Jetson Orin Nano not reporting CPU/GPU information correctly

Robotics & Edge Computing Jetson & Embedded Systems Jetson Orin Nano
1

Hey all, dmidecode and lspci on the Jetson Orin Nano with Jetpack 6.1 don’t show up all the info I’d expect (e.g. there’s limited CPU information, just version “Orin”; CUDA/Tensor cores aren’t visible). Similarly, nvidia-smi shows a pretty basic Orin (nvgpu), with all other fields to N/A. (The last one is I guess because NVML doesn’t support Jetsons).

Is this on purpose? Is there a way to manually update these?

Background:
Fiddling around with Nomad as an orchestrator to deploy stuff on the Jetson Orin Nano, I discovered that the CPU and GPU are not getting picked up. This led me down a rabbit hole through GitHub - klauspost/cpuid: CPU feature identification for Go, and then ended up in dmidecode and lspci. I would like for more detail to be presented to be able to target different devices in a fleet (it’s still doable right now, but would require manually tagging each device with the hardware it has, which I prefer not to).

dmidecode is very vague on the CPU, and lspci is even more so:

# dmidecode 3.3
Getting SMBIOS data from sysfs.
SMBIOS 3.6.0 present.
# SMBIOS implementations newer than version 3.5.0 are not
# fully supported by this version of dmidecode.

Handle 0x000A, DMI type 4, 50 bytes
Processor Information
	Socket Designation: Not Specified
	Type: Central Processor
	Family: ARMv8
	Manufacturer: NVIDIA
	ID: 34 02 6B 03 01 04 00 00
	Signature: JEP-106 Bank 0x03 Manufacturer 0x6b, SoC ID 0x0234, SoC Revision 0x00000401
	Version: Orin
	Voltage: Unknown
	External Clock: 31 MHz
	Max Speed: 1728 MHz
	Current Speed: 1728 MHz
	Status: Populated, Enabled
	Upgrade: Unknown
	L1 Cache Handle: 0x0007
	L2 Cache Handle: 0x0008
	L3 Cache Handle: 0x0009
	Serial Number: Not Specified
	Asset Tag: Not Specified
	Part Number: Not Specified
	Core Count: 6
	Core Enabled: 6
	Thread Count: 6
	Characteristics:
		64-bit capable
		Multi-Core
		Execute Protection
		Enhanced Virtualization
		Arm64 SoC ID

(note the “Version: Orin”, “Voltage: Unknown”).

lspci just lists a few “NVIDIA Corporation Device xxx”:

0001:00:00.0 PCI bridge: NVIDIA Corporation Device 229e (rev a1)
0001:01:00.0 Network controller: Realtek Semiconductor Co., Ltd. RTL8822CE 802.11ac PCIe Wireless Network Adapter
0004:00:00.0 PCI bridge: NVIDIA Corporation Device 229c (rev a1)
0004:01:00.0 Non-Volatile memory controller: MAXIO Technology (Hangzhou) Ltd. NVMe SSD Controller MAP1202 (rev 01)
0008:00:00.0 PCI bridge: NVIDIA Corporation Device 229c (rev a1)
0008:01:00.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168/8411 PCI Express Gigabit Ethernet Controller (rev 15)

Just for a comparison, a random mini PC with an Intel CPU with integrated graphics shows this in lspci:

00:02.0 VGA compatible controller: Intel Corporation JasperLake [UHD Graphics] (rev 01)

and more detail in dmidecode:

        Version: Intel(R) Celeron(R) N5095 @ 2.00GHz
        Voltage: 1.1 V
2

Note that Jetsons use an integrated GPU (iGPU) which is wired directly to the memory controller. Other systems use a discrete GPU (dGPU) wired only through PCI. The original nvidia-smi is only for use with a dGPU. It is only recently that a minimal nvidia-smi was added for partial information of the GPU, so it is limited due to that.

And of course lspci won’t show a GPU that isn’t on the PCI bus.

3

That’s why I included the comparable outputs from a mini PC with an Intel CPU with an iGPU. It includes the GPU in lspci, and dmidecode includes more CPU information than the Jetson Orin. My question is mostly about dmidecode, because that’s entirely vendor supplied information and can include whatever NVIDIA decide.

5

Hi,
Please refer to this:
Nvidia-smi does not show meaningfull data - #11 by AastaLLL

6

Hey, thanks for that, nvidia-smi (and maybe NVML) supporting the Jetsons more would be great.

However, is that related to dmidecode and lspci not showing much, if any, information about the Jetson CPU and iGPU?

7

Hi,
Information about CPU and iGPU is stated in module data sheet:

Jetson Download Center | NVIDIA Developer

You can run $ sudo tegrastats to get runtime frequency and loading.

9

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.