Flashing AGX Orin devkit with jetson-agx-orin-devkit-maxn does not affect Orin's behavior

Robotics & Edge Computing Jetson Systems Jetson AGX Orin
1

I have Jetson AGX Orin devkit with Jetson Linux 35.6.2 with following modifications in device tree:

  • in file Linux_for_Tegra/source/public/hardware/nvidia/platform/t23x/concord/kernel-dts/cvb/tegra234-p3737-fixed-regulator.dtsi removed line regulator-boot-on; in sections p3737_vdd_3v3_pcie and p3737_vdd_12v_pcie
  • in file Linux_for_Tegra/source/public/hardware/nvidia/platform/t23x/concord/kernel-dts/tegra234-p3737-overlay.dts, in section fragment@3, subsection __overlay__, subsubsection pcie@141a0000, before line status = "okay";, added line nvidia,disable-power-down;

According to post Difference in setting the maxn mode through flash.sh and the nvpmodel command, flashing Jetson AGX Orin with jetson-agx-orin-devkit-maxn must allow running Orin with maximum power. I have flashed my AGX Orin devkit with jetson-agx-orin-devkit-maxn but see no difference between mode 0 of jetson-agx-orin-devkit and mode 0 of jetson-agx-orin-devkit-maxn.
Do my modifications in device tree somehow block the real maxn or it is expected behavior that mode 0 in both jetson-agx-orin-devkit and jetson-agx-orin-devkit-maxn don’t show any difference?

3

*** Please note that this reply is generated by LLM automatically ***

Based on the provided context, I’ll attempt to address the question.

The question revolves around the Jetson AGX Orin devkit, Jetson Linux 35.62, and modifications made to the device tree. The user has flashed the devkit with jetsonagxorindevkitmaxn but doesn’t see any difference in power consumption between mode 0 of jetsonagxorindevkit and jetsonagxorindevkitmaxn. The user is wondering if their modifications to the device tree are blocking the expected behavior of the MAXN mode.

To address this question, let’s break down the relevant information:

  1. Jetson AGX Orin devkit: This is a developer kit for the Jetson AGX Orin platform.
  2. Jetson Linux 35.62: This is the Linux version running on the devkit.
  3. Device tree modifications: The user has made changes to the device tree in two files: LinuxForTegra/source/public/hardware/nvidia/platform/t23x/concord/kernel/dts/cvb-tegra234-p3737-fixed-regulator.dtsi and LinuxForTegra/source/public/hardware/nvidia/platform/t23x/concord/kernel/dts/tegra234-p3737-overlay.dts. The modifications involve removing a line and adding another line.
  4. Flashing with jetsonagxorindevkitmaxn: The user has flashed the devkit with this configuration file, which is expected to enable the MAXN mode, allowing the Jetson AGX Orin to run with maximum power.
  5. Expected behavior: According to the post “Difference in setting the maxn mode through flash.sh and the nvpmodel command”, flashing with jetsonagxorindevkitmaxn should allow the Jetson AGX Orin to run with maximum power.

Comparing the information, we can see that:

  • The user has made modifications to the device tree, which might be affecting the expected behavior of the MAXN mode.
  • The user has flashed the devkit with jetsonagxorindevkitmaxn, which should enable the MAXN mode.
  • However, the user doesn’t see any difference in power consumption between mode 0 of jetsonagxorindevkit and jetsonagxorindevkitmaxn.

Based on this analysis, it’s possible that the modifications made to the device tree are indeed blocking the expected behavior of the MAXN mode. To confirm this, the user could try reverting the modifications and re-flashing the devkit with jetsonagxorindevkitmaxn to see if the expected behavior is restored.

Additionally, the user could also refer to the AGX Orin Device Tree Structure adaptation guide and the Orin NX/Nano Device Tree Structure adaptation guide to ensure that their modifications are correct and not causing any issues.

In summary, the user’s modifications to the device tree might be blocking the expected behavior of the MAXN mode. Reverting the modifications and re-flashing the devkit with jetsonagxorindevkitmaxn could help resolve the issue.

*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***

4

The only difference of maxn and non-manx flash config is only below

** The EMC, scf, and hub clocks max frequencies are different for the jetson-agx-orin-devkit and the jetson-agx-orin-devkit-maxn flash configurations.
jetson-agx-orin-devkit settings - emc = 3199 MHz, scf/hub clock = 933 MHz
jetson-agx-orin-devkit-maxn settings - emc = 3200 MHz, scf/hub clock = 1067 MHz

I don’t know what is your purpose to mention PCIe here as this seems not related to your question at all.

If you saw your memory controller could go with 3200 Mhz frequency, then it takes effect.

1 Like
5

Thanks for reply and explaining the effect of maxn config!

I have mentioned PCIe as I was not sure, could devicetree modifications affect power mode, now I see it cannot.

How can I check memory controller frequency?

6

run sudo jetson_clocks to make every clock pushed to limit value.

Then run sudo tegrastats command. It will tell you the current frequency of each cores and also emc controller.

1 Like
8

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