Tesla M40 GPU Overview

NVIDIA Tesla M40 GPU graphics card was released on November 10, 2015. The card is designed on the Maxwell 2.0 architecture, the minimum clock speed is 948 MHz. There is an acceleration system and the ability to work in turbo mode or when overclocking at 1114 MHz. More information on compatibility, dimensions, technology, and scope of use is below.

Tesla M40

Technical parameters

The following information shows the power specifications of NVIDIA Tesla M40. A shorter manufacturing process of the chip is preferable. The core clock frequency characterizes the performance (the higher it is, the faster the performance of the video card). The number of transistors is directly related to the speed of calculations.

Core clock frequency:

948 MHz

Overclocking speed:

1114 MHz

Number of transistors:

8,000 million

Process technology:

28 nm

Number of texels processed per second:


Floating point operations:

6,844 gflops

Conveyors / CUDA Cores:


Rated Thermal Power:

250 Watts

Interfaces and Compatibility

There are many sizes of PC and notebook cases, so it is important to correctly assess your graphics card length as well as the type of interface to connect it when buying. The M40 has the following parameters:


PCIe 3.0 x16


267 mm

Optional power supply:

1x 6-pin + 1x 8-pin


Built-in memory allows you to store data when performing calculations. The higher its value, the more powerful and fast the graphics card is.

Memory type:


Maximum RAM capacity:

12 Gb

Memory bus capacity:

384 Bit

Memory Frequency:

6008 MHz

Memory bandwidth:

288.4 Gb/s

Port and display support

The main part of modern graphics cards is equipped with several types of connections and optional ports such as HDMI and DVI. You should be aware of these features so that you don't have any difficulties when connecting your graphics card to a monitor or other peripheral devices.

Display Connections:

No outputs


Tesla M40 is a powerful graphics accelerator, which is suitable for deep learning of neural networks. It reduces the training time of neural networks by several days. Performance is scalable with GPUDirect support. The card is suitable for 24/7 operation.

In addition, M40 can be used to organize remote workstations, suitable for resource-intensive applications for 3D graphics and rendering.

Cloud4Y offers you to rent a server equipped with NVIDIA Tesla M40 GPU and other models.

Is useful article?
author: John
published: 03/14/2022
Last articles
Scroll up!