Infrastructure Blueprint Capabilities#

Collecting a Platform Report with system-info.sh#

system-info.sh is a diagnostic script for Intel® Panther Lake (PTL) systems provisioned using Infrastructure Blueprint. After provisioning, the script is available on the target system at /opt/edge/developer/tools/system-info/.

Summary of Tools#

Common tools: bash, lscpu, lsblk, ip, etc.
Optional tools for a more complete report: dmidecode, turbostat, intel_gpu_top, vulkaninfo, vainfo, clinfo, fwupdmgr
sudo recommended for full visibility (firmware, DMI, turbostat, dmesg)

Running the script#

cd /opt/edge/developer/tools/system-info
sudo ./system-info.sh

Save output to a file:

sudo ./system-info.sh > sys-info.txt 2>&1

Note: If PTL is not detected (CPUID mismatch), the script still runs and reports what it finds. Some sections show “not installed” warnings when optional tools are missing.

Output Sections Reference#

The script produces the following sections. Use this table to navigate the output.

Section	What it covers
SYSTEM INFO	Script version, hostname, kernel, OS, uptime, `hostnamectl` output
PANTHER LAKE PLATFORM CHECK	CPUID validation (family/model/stepping), microcode, Secure Boot state, PTL-relevant firmware blobs (`xe`, `huc`, `gsc`, `vpu_50xx`)
CPU INFO	`lscpu` full output, hybrid P/E/LP-E core topology and capacities, per-CPU frequency table, `intel_pstate` governor/HWP settings, cache hierarchy (L1/L2/L3), ISA flags (AVX, AVX-VNNI, AES, SHA, etc.), hardware vulnerability mitigations, live CPU usage, top 5 CPU-consuming processes, `turbostat` summary
MEMORY INFO	`free -h`, key `/proc/meminfo` fields (hugepages, swap, slabs), DIMM details from `dmidecode` (type, speed, manufacturer, part number), NUMA topology
STORAGE INFO	`lsblk` block device tree with filesystem and mount points, `df -h` disk usage, NVMe device info (`nvme-cli`), SMART data
NETWORK INFO	Interface list with IP addresses, default routes, wireless info (`iw`), PCI network device drivers
INTEL GPU INFO	PCI VGA devices, `/sys/class/drm` device details (vendor/device/revision, GT0/GT1 frequencies), `xe` kernel module info, `intel_gpu_top` utilisation sample, OpenGL/Mesa renderer, Vulkan info, VA-API profiles and entrypoints (`vainfo`)
INTEL NPU INFO	PCI accelerator device (`8086:b03e`), `/sys/class/accel` details, `intel_vpu` driver version and firmware version, NPU firmware blobs, relevant `dmesg` messages
INTEL COMPUTE / AI RUNTIMES	OpenCL platform/device details (`clinfo`) — EUs, device IP, USM capabilities; Level Zero library inventory; OpenVINO version and available device list (`CPU`, `GPU.0`–`GPU.7`, `NPU`); oneAPI/DPC++ runtime library paths
INTEL USERSPACE PACKAGES (dpkg)	Installed Intel packages grouped by: CPU/platform/monitoring, GPU/media/display, NPU/AI/OpenVINO/oneAPI, kernel and firmware
THERMALS, POWER, FANS	`lm-sensors` output, thermal zone temperatures, cooling device states, RAPL powercap zones (long-term/short-term power limits), battery/power supply state
FIRMWARE / BIOS / SECURITY	SMBIOS CPU and board details (`dmidecode`), UEFI boot confirmation, TPM state, `fwupd` firmware versions for CPU microcode, display controller, NVMe SSD, system firmware, BootGuard
PCI / USB DEVICE SUMMARY	Full Intel® PCI device list (BDF, class, device ID), all PCI devices, USB bus/device topology
DMESG: LAST 30 INTEL-RELATED LINES	Filtered dmesg lines for `xe`, `intel_vpu`, and related Intel driver messages
RECOMMENDED PACKAGES FOR INTEL PTL	`apt install` commands grouped by: kernel/firmware, core diagnostics, GPU/media, OpenCL/Level Zero, NPU/OpenVINO, useful extras

Provisioned System Profile#

The following tables describe what is expected to be present on a system that has been provisioned using the Infrastructure Blueprint. They cover the four key layers of a provisioned edge node: the underlying platform components, the AI and compute environment, the system services, and the orchestration stack.

Platform Components#

Component	Detail
Platform	Intel® Panther Lake Client Platform
ISA extensions	SSE4.2, AVX, AVX2, AVX-VNNI, AES-NI, SHA-NI, VAES, VPCLMULQDQ, GFNI, MOVDIRI, MOVDIR64B (no AVX-512 / AMX)
CPU governor	`intel_pstate` / `powersave`; HWP active, turbo enabled;
OS	Ubuntu OS Version 24.04 LTS (`minimal-desktop-ubuntu`)
Kernel	`linux-image-6.18-intel 260427T075939Z-r2`; command line: `xe.max_vfs=7 xe.force_probe=* modprobe.blacklist=i915 udmabuf.list_limit=8192`
iGPU	`xe` driver 1.1.0; device `8086:b08f`; 8 Physical Functions (PFs), 7 SR-IOV Virtual Functions (VFs); persisted via `intel-sriov-vf.service`
iGPU firmware	`ptl_guc_70.bin.zst`, `ptl_huc.bin.zst`, `ptl_gsc_1.bin.zst`
NPU (NPU 5)	`intel_vpu` 1.0.0 (in-kernel); firmware `vpu_50xx_v1.bin` (Mar 2026); `intel-level-zero-npu 1.32.0`
Ethernet	Intel® I226-V (`8086:57b4`); `igc` driver; managed via netplan/NetworkManager
Firmware	`PTLPFWI1.R00.3393.D60.2511181224` (2025-11-18); Secure Boot disabled (Setup Mode)

AI and Compute Environment#

Component	Detail
OpenVINO™ Runtime and OpenVINO™ toolkit	`2025.4.1-20426` runtime and toolkit; inference targets: `CPU`, `GPU.0`–`GPU.7`, `NPU` (Intel® AI Boost)
OpenCL	OpenCL 3.0 via `intel-opencl-icd 26.05.37020.3`; device IP `0x7800004` (Xe3); DP4A and DPAS; USM supported
Level Zero	`level-zero 1.22.4` and `level-zero-devel`; `libze_intel_gpu` and `libze_intel_npu` loaded
oneAPI Deep Neural Network Library (oneDNN)	`intel-oneapi-dnnl 2026.0.0-688` and `-devel`
oneAPI TBB	`intel-oneapi-tbb 2023.0.0-724`
VA-API / media	iHD driver `25.4.6`; `intel-media-va-driver-non-free`; `libvpl2 2.16.0` (oneVPL); decode: H.264, HEVC, VP9, AV1, VVC, MPEG-2, JPEG; encode: H.264, HEVC, VP9, AV1, JPEG
GStreamer framework	Full plugin set that comprises base, good, bad, ugly, OpenCV, RTSP, and Qt5
Mesa	`mesa-vulkan-drivers 25.3.4`, `mesa-va-drivers 25.2.8`
Container Device Interface (CDI)	GPU specification generator written in Go programming language and built from source; NPU generator script
Developer tools	`edge-node-infrastructure-blueprint` repo at `/opt/edge/developer/`; `system-info.sh` at `/opt/edge/developer/tools/system-info/`

Services#

Service	Detail
Network Time Protocol (NTP)	`chrony` installed and enabled; configurable via `/etc/chrony/chrony.conf`
Precision Time Protocol (PTP)	`linuxptp` available for precision time protocol
Container runtime	Docker CE, containerd, Buildx and the Compose plugin (active when `host_type=container`)
Kubernetes sever	K3s single-node server (active when `host_type=kubernetes`); traefik disabled
SR-IOV	`intel-sriov-vf.service` — provisions and persists 7 GPU VFs across reboots
Power monitoring and tuning	`powertop`, `pcm`; power tuning scripts (`battery`, `balanced`, `performance`, `graphical` profiles)
GPU monitoring	`intel-gpu-tools 1.28` (`intel_gpu_top`)
Network performance and profiling	`iperf3`, `linuxptp`, `tcpdump`

Kernel Performance and Debug Tooling#

User-space tools sourced from the Intel® Linux overlay (linux-tools/) and aligned with the installed kernel linux-image-6.18-intel build. They provide tracing, profiling, BPF, CPU power control, and related diagnostics tied to the running kernel.

Package	Detail
`bpftool`	Inspect and manage eBPF programs, maps, links, and BTF metadata loaded in the kernel
`hyperv-daemons`	Microsoft Hyper-V integration daemons (KVP, VSS, fcopy) for running the image as a Hyper-V guest
`intel-sdsi`	Intel® Software Defined Silicon (SDSi) provisioning and feature-activation utility
`libcpupower1`	Shared runtime library used by `cpupower` and other CPU frequency / idle tools
`libcpupower-dev`	Development headers for `libcpupower` to build custom CPU power tooling
`linux-bpf-dev`	Kernel BPF UAPI headers required to compile BPF programs against this kernel
`linux-config-6.18`	Exact `.config` used to build the installed `6.18-intel` kernel
`linux-cpupower`	`cpupower` CLI for CPU frequency governors, idle states, and turbo / HWP control
`linux-kbuild`	Kernel build infrastructure (scripts, host tools) needed for out-of-tree module builds
`linux-misc-tools`	Miscellaneous in-tree user-space tools (e.g. `x86_energy_perf_policy`, `tmon`, `bootconfig`, `gpio`/`iio`/`pci`/`spi`/`usb`/`wmi` utilities)
`linux-perf`	`perf` profiler for hardware counters, sampling, tracing, and flame-graph workflows
`rtla`	Real-Time Linux Analysis tool for latency tracing (`osnoise`, `timerlat`, `hwnoise`)
`usbip`	USB-over-IP client/server for exporting and attaching USB devices across the network

Orchestration#

Component	Detail
Host type dispatch	`kubernetes`: K3s and Helm and device plugins; `container`: Docker and containerd
Helm tool	version 3.x — deployed via `get-helm-3` during provisioning
Intel® device plugins	Node Feature Discovery (NFD), GPU plugin, NPU plugin — deployed as Helm charts
SR-IOV accelerated containers	VF provisioning and CDI GPU specifications help enable passthrough to containers via device plugin
NPU accelerated containers	CDI NPU generator and Intel® NPU device plugin for workload scheduling to NPU
Provisioning scripts	`/opt/edge/scripts/` — `kubernetes-provision.sh`, `container-provision.sh`, `setup-kernel-depended-pkgs.sh`