1.1.126