Continuing Power Mode Issues on Jetpack 7.2

I have followed a number of the forum threads on the power mode issue but have not found a solution I can make work.

I have a new Jetson Orin Nano, which I flash from the SDK manager, from a Ubuntu machine without issue.

When I boot the newly flashed system, all power modes are available using the power mode icon in upper right. I can switch between modes using the icon, and no reboot is requested to switch between modes.

If I run a heavy Ollama model at MAXN SUPER, and the system throttles, then the powermodes break for me. I am unable to change power modes. I try to change modes, and a dialog box comes up and says I must reboot. I click ‘Yes’. But system does not reboot. I reboot manually, and the system comes up in the ‘stuck’ power mode.

I have gone all the way back to reflashing the OS from SDK manager several times, but still end up in this stuck power mode if I ever thermally throttle the board.

I would appreciate guidance on getting the power mode icon and power mode states working.
Thank you,
Paul Mcwhorter

Hi @paul.mcwhorter,

Since this is JetPack 7.2, please use the Jetson Linux r39.2 power mode controls as the reference:

For Jetson Orin Nano Super, MAXN_SUPER is tied to the jetson-orin-nano-devkit-super / jetson-orin-nano-devkit-super-maxn configuration. Some power mode changes may also require a reboot when the GPU configuration changes.

Because you flashed with SDK Manager and all modes are available immediately after a fresh flash, please first collect the failed-state configuration rather than applying a config-file workaround. Please reproduce the issue once, then collect the following outputs while the system is in the stuck state and before reflashing:

cat /etc/nv_tegra_release
cat /etc/nv_boot_control.conf
sudo /usr/sbin/nvpmodel -q --verbose
sudo /usr/sbin/nvpmodel -m 2

For the last command, if the verbose query shows that MAXN_SUPER has a different mode ID, please use that mode ID instead. If the command asks for reboot confirmation, please type YES and capture the full terminal output, including whether the system actually reboots.

Please also avoid editing /etc/nv_boot_control.conf or reflashing again before collecting these outputs, since that would remove the failed state we need to inspect.

Thanks.

Thank you for your help:

  1. I have flashed a fresh Jetpack 7.2 Using the SDK Manager
  2. I have booted up, and the power modes all show under the power icon, and I am free to switch power modes by the power icon in upper right
  3. I run a heavy ollama model, and the System thermally throttles.
  4. I try to change power modes using the icon, and it demands a reboot by a popup.
  5. I click ‘Yes’ but it does not reboot. Now I am in the classic, ‘stuck’ power mode.
  6. pjm@localhost:~$ cat /etc/nv_tegra_release

pjm@localhost:~$ cat /etc/nv_tegra_release

R39 (release), REVISION: 2.0, GCID: 45755727, BOARD: generic, EABI: aarch64, DATE: Mon Jun 1 09:28:48 PM UTC 2026

KERNEL_VARIANT: oot

TARGET_USERSPACE_LIB_DIR=nvidia
TARGET_USERSPACE_LIB_DIR_PATH=usr/lib/aarch64-linux-gnu/nvidia
pjm@localhost:~$ cat /etc/nv_boot_control.conf
TNSPEC 3767-300-0005-M.1-1-1-jetson-orin-nano-devkit-super-
COMPATIBLE_SPEC 3767–0005–1–jetson-orin-nano-devkit-super-
TEGRA_BOOT_STORAGE nvme0n1
TEGRA_CHIPID 0x23
TEGRA_OTA_BOOT_DEVICE /dev/mtdblock0
TEGRA_OTA_GPT_DEVICE /dev/mtdblock0
pjm@localhost:~$
pjm@localhost:~$ sudo /usr/sbin/nvpmodel -q --verbose
NVPM VERB: Config file: /etc/nvpmodel.conf
NVPM VERB: parsing done for /etc/nvpmodel.conf
NVPM VERB: Current mode: NV Power Mode: MAXN_SUPER
2
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_0: PATH /sys/devices/system/cpu/cpu0/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_1: PATH /sys/devices/system/cpu/cpu1/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_2: PATH /sys/devices/system/cpu/cpu2/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_3: PATH /sys/devices/system/cpu/cpu3/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_4: PATH /sys/devices/system/cpu/cpu4/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM CPU_ONLINE: ARG CORE_5: PATH /sys/devices/system/cpu/cpu5/online: REAL_VAL: 1 CONF_VAL: 1
NVPM VERB: PARAM FBP_POWER_GATING: ARG FBP_PG_MASK: PATH /sys/devices/platform/gpu.0/fbp_pg_mask: REAL_VAL: 0 CONF_VAL: 2
NVPM VERB: PARAM TPC_POWER_GATING: ARG TPC_PG_MASK: PATH /sys/devices/platform/gpu.0/tpc_pg_mask: REAL_VAL: 0 CONF_VAL: 240
NVPM VERB: PARAM GPU_POWER_CONTROL_ENABLE: ARG GPU_PWR_CNTL_EN: PATH /sys/devices/platform/gpu.0/power/control: REAL_VAL: auto CONF_VAL: on
NVPM VERB: PARAM CPU_A78_0: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_0: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM CPU_A78_1: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu1/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_1: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu1/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM CPU_A78_2: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu2/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_2: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu2/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM CPU_A78_3: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu3/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_3: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu3/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM CPU_A78_4: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu4/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_4: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu4/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM CPU_A78_5: ARG MIN_FREQ: PATH /sys/devices/system/cpu/cpu5/cpufreq/scaling_min_freq: REAL_VAL: 729600 CONF_VAL: 729600
NVPM VERB: PARAM CPU_A78_5: ARG MAX_FREQ: PATH /sys/devices/system/cpu/cpu5/cpufreq/scaling_max_freq: REAL_VAL: 1728000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM GPU: ARG MIN_FREQ: PATH /sys/devices/platform/17000000.gpu/devfreq_dev/min_freq: REAL_VAL: 306000000 CONF_VAL: 0
NVPM VERB: PARAM GPU: ARG MAX_FREQ: PATH /sys/devices/platform/17000000.gpu/devfreq_dev/max_freq: REAL_VAL: 1020000000 CONF_VAL: 9223372036854775807
NVPM VERB: PARAM GPU_POWER_CONTROL_DISABLE: ARG GPU_PWR_CNTL_DIS: PATH /sys/devices/platform/gpu.0/power/control: REAL_VAL: auto CONF_VAL: auto
NVPM VERB: PARAM EMC: ARG MAX_FREQ: PATH /sys/kernel/nvpmodel_clk_cap/emc: REAL_VAL: 3199000000 CONF_VAL: 9223372036854775807
pjm@localhost:~$

Now when I issue the change power mode command, it demands a reboot:
pjm@localhost:~$ sudo /usr/sbin/nvpmodel -m 0
NVPM WARN: Golden image context is already created
NVPM WARN: Reboot required for changing to this power mode: 0
NVPM WARN: DO YOU WANT TO REBOOT NOW? enter YES/yes to confirm:

I type ‘YES’ and it attempts a reboot, but goes to completely black screen, with not even a blinking cursor, and hangs. After several attempts, it does eventually reboot, and does have the new requested power mode from before the reboot. The power select icon in upper right still does not work. Selecting a different power mode creates a popup demanding a rebbot, clicking yes does nothing.

Hi Paul,

The reboot prompt is expected after the GPU has been used

The message below matches the documented nvpmodel behavior:

NVPM WARN: Golden image context is already created
NVPM WARN: Reboot required for changing to this power mode: 0

After the GPU golden context has been created, a power-mode change that requires a different GPU power-gating configuration must be completed through a reboot. The Ollama workload likely creates that GPU context; the outputs do not show that thermal throttling corrupted the power-mode configuration.

The black-screen reboot may match a Jetson Linux r39.2 known issue

Jetson Linux r39.2 release-note issue 6236259 documents that applying a power mode with an EMC frequency below Fmax during boot can cause a reboot failure, especially when a display is connected. Your transition from MAXN_SUPER to mode 0 similarly changes the EMC limit from 3199 MHz to 2133 MHz.

The release note explicitly lists the Orin Nano 8GB 7W mode, while your test selected the Super configuration’s 15W mode, so we should verify that the same display-related condition applies here.

Recommendation: Since mode 0 was eventually applied and persisted, please perform one headless boot to check the display-related condition. Shut down the system cleanly:

sudo poweroff

After it powers off, disconnect the monitor, power the system on, wait for it to finish booting, reconnect the monitor, and then run:

sudo /usr/sbin/nvpmodel -q

Please report whether the headless boot completes on the first attempt and which mode is reported afterward. There is no need to reflash or modify /etc/nv_boot_control.conf for this test.

Thanks.

Thank you for your help.

I can confirm that booting without a monitor connected seems to clear up the various power mode issues I have been having.

Let me summarize the issues I was having, for others who might read this thread:

  1. After running a heavy LLM model leading to a thermal throttle, the system begins to behave unpredictably. Boot failures, extremely long boot times, power select icon in upper right of menu bar ceases to function. Upon booting after a thermal throttle requested power mode changes are not respected.

SOLUTION:

Upon heavy thermal throttling, on the next reboot, reboot without a monitor attached. After Orin boots, plug monitor in and things work again as expected.

Thank you for your help.