1. Installation
This page describes software installation for Intel® Data Center GPU Max Series and Intel® Data Center GPU Flex Series. For information about other Intel GPUs, see Client GPU Installation. The steps described here are not validated for client GPUs.
NOTE: For readability purposes, Intel® Data Center GPU Max Series might be refered to as Max, while Intel® Data Center GPU Flex Series might be referred to as Flex.
1.1. Supported Hardware
Linux* operating system distribution support varies by hardware family. The following table indicates which streams are supported for each hardware family and OS combination. Support indicates Intel has validated that combination. Other combinations may work, however they are not explicitly tested by Intel and issues reported against non-supported combination are subject to Intel’s evaluation prior to fixing in the corresponding release stream.
1.1.1. Intel® Data Center GPU Max Series and Intel® Data Center GPU Flex Series
Usage |
Ubuntu Server |
Red Hat Enterprise Linux |
SUSE Linux Enterprise Server |
|---|---|---|---|
Max |
LTS, Production, Rolling |
LTS, Production, Rolling |
LTS, Production, Rolling |
Flex |
LTS, Production, Rolling |
LTS, Production, Rolling |
N/A |
1.1.2. Intel® Arc™ Graphics
Usage |
Ubuntu Desktop (HWE) |
Red Hat Enterprise Linux |
SUSE Linux Enterprise Server |
|---|---|---|---|
Client |
Rolling |
N/A |
N/A |
For client install (Intel® Arc™ Graphics) steps see Client GPU Installation. To identify which hardware you have see Hardware Devices.
1.2. Release Streams
Three types of release streams are available for Intel® Data Center GPUs.
Rolling |
Production |
Long Term Support - LTS |
|
|---|---|---|---|
Purpose |
Early adopters who want to evaluate new features and where new hardware enablement first appears. |
For production use with 6 months * of support of bug/security fixes, no new features |
For production use with 3 years* support of bug/security fixes, no new features |
Release lifetime* |
Supported only until next published rolling release. |
6 months* |
3 years* |
The instructions on this page will install the LTS version, which is recommended for most users. For information on configuring your system to install from a different stream, see Release Streams.
For client install steps see Client GPU Installation.
1.3. Install Steps
Instructions for each OS follow this sequence:
Install the Intel GPU driver repository.
Install GPU drivers for bare metal access and management.
Adjust kernel boot parameters as needed.
Install the oneAPI toolkit and other components as required.
NOTE: The Intel GPU driver releases are designed to use your operating system’s package manager for installation and dependency resolution.
1.3.1. Red Hat Enterprise Linux Install Steps
These steps describe installation of the unified LTS driver release for Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series on Red Hat Enterprise Linux.
Supported Red Hat Enterprise Linux (RHEL) versions:
Red Hat Enterprise Linux 8.6, 8.8, 8.9, 9.0, 9.2, 9.3
NOTE: To install the kernel modules provided, your system needs to support the Dynamic Kernel Module System (DKMS), which is not provided by the official Red Hat Enterprise Linux package repositories. You can install DKMS onto your system through the Extra Packages for Enterprise Linux (EPEL) project.
If DKMS is not available, the following instructions will fail with a package dependency not being met for DKMS.
The following steps pick one of the options for DKMS install, but this can be changed to any of the other methods described in the previous link to match your system setup.
1.3.1.1. Red Hat Enterprise Linux Package repository
To add the online network package repository for the LTS releases:
. /etc/os-release
if [[ ! " 8.6 8.8 8.9 9.0 9.2 9.3 " =~ " ${VERSION_ID} " ]]; then
echo "RHEL version ${VERSION_ID} not supported"
else
echo "Installing online network repository for ${VERSION_ID}"
sudo dnf install -y 'dnf-command(config-manager)'
sudo dnf config-manager --add-repo \
https://repositories.intel.com/gpu/rhel/${VERSION_ID}/lts/2350/unified/intel-gpu-${VERSION_ID}.repo
fi
Optional: Instead of the online network package repository steps above, you can download and use the offline repository:
. /etc/os-release
if [[ ! " 8.6 8.8 8.9 9.0 9.2 9.3 " =~ " ${VERSION_ID} " ]]; then
echo "RHEL version ${VERSION_ID} not supported"
else
wget https://repositories.intel.com/gpu/rhel/${VERSION_ID}/lts/2350/intel-gpu-rhel-${VERSION_ID}-2350.run
chmod +x intel-gpu-rhel-${VERSION_ID}-2350.run
sudo ./intel-gpu-rhel-${VERSION_ID}-2350.run
fi
This will extract and enable the Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series repository in your OS repository management configuration.
NOTE: Apart from enabling the repository, the offline installer does not make any other changes to the OS configuration.
To add the EPEL repository needed for DKMS:
. /etc/os-release
sudo dnf -y install https://dl.fedoraproject.org/pub/epel/epel-release-latest-${VERSION_ID%.*}.noarch.rpm
sudo dnf config-manager --disable epel
NOTE: After installing EPEL, it is disabled by default as EPEL currently has packages which conflict with the packages provided by Intel. It is enabled for a single command below to allow DKMS to be installed.
1.3.1.2. Red Hat Enterprise Linux Package installation
The kernel and xpu-smi packages can be installed on a bare metal system. Installation on the host is sufficient for hardware management and support of the runtimes in containers and bare metal.
sudo dnf install -y kernel-headers flex bison
sudo dnf install --enablerepo epel -y intel-fw-gpu intel-i915-dkms xpu-smi
sudo reboot
NOTE: There are known limitations associated with the installation of the intel-i915-dkms package. You may encounter a warning about an unknown symbol during the installation process. Please be aware that this is a known issue and does not affect the functionality of the software.
Compute and Media Runtimes
sudo dnf install -y \
intel-opencl intel-media intel-mediasdk libmfxgen1 libvpl2\
level-zero intel-level-zero-gpu mesa-dri-drivers mesa-vulkan-drivers \
mesa-vdpau-drivers libdrm mesa-libEGL mesa-libgbm mesa-libGL \
mesa-libxatracker libvpl-tools intel-metrics-discovery \
intel-metrics-library intel-igc-core intel-igc-cm \
libva libva-utils intel-gmmlib libmetee intel-gsc intel-ocloc
Development packages
sudo dnf install -y --refresh \
intel-igc-opencl-devel level-zero-devel intel-gsc-devel libmetee-devel \
level-zero-devel
Install verification tools
sudo dnf install -y hwinfo clinfo
1.3.2. SUSE Linux Enterprise Server Install Steps
These steps describe installation of the unified LTS driver release for Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series on SUSE Linux Enterprise Server.
Supported SUSE Linux Enterprise Server (SLES) versions:
SUSE Linux Enterprise Server 15 SP4, 15 SP5
1.3.2.1. SUSE Linux Enterprise Server Package repository
To add the online network package repository for the LTS releases:
. /etc/os-release
VERSION_SP=${VERSION_ID//./sp}
if [[ ! " 15sp4 15sp5 " =~ " ${VERSION_SP} " ]]; then
echo "SLES version ${VERSION_ID} not supported"
else
sudo zypper addrepo -r \
https://repositories.intel.com/gpu/sles/${VERSION_SP}/lts/2350/unified/intel-gpu-${VERSION_SP}.repo
sudo zypper --non-interactive --gpg-auto-import-keys refresh
fi
Optional: Instead of the online network package repository steps above, you can download and use the offline repository:
. /etc/os-release
VERSION_SP=${VERSION_ID//./sp}
if [[ ! " 15sp4 15sp5 " =~ " ${VERSION_SP} " ]]; then
echo "SLES version ${VERSION_ID} not supported"
else
wget https://repositories.intel.com/gpu/sles/${VERSION_SP}/lts/2350/intel-gpu-sles-${VERSION_SP}-2350.run
chmod +x intel-gpu-sles-${VERSION_SP}-2350.run
sudo ./intel-gpu-sles-${VERSION_SP}-2350.run
fi
This will extract and enable the Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series repository in your OS repository management configuration.
NOTE: Apart from enabling the repository, the offline installer does not make any other changes to the OS configuration.
1.3.2.2. SUSE Linux Enterprise Server Package installation
Prior to installing the packages for Intel GPUs on SUSE Linux Enterprise Server, you need to ensure the following Modules and Extensions are enabled in your SUSE subscription in order for all required packages to be available:
Basesystem Module
Development Tools Module
Product-WE
Package-HUB
For information on configuring extensions and modules, refer to SUSE Modules Article.
Once your system has the required modules and extensions enabled, you can continue with installing the GPU packages.
The kernel and xpu-smi packages can be installed on a bare metal system. Installation on the host is sufficient for hardware management and support of the runtimes in containers and bare metal.
Prior to installing intel-i915-dkms below, if you are not running the latest kernel from SUSE,
you need to ensure your system has the kernel devel package for your kernel. To verify that
package is installed, you can run: zypper install --oldpackage kernel-default-devel-$(uname -r)
sudo zypper install -y lsb-release linux-kernel-headers flex bison intel-fw-gpu intel-i915-dkms xpu-smi
NOTE: When installing a kernel module which has not been supplied by SUSE, for example the
module that is built when installing the intel-i915-dkms package, you must set
the system configuration allow_unsupported_modules to 1. Failure to enable that option will
prevent the kernel from loading the intel-i915-dkms generated kernel module.
The instructions below are based on the instructions supplied by SUSE:
# Copy the default into /etc.
sudo cp /lib/modprobe.d/10-unsupported-modules.conf /etc/modprobe.d/10-unsupported-modules.conf
# Modify the default to set allow_unsupported_modules to 1
sudo sed -i -E 's,allow_unsupported_modules 0,allow_unsupported_modules 1,g' \
/etc/modprobe.d/10-unsupported-modules.conf
If you will be booting your system from an initrd and have dracut configured, make sure to re-run dracut after changing the above parameter:
# Update the initrd
sudo dracut -f
For additional details related to the allow_unsupported_modules option, see SUSE’s Working with unsupported modules.
Once you have installed intel-i915-dkms and ensured allow_unsupported_modules is set to 1, you can reboot the system to load the new module:
sudo reboot
Compute and Media Runtimes
sudo zypper install -y \
intel-level-zero-gpu level-zero intel-gsc intel-opencl intel-ocloc \
intel-media-driver libigfxcmrt7 libvpl2 libvpl-tools libmfxgen1 libmfx1
Development Packages
sudo zypper install -y \
libigdfcl-devel intel-igc-cm libigfxcmrt-devel level-zero-devel
Install Verification Tools
sudo zypper install -y clinfo libOpenCL1 libva-utils hwinfo
1.3.3. Ubuntu Install Steps
These steps describe installation of the unified LTS driver release for Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series on Ubuntu Server.
NOTE: Datacenter installs are validated with Ubuntu Server. Client installs are validated with Ubuntu Desktop using the Rolling Stable release. For client usage, see Client GPU Installation.
Supported Ubuntu versions:
Ubuntu Server 22.04.1 (5.15 LTS kernel)
1.3.3.1. Ubuntu Package Repository
Make sure prerequisites to add repository access are available:
sudo apt update
sudo apt install -y gpg-agent wget
To add the online network package repository for the LTS releases:
. /etc/os-release
if [[ ! " jammy " =~ " ${VERSION_CODENAME} " ]]; then
echo "Ubuntu version ${VERSION_CODENAME} not supported"
else
wget -qO - https://repositories.intel.com/gpu/intel-graphics.key | \
sudo gpg --dearmor --output /usr/share/keyrings/intel-graphics.gpg
echo "deb [arch=amd64 signed-by=/usr/share/keyrings/intel-graphics.gpg] https://repositories.intel.com/gpu/ubuntu ${VERSION_CODENAME}/lts/2350 unified" | \
sudo tee /etc/apt/sources.list.d/intel-gpu-${VERSION_CODENAME}.list
sudo apt update
fi
Optional: Instead of the online network package repository steps above, you can download and use the offline repository:
. /etc/os-release
if [[ ! " jammy " =~ " ${VERSION_CODENAME} " ]]; then
echo "Ubuntu version ${VERSION_CODENAME} not supported"
else
wget https://repositories.intel.com/gpu/ubuntu/dists/jammy/lts/2350/intel-gpu-ubuntu-${VERSION_CODENAME}-2350.run
chmod +x intel-gpu-ubuntu-${VERSION_CODENAME}-2350.run
sudo ./intel-gpu-ubuntu-${VERSION_CODENAME}-2350.run
fi
This will extract and enable the Intel® Data Center GPU Flex Series and Intel® Data Center GPU Max Series repository in your OS repository management configuration.
NOTE: Apart from enabling the repository, the offline installer does not make any other changes to the OS configuration.
1.3.3.2. Ubuntu Package Installation
The kernel and xpu-smi packages can be installed on a bare metal system. Installation on the host is sufficient for hardware management and support of the runtimes in containers and bare metal.
sudo apt install -y \
linux-headers-$(uname -r) \
linux-modules-extra-$(uname -r) \
flex bison \
intel-fw-gpu intel-i915-dkms xpu-smi
sudo reboot
Compute and Media Runtimes
sudo apt install -y \
intel-opencl-icd intel-level-zero-gpu level-zero \
intel-media-va-driver-non-free libmfx1 libmfxgen1 libvpl2 \
libegl-mesa0 libegl1-mesa libegl1-mesa-dev libgbm1 libgl1-mesa-dev libgl1-mesa-dri \
libglapi-mesa libgles2-mesa-dev libglx-mesa0 libigdgmm12 libxatracker2 mesa-va-drivers \
mesa-vdpau-drivers mesa-vulkan-drivers va-driver-all vainfo hwinfo clinfo
Development Packages
sudo apt install -y \
libigc-dev intel-igc-cm libigdfcl-dev libigfxcmrt-dev level-zero-dev
1.3.4. Configuring Render Group Membership
To access GPU capabilities, the active user must be able to access DRM render nodes in /dev/dri.
The following command will list the group assigned ownership of the render nodes and will also list the groups the active user is a member of:
stat -c "%G" /dev/dri/render*
groups ${USER}
If the active user is not a member of the same group used by the DRM render nodes (usually ‘render’), add the user to the render node group.
sudo gpasswd -a ${USER} render
This will be sufficient for shells created after this update. To change the group ID of the current shell:
newgrp render
1.4. Verify Install
1.4.1. To Verify that Expected Hardware is Working with the i915 Driver (Flex and Max)
hwinfo --display
On SLES, hwinfo might be installed in /usr/sbin and not in the default
user path so you need to run via:
/usr/sbin/hwinfo --display
Here is an example output for Intel® Data Center GPU Max 1550 (device ID 0x0bd5):
51: PCI 8c00.0: 0380 Display controller
[Created at pci.386]
Unique ID: JefI.QAjErpDk4H4
Parent ID: juVd.xbjkZcxCQYD
SysFS ID: /devices/pci0000:89/0000:89:02.0/0000:8a:00.0/0000:8b:01.0/0000:8c:00.0
SysFS BusID: 0000:8c:00.0
Hardware Class: graphics card
Model: "Intel Display controller"
Vendor: pci 0x8086 "Intel Corporation"
Device: pci 0x0bd5
SubVendor: pci 0x8086 "Intel Corporation"
SubDevice: pci 0x0000
Revision: 0x2f
Driver: "i915"
Driver Modules: "i915"
Memory Range: 0x23fe7e000000-0x23fe7fffffff (ro,non-prefetchable)
Memory Range: 0x236000000000-0x237fffffffff (ro,non-prefetchable)
IRQ: 138 (447 events)
Module Alias: "pci:v00008086d00000BD5sv00008086sd00000000bc03sc80i00"
Driver Info #0:
Driver Status: i915 is active
Driver Activation Cmd: "modprobe i915"
Config Status: cfg=new, avail=yes, need=no, active=unknown
Attached to: #26 (PCI bridge)
A list of device IDs is included in the Graphics processor table section.
1.4.2. xpu-smi Device Information and Telemetry
The Intel® XPU Manager (Intel® XPUM) tool provides a wide range of functionality covering system administration, GPU monitoring, diagnostics, and configuration for Intel data center GPUs. It can be used in full featured mode with a RESTful API as well as via the simplified xpu-smi tool.
Some example xpu-smi command lines are shown next.
Available GPU device discovery:
$ xpu-smi discovery
+-----------+--------------------------------------------------------------------------------------+
| Device ID | Device Information |
+-----------+--------------------------------------------------------------------------------------+
| 0 | Device Name: Intel(R) Data Center GPU Flex 170 |
| | Vendor Name: Intel(R) Corporation |
| | UUID: 00000000-0000-0000-6769-df256e271362 |
| | PCI BDF Address: 0000:4d:00.0 |
| | DRM Device: /dev/dri/card1 |
| | Function Type: physical |
+-----------+--------------------------------------------------------------------------------------+
Detailed device information, including installed driver and firmware versions:
$ sudo xpu-smi discovery -d 0
+-----------+--------------------------------------------------------------------------------------+
| Device ID | Device Information |
+-----------+--------------------------------------------------------------------------------------+
| 0 | Device Type: GPU |
| | Device Name: Intel(R) Data Center GPU Flex 170 |
| | Vendor Name: Intel(R) Corporation |
| | UUID: 00000000-0000-0000-6769-df256e271362 |
| | Serial Number: LQAC13401787 |
| | Core Clock Rate: 2050 MHz |
| | Stepping: C0 |
| | |
| | Driver Version: I915_23.4.15_PSB_230307.15 |
| | Kernel Version: 5.15.0-47-generic |
| | GFX Firmware Name: GFX |
| | GFX Firmware Version: DG02_1.3267 |
| | GFX Firmware Status: normal |
| | GFX Data Firmware Name: GFX_DATA |
| | GFX Data Firmware Version: 0x46b |
| | GFX PSC Firmware Name: GFX_PSCBIN |
| | GFX PSC Firmware Version: |
| | AMC Firmware Name: AMC |
| | AMC Firmware Version: |
| | |
| | PCI BDF Address: 0000:4d:00.0 |
| | PCI Slot: J37 - Riser 1, Slot 1 |
| | PCIe Generation: 4 |
| | PCIe Max Link Width: 16 |
| | OAM Socket ID: |
| | |
| | Memory Physical Size: 14248.00 MiB |
| | Max Mem Alloc Size: 4095.99 MiB |
| | ECC State: enabled |
| | Number of Memory Channels: 2 |
| | Memory Bus Width: 128 |
| | Max Hardware Contexts: 65536 |
| | Max Command Queue Priority: 0 |
| | |
| | Number of EUs: 512 |
| | Number of Tiles: 1 |
| | Number of Slices: 1 |
| | Number of Sub Slices per Slice: 32 |
| | Number of Threads per EU: 8 |
| | Physical EU SIMD Width: 8 |
| | Number of Media Engines: 2 |
| | Number of Media Enhancement Engines: 2 |
| | |
| | Number of Xe Link ports: |
| | Max Tx/Rx Speed per Xe Link port: |
| | Number of Lanes per Xe Link port: |
+-----------+--------------------------------------------------------------------------------------+
GPU telemetry:
$sudo xpu-smi stats -d 0
+-----------------------------+--------------------------------------------------------------------+
| Device ID | 0 |
+-----------------------------+--------------------------------------------------------------------+
| GPU Utilization (%) | 0 |
| EU Array Active (%) | |
| EU Array Stall (%) | |
| EU Array Idle (%) | |
| | |
| Compute Engine Util (%) | 0; Engine 0: 0, Engine 1: 0, Engine 2: 0, Engine 3: 0 |
| Render Engine Util (%) | 0; Engine 0: 0 |
| Media Engine Util (%) | 0 |
| Decoder Engine Util (%) | Engine 0: 0, Engine 1: 0 |
| Encoder Engine Util (%) | Engine 0: 0, Engine 1: 0 |
| Copy Engine Util (%) | 0; Engine 0: 0 |
| Media EM Engine Util (%) | Engine 0: 0, Engine 1: 0 |
| 3D Engine Util (%) | |
+-----------------------------+--------------------------------------------------------------------+
| Reset | |
| Programming Errors | |
| Driver Errors | |
| Cache Errors Correctable | |
| Cache Errors Uncorrectable | |
| Mem Errors Correctable | |
| Mem Errors Uncorrectable | |
+-----------------------------+--------------------------------------------------------------------+
| GPU Power (W) | 44 |
| GPU Frequency (MHz) | 2050 |
| GPU Core Temperature (C) | 40 |
| GPU Memory Temperature (C) | |
| GPU Memory Read (kB/s) | 1346 |
| GPU Memory Write (kB/s) | 286 |
| GPU Memory Bandwidth (%) | 0 |
| GPU Memory Used (MiB) | 26 |
| Xe Link Throughput (kB/s) | |
+-----------------------------+--------------------------------------------------------------------+
For more information on Intel® XPUM, see
1.4.3. To Smoke Test the Compute Stack (Flex and Max)
clinfo | head -n 5
Example output:
Number of platforms 1
Platform Name Intel(R) OpenCL HD Graphics
Platform Vendor Intel(R) Corporation
Platform Version OpenCL 3.0
Platform Profile FULL_PROFILE
Running the same command without ‘head’ should display many pages of GPGPU compute capability summary.
1.4.4. To Smoke Test the Media Stack (Flex only)
vainfo
Intel® Data Center GPU Max Series does not include codec capabilities, so the expected output has minimal entry points.
vainfo: VA-API version: 1.18 (libva 2.17.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 23.1.4 (12e141d)
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
Intel® Data Center GPU Flex Series and client GPUs provide hardware codecs, so many entrypoints are expected from vainfo output.
Example output from Intel® Data Center GPU Flex Series:
vainfo: VA-API version: 1.18 (libva 2.17.0)
vainfo: Driver version: Intel iHD driver for Intel(R) Gen Graphics - 23.1.4 (12e141d)
vainfo: Supported profile and entrypoints
VAProfileNone : VAEntrypointVideoProc
VAProfileNone : VAEntrypointStats
VAProfileMPEG2Simple : VAEntrypointVLD
VAProfileMPEG2Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointVLD
VAProfileH264Main : VAEntrypointEncSliceLP
VAProfileH264High : VAEntrypointVLD
VAProfileH264High : VAEntrypointEncSliceLP
VAProfileJPEGBaseline : VAEntrypointVLD
VAProfileJPEGBaseline : VAEntrypointEncPicture
VAProfileH264ConstrainedBaseline: VAEntrypointVLD
VAProfileH264ConstrainedBaseline: VAEntrypointEncSliceLP
VAProfileHEVCMain : VAEntrypointVLD
VAProfileHEVCMain : VAEntrypointEncSliceLP
VAProfileHEVCMain10 : VAEntrypointVLD
VAProfileHEVCMain10 : VAEntrypointEncSliceLP
VAProfileVP9Profile0 : VAEntrypointVLD
VAProfileVP9Profile0 : VAEntrypointEncSliceLP
VAProfileVP9Profile1 : VAEntrypointVLD
VAProfileVP9Profile1 : VAEntrypointEncSliceLP
VAProfileVP9Profile2 : VAEntrypointVLD
VAProfileVP9Profile2 : VAEntrypointEncSliceLP
VAProfileVP9Profile3 : VAEntrypointVLD
VAProfileVP9Profile3 : VAEntrypointEncSliceLP
VAProfileHEVCMain12 : VAEntrypointVLD
VAProfileHEVCMain422_10 : VAEntrypointVLD
VAProfileHEVCMain422_12 : VAEntrypointVLD
VAProfileHEVCMain444 : VAEntrypointVLD
VAProfileHEVCMain444 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_10 : VAEntrypointVLD
VAProfileHEVCMain444_10 : VAEntrypointEncSliceLP
VAProfileHEVCMain444_12 : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointVLD
VAProfileHEVCSccMain : VAEntrypointEncSliceLP
VAProfileHEVCSccMain10 : VAEntrypointVLD
VAProfileHEVCSccMain10 : VAEntrypointEncSliceLP
VAProfileHEVCSccMain444 : VAEntrypointVLD
VAProfileHEVCSccMain444 : VAEntrypointEncSliceLP
VAProfileAV1Profile0 : VAEntrypointVLD
VAProfileAV1Profile0 : VAEntrypointEncSliceLP
VAProfileHEVCSccMain444_10 : VAEntrypointVLD
VAProfileHEVCSccMain444_10 : VAEntrypointEncSliceLP
1.4.5. Verify VSEC is in use for Intel® Data Center GPU Max Series
To access the full range of Intel® Data Center GPU Max Series telemetry features the intel_vsec module should be used instead of intel_pmt. The intel_vsec module supports Max telemetry features while intel_pmt focuses on CPU telemetry.
To check if VSEC change is needed, review output from xpu-smi discovery -d 0.
If you see output like the following, there may be a vsec issue for the device serial number.
Serial Number: unknown
1.4.5.1. Test that the intel_vsec Module Loads and is Associated with a PCI Device:
for d in 8086:09A7 8086:4F93 8086:4F95; do sudo lspci -k -d $d; done
Should see: 05:00.0 Memory controller: Intel Corporation Device 09A7 Kernel driver in use: intel-vsec Kernel modules: intel_vsec
If intel_pmt is in use as a kernel driver instead of intel_vsec(example below), change the kernel driver in use.
Example: 05:00.1 System peripheral: Intel Corporation Device 09a7 Kernel driver in use: intel-pmt Kernel modules: intel_pmt, intel_vsec
1.4.5.2. To Change Kernel Driver Module:
Install driverctl tool:
For Red Hat Enterprise Linux:
sudo dnf install driverctl
For SUSE Linux Enterprise Server 15: A driverctl package is not available for SUSE Linux Enterprise Server15. Instead, install from the driverctl repo:
git clone https://gitlab.com/driverctl/driverctl.git
cd driverctl
sudo make install
For Ubuntu:
sudo apt install driverctl
Next, check which device the intel-pmt module is linked to.
sudo driverctl list-devices | grep -iE "pmt"
Should see: 0000:8e:00.1 intel-pmt
You may see a different device address than 0000:8e:00.1. Use your system’s device address in the following command.
sudo driverctl set-override 0000:8e:00.1 "intel_vsec"
1.5. IFWI Update
The Intel® XPUM utility can be used to flash IFWI to a Flex or Max GPU.
Check GFX FW version on devices (that is, for each GPU):
sudo xpu-smi discovery -d 0
sudo xpu-smi discovery -d 1
Find the latest firmware version for your hardware from your Intel or OEM portal. If it is newer than what is currently on your device, install the new firmware with xpu-smi updatefw. Example:
sudo xpu-smi updatefw -d 0 -t GFX -f
/home/intel/ATS_M75_128_B0_PVT_ES_017_gfx_fwupdate_SOC2.bin -y
sudo xpu-smi updatefw -d 0 -t GFX_PSCBIN -f /home/test/PVC_Tuscany_oam_cbb_otf_53G_220803.pscbin
sudo xpu-smi updatefw -d 0 -t GFX -f /home/test/PVC.Fwupdate_Prod_2023.WW26.3_Tuscany_Pcie.bin
updatefw options:
sudo xpu-smi updatefw
Update GPU firmware
Usage: xpu-smi updatefw [Options]
xpu-smi updatefw -d [deviceId] -t GFX -f [imageFilePath]
xpu-smi updatefw -d [pciBdfAddress] -t GFX -f [imageFilePath]
Options:
-h,--help Print this help message and exit
-j,--json Print result in JSON format
-d,--device The device ID or PCI BDF address
-t,--type The firmware name. Valid options: GFX, GFX_DATA, GFX_CODE_DATA, GFX_PSCBIN, AMC. AMC firmware update just works on Intel M50CYP server (BMC firmware version is 2.82 or newer) and Supermicro SYS-620C-TN12R server (BMC firmware version is 11.01 or newer).
-f,--file The firmware image file path on this server
-u,--username Username used to authenticate for host redfish access
-p,--password Password used to authenticate for host redfish access
-y,--assumeyes Assume that the answer to any question which would be asked is yes
--force Force GFX firmware update. This parameter only works for GFX firmware.
1.6. Architecture Overview
1.6.1. Introduction
Solutions built using Intel GPU products rely on several software components working together. The Intel GPU Base (bottom rectangle) provides standard interfaces to higher level use cases (top rectangle).
The following sections describe the GPU Base block diagram components.
1.6.2. Intel Data Center GPUs
Intel® Data Center GPU Max Series Intel’s highest performing, highest density, general-purpose discrete GPU, which packs over 100 billion transistors into a package and contains up to 128 Xe Cores – Intel’s foundational GPU compute building block.
Intel® Data Center GPU Flex Series Intel’s general-purpose GPU optimized for media stream density and quality. Infused with server capabilities, Intel® Data Center GPU Flex Series enable high levels of reliability, availability, and scalability. Hardware includes Xe Cores and media acceleration.
1.6.3. Intel Kernel Driver(s)
The Linux* kernel driver(s) provide the software connection to the Intel GPU hardware. The kernel driver(s) are provided today as Dynamic Kernel Module Support (DKMS) drivers “out-of-tree” from the Linux kernel. See Kernel Driver Types for more information on the kernel driver.
1.6.4. Compute and Media Stacks
The Compute and Media usermode library stacks provide higher level software support of industry standard APIs.
1.6.5. Intel® XPU Manager (Intel® XPUM)
Use Intel® XPUM for managing the physical hardware, including firmware updates, monitoring,diagnostics, and configuration. Intel® XPUM uses the oneAPI Level Zero Sysman API, and exposes standard interfaces for telemetry, including Prometheus.
1.7. Additional Information
1.7.1. Install the oneAPI Toolkit
For general instructions, see Intel® oneAPI Toolkits and Components Installation Guide for Linux* OS.
For instructions specific to using the system package manager to install oneAPI, see Intel® oneAPI Toolkits and Components Install Using Package Managers.
1.7.2. Kernel Boot Parameters
1.7.2.1. Long Running Compute Workloads
The Intel graphics kernel driver defaults to preventing long running
compute workloads from executing longer than four seconds in order to
prevent display usages from blocking desktop operations due to an errant
shader. This setting is controlled via the i915 kernel module parameter
enable_hangcheck.
To check the current setting:
sudo cat /sys/module/i915/parameters/enable_hangcheck
A value of Y (default) indicates that the hang check is currently active.
You can change the setting at runtime via:
echo N | sudo tee /sys/module/i915/parameters/enable_hangcheck
To persist the value after reboot, the default setting can be changed by
passing i915.enable_hangcheck=0 in your GRUB_CMDLINE_LINUX_DEFAULT
setting. You can add the option using instructions provided by your operating
system provider, or by editing the value of GRUB_CMDLINE_LINUX_DEFAULT in
/etc/default/grub to include i915.enable_hangcheck=0.
After editing, the line for GRUB_CMDLINE_LINUX_DEFAULT should look similar to the following:
GRUB_CMDLINE_LINUX_DEFAULT="i915.enable_hangcheck=0"
After editing, update the GRUB boot files:
Ubuntu:
update-grubRed Hat Enterprise Linux and SUSE Linux Enterprise Server:
grub2-mkconfig -o /boot/grub2/grub.cfg
After rebooting, you should now see i915.enable_hangcheck=0 in /proc/cmdline
and the value in /sys/module/i915/parameters/enable_hangcheck should
be 0.
1.7.2.2. Multi-Card Deployments
Some systems may have compatibility issues between the system BIOS and the
Linux kernel MMIO BAR re-allocation, which prevents the Intel® Data Center GPUs
from being accessible once the system has booted. If you are
experiencing problems with multi-card solutions and the Intel GPU devices
not initializing (entries for the device are not enumerating in /dev/dri),
you can work around the issue by adding pci=realloc=off to your kernel’s
boot parameter.
To check if your system already has pci=realloc=off, you can check for the
string via cat /proc/cmdline. If it does not, you can add the option
using instructions provided by your operating system provider, or by
editing the value of GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub
to include pci=realloc=off.
After editing, the line for GRUB_CMDLINE_LINUX_DEFAULT should look similar to the following:
GRUB_CMDLINE_LINUX_DEFAULT="pci=realloc=off"
After editing, update the GRUB boot files:
Ubuntu:
update-grubRed Hat Enterprise Linux and SUSE Linux Enterprise Server:
grub2-mkconfig -o /boot/grub2/grub.cfg
After rebooting, you should now see pci=realloc=off in /proc/cmdline.
1.7.3. Host System Support
Packages provided on this site have been validated with the following platform configurations:
Host system |
GPU |
|---|---|
12th Generation Intel® Core™ Processors (Codename Alder Lake-H, Alder Lake-P, Alder Lake-U, Alder Lake-S, Alder Lake-HX) |
Intel® Arc™ A-Series Graphics (Codename Alchemist) |
11th Generation Intel® Core Processors (Codename Tiger Lake) |
Intel® Arc™ A-Series Graphics (Codename Alchemist) |
3rd Gen Intel® Xeon Scalable Ice Lake (M50CYP Family) |
Intel® Data Center GPU Flex Series |