GPU Driver

Overview

Before using a GPU-accelerated ECS, make sure that a GPU driver has been installed on the ECS for GPU acceleration.

GPU-accelerated ECSs support GRID and Tesla drivers.

  • To use graphics acceleration, such as OpenGL, DirectX, or Vulkan, install a GRID driver and separately purchase and configure a GRID license. The GRID driver with a vDWS license also supports CUDA for both computing and graphics acceleration.

    • A graphics-accelerated (G series) ECS created using a Windows public image has had a GRID driver of a specified version installed by default, but the GRID license must be purchased and configured separately. Before using such an ECS, check whether the desired driver has been installed on it and whether the version of the installed driver meets service requirements.

    • A graphics-accelerated (G series) ECS created using a Linux public image does not have a GRID driver installed by default. To install a GRID driver, see Manually Installing a GRID Driver on a GPU-accelerated ECS.

    • To install a GRID driver on a GPU-accelerated ECS created using a private image, see Manually Installing a GRID Driver on a GPU-accelerated ECS.

  • To use computing acceleration, install a Tesla driver.

Table 1 Acceleration supported by GPU drivers

Driver

License

CUDA

OpenGL

DirectX

Vulkan

Application Scenario

Description

GRID

Required

Supported

Supported

Supported

Supported

3D rendering, graphics workstation, and game acceleration

The GRID driver must be paid and requires a license to accelerate graphics and image applications.

Tesla

Not required

Supported

Not supported

Not supported

Not supported

Scientific computing, deep learning training, and inference

The Tesla driver is downloaded free of charge and usually used with NVIDIA CUDA SDKs to accelerate general computing applications.

last updated: 2024-12-17 08:54 UTC - commit: 60d73b5e88a7b9e2d4d3346634d7fac6006f7365