topic Re:GPU driver conflicts in Analyzers

GPU driver conflicts

itl-dag — Thu, 29 Feb 2024 02:32:00 GMT

Following the OneAPI Installation guide from:

https://www.intel.com/content/www/us/en/developer/articles/guide/installation-guide-for-oneapi-toolkits.html

1. I confirmed that my hardware devices are supported client GPUs
david@linux-desktop:~$ lspci -nn |grep -Ei 'VGA|DISPLAY'
00:02.0 VGA compatible controller [0300]: Intel Corporation Device [8086:a780] (rev 04)
03:00.0 VGA compatible controller [0300]: Intel Corporation Device [8086:56a0] (rev 08)
(a780 signifies Intel UHD Graphics Xe Raptor Lake-S and 56a0 signifies Intel Arc A770 Graphics XeHPG Alchemist)
2. Before you begin (installing OneAPI Base Package) - prerequisites
a) Install GPU drivers - took me to:
https://dgpu-docs.intel.com/driver/client/overview.html

I followed instructions for Customized Display Stack with Ubuntu 22.04 LTS HWE at 2.1.2, 2.1.3, 2.1.4.
As I assumed I would need GPU Debugging, I also followed section 2.1.5.1 to install the Out-Of-Tree Driver.

sudo apt update && sudo apt install -y --install-suggests linux-generic-hwe-22.04

The kernel actually installed was 6.5.0-18-generic, which subsequently rose to 6.5.0-21-generic.

The first 5 steps seemed to execute but on step 6:
sudo apt install -y \
intel-i915-dkms intel-platform-vsec-dkms \
intel-platform-cse-dkms intel-fw-gpu

I got the output:
The following packages have unmet dependencies.
Intel-i915-dkms : Conflicts: intel-platform-cse-dkms
Conflicts: intel-platform-vsec-dkms
E: Unable to correct problems, you have held broken packages.

Did I do something wrong? Or is there something else I should have done first?

Re: GPU driver conflicts

Juhi_B_Intel — Tue, 26 Mar 2024 14:06:18 GMT

@itl-dag The Base toolkit forum is not actively monitored, so I have moved your question to the relevant forum where more community members can take a look at it - since its related to installation.

Re:GPU driver conflicts

Vipin_S_Intel — Tue, 26 Mar 2024 19:18:46 GMT

Hi David, Could you please share the below details with us?

Exact version of Intel® oneAPI Base toolkit.
Is Intel® oneAPI Base Toolkit installed or not?
Operating System build on which you are installing/installed the product.
Other versions of Intel® Parallel Studio XE or Intel® oneAPI already installed on the system.
Is the installer an online or an offline package?
Installation Log files.
- The default path for a log file for a oneAPI installation is: C:\Program Files (x86)\Intel\oneAPI\logs

To assist you further, we would require these details.

Re: Re:GPU driver conflicts

itl-dag — Wed, 27 Mar 2024 02:02:23 GMT

Thanks for getting back to me.

As I explained in my post, the issue is not with the OneAPI Basekit installation itself, I have successfully compiled and run OneAPI example code. However, this is not much use without the resources needed for the vTune diagnostic package and suitable debugger, including the correct GPU drivers and permissions.

I am currently running OneAPI version 2024.0 on Ubuntu 22.04, with kernel 6.5.0-21-generic.

I attach the log from vTune self-checker output.

Any help / suggestions would be gratefully received.

David.

Re:GPU driver conflicts

Vipin_S_Intel — Wed, 27 Mar 2024 19:04:02 GMT

Hi David, we would like to inform you that we are routing your query to the dedicated team for further assistance.

Re: GPU driver conflicts

yuzhang3_intel — Thu, 28 Mar 2024 02:37:07 GMT

Try with below command lines below first and run self-check again.

echo 0 > /proc/sys/kernel/perf_event_paranoid
echo 0 > /proc/sys/kernel/kptr_restrict
echo 0 > /proc/sys/kernel/yama/ptrace_scope
echo 0 > /proc/sys/dev/i915/perf_stream_paranoid

Re: GPU driver conflicts

itl-dag — Thu, 28 Mar 2024 03:56:29 GMT

When running all command lines, even prefixed by sudo, got the response...

bash: [target file]: Permission denied

... so I assume there is no point running the self-check again, as it looks like nothing happened.

The attached file might give you further information about the issues.

It outlines a history of actions and system output from trying to solve vTune driver problems up to this point, from the best information I could find at the time.

Thanks,

David.

Re: GPU driver conflicts

yuzhang3_intel — Thu, 28 Mar 2024 05:31:41 GMT

From the log, VTune driver installation failed. This is because your kernel was built with MODULE_SIG_FORCE=y, so signed kernel modules are required. You need to resolve the issue first and then configure the kernel files as mentioned above.

--------------------- Loading PAX driver ---------------------
Detecting Secure Boot status...
Secure Boot is enabled
Checking for PMU arbitration service (PAX) ... not detected.
Attempting to start PAX service ...
Executing: insmod ./pax/pax-x32_64-6.5.0-21-genericsmp.ko
insmod: ERROR: could not insert module ./pax/pax-x32_64-6.5.0-21-genericsmp.ko: Key was rejected by service
ERROR: pax driver failed to load!
The driver must be signed because MODULE_SIG_FORCE kernel config is set on the system.
ERROR: failed to start or connect to required PAX service