How to deal with the problem that the Nvidia driver is installed successfully but the display driver is not installed

The content of this article is about how to deal with the problem that the Nvidia driver is installed successfully but the display driver is not installed. It has certain reference value. Friends in need can refer to it. I hope it will be helpful to you.

GPU instance FAQ

The Nvidia driver is installed successfully, but nvidia-smi is invalid and the driver is not installed.

Problem analysis: The kernel-devel and kernel versions are inconsistent, resulting in driver compilation errors during the installation of driver rpm.

Solution: Check the kernel version number, download the corresponding kernel-devel package online, and run rpm –qa | grep kernel in the instance to check whether the versions are consistent. After confirming the version, reinstall the driver.

The Windows instances of gn5 and gn5i were disconnected when installing the official driver.

Problem analysis: Nvidia has restricted graphics functions and requires specific drivers and certified licenses.

Solution: Currently, the test version of the driver is only available in specific availability zones (North China 5 and North China 2 Availability Zone E) to ensure continuous network access. The test driver is valid for 7-100 days. A certified license is required after 100 days.

The following are test driver download addresses for different versions of Windows:

Windows Server 2016

Window Server 2008 or 2012

Windows graphics does not support features such as DirectX.

Problem analysis: For Windows systems, after the GPU driver installation takes effect, the remote connection (RDP) protocol that comes with Windows does not support DirectX, OpenGL and other related applications.

Solution: You need to install the VNC service and client yourself, or other supported protocols, such as PCOIP, XenDeskop HDX 3D, etc.

The ga1 instance uses the cloud market to pre-install the AMD driver image, but when entering the Windows Device Manager, it shows that the GPU driver failed to load normally.

After logging in to the instance, open Device Manager, right-click the GPU device name, click Update Driver Software, and then select Automatically search for updated driver software. After the Windows system completes the device driver update, restart the system.

Using the cloud market to pre-install the AMD driver image, a blue screen occurs when starting the instance.

Alibaba Cloud will update the driver included in the pre-installed driver image in the cloud market to fix this problem. If you are using an older version of the image, you can avoid this problem by updating the GPU driver.

If a blue screen occurs and the instance cannot be started, you must enter safe mode, uninstall the previous driver, start the Windows system normally, and then install the new version of the GPU driver.

Does the GPU instance support the Android emulator?

not support.

Can I change the configuration of the GPU cloud server?

The configuration of GPU cloud server specification families with local disks, such as ga1 and gn5, cannot be changed. For GPU cloud server instances without local disks, you can change their specifications.

Does the pay-as-you-go GPU cloud server support the no-charge function for downtime?

GPU cloud server specification family without local disk, including gn4 (Nvidia M40), gn5i (Nvidia P4) and gn6v (Nvidia V100), supports pay-as-you-go instances without any charge for downtime Function. For more information about this feature, please refer to No charge for downtime of pay-as-you-go instances.

How to view GPU monitoring data?

You can view GPU monitoring data through the cloud monitoring console or API. For details, see GPU monitoring data query.

The above is the detailed content of How to deal with the problem that the Nvidia driver is installed successfully but the display driver is not installed. For more information, please follow other related articles on the PHP Chinese website!