When Ubuntu fails to recognize an NVIDIA GeForce RTX 4090 during boot, the issue typically stems from missing proprietary drivers, Secure Boot restrictions, or kernel module conflicts. This guide outlines diagnostic steps and configuration adjustments to restore GPU functionality without relying on unverified package names.
Initial Diagnostics
Before attempting complex fixes, verify whether the system detects the hardware at all. Open a terminal and runlspci | grep -i nvidia. If the output lists your device, the kernel sees the card physically. If it returns nothing, check BIOS settings for PCIe slot configuration or try a different slot. Next, check for driver loading issues by running dmesg | grep -i nvidia. Look for error messages related to firmware loading failures or permission denied errors. These logs provide critical clues about why the operating system ignores the GPU despite its physical presence.
Another essential command is nvidia-smi. If this command is not found, the proprietary drivers are likely absent. If the command runs but shows no connected GPUs, the driver is installed but failing to initialize the hardware. This distinction helps determine if you need to install software or troubleshoot a runtime conflict. Ensure that the open-source Nouveau driver is not interfering. You can check active modules with lsmod | grep nouveau. If Nouveau is loaded, it often blocks NVIDIA drivers from attaching to the card.
Secure Boot and Kernel Modules
Secure Boot is a frequent culprit for GPU recognition failures on modern Linux distributions. It prevents unsigned kernel modules from loading. Since official NVIDIA drivers are not always signed with keys accepted by default Secure Boot configurations, the kernel may reject them silently. Enter your system BIOS/UEFI settings and disable Secure Boot temporarily to test if the GPU becomes visible. If disabling Secure Boot resolves the issue, you have two options: keep it disabled or sign the NVIDIA modules manually. To manage kernel modules effectively, ensure that Nouveau is blacklisted. Create or edit the file/etc/modprobe.d/blacklist-nouveau.conf and add the following lines:
blacklist nouveau
options nouveau modeset=0sudo update-initramfs -u and reboot. This step prevents the open-source driver from claiming the device before the proprietary NVIDIA driver can load. Verify the blacklist is effective by checking lsmod | grep nouveau after reboot; it should return empty.
Driver Installation Verification
If Secure Boot was not the issue, focus on the driver installation itself. Ubuntu usually provides NVIDIA drivers through the "Software & Updates" application under the "Additional Drivers" tab. Select the recommended proprietary driver version and apply changes. Alternatively, use the command line to search for available drivers withubuntu-drivers devices. This command lists compatible drivers and recommends the best one based on your hardware and kernel version.
Install the recommended driver using sudo ubuntu-drivers autoinstall. This method ensures that the correct dependencies are pulled in automatically. After installation, reboot the system. Check the status again with nvidia-smi. If it now displays your GPU details, the problem is resolved. If not, check the journal logs for specific failure points using journalctl -xe | grep nvidia. Common errors include mismatched kernel headers or incomplete dependency chains.
Kernel Compatibility
Sometimes, a recent kernel update breaks compatibility with existing drivers. If the issue started after a system update, consider reverting to a previous kernel version from the GRUB menu. Access the advanced options during boot and select an older kernel. Test if the GPU is recognized in that environment. If it works, the new kernel introduced a regression. In such cases, wait for a kernel update that fixes the issue or manually patch the driver against the newer kernel headers. Additionally, ensure that your system has the necessary build tools for compiling driver modules. Installbuild-essential and linux-headers-generic if they are missing. These packages are required for dynamic kernel module support (DKMS) to function correctly. Without them, driver updates may fail silently, leaving the system without proper GPU acceleration.
Frequently Asked Questions
Why does lspci see the GPU but nvidia-smi does not?
This indicates that the kernel recognizes the hardware but cannot load the necessary proprietary driver module. The issue is often caused by the Nouveau driver taking precedence or Secure Boot blocking the unsigned NVIDIA module. Blacklisting Nouveau and verifying driver signatures usually resolves this discrepancy.
Can I keep Secure Boot enabled while using an RTX 4090?
Yes, but it requires manual signing of the NVIDIA kernel modules. You must generate a Machine Owner Key (MOK), enroll it in the BIOS, and then sign the driver files using the kmodsign utility. This process is complex and prone to errors, so many users prefer disabling Secure Boot for gaming rigs.
What should I do if nvidia-smi reports a timeout error?
A timeout error suggests the driver failed to communicate with the hardware within the expected timeframe. This often happens due to power management conflicts or improper shutdowns. Try resetting the PCI bus by running sudo rmmod nvidia_uvm nvidia_drm nvidia_modeset nvidia followed by reloading them, or simply rebooting the system to clear any stuck states.
Does Ubuntu Server require different steps for GPU detection?
No, the underlying mechanism for driver loading is identical between Desktop and Server editions. However, Server installations lack a graphical interface by default, so you must rely entirely on command-line tools like nvidia-smi and lspci for verification. Ensure that the nvidia-utils package is installed to access these utilities.
How do I know if my kernel headers are up to date?
You can check the installed header versions by listing files in /usr/src or running dpkg -l | grep linux-headers. Compare this output with your current kernel version obtained via uname -r. If they do not match exactly, reinstall the headers matching your running kernel to ensure DKMS can compile modules correctly.