jetson-customize-clocks
Use to lock/cap Jetson CPU/GPU/EMC clocks, toggle EMC/CPU DVFS, or change cpufreq governors by editing BPMP DTB and nvpower.sh pre-flash. Do NOT use for live tuning or nvpmodel edits.
npx skills add NVIDIA/skills --skill jetson-customize-clocks --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Instructs the agent to customize clock behavior on a Jetson target by editing two pre-flash components: the BPMP DTB (to set per-clock ceilings and EMC DVFS behavior) and nvpower.sh (to configure cpufreq/devfreq governors and boot-time rates). It supports operations to lock specific clocks, disable/enable EMC DVFS, and pin to MAXN for stress tests, all via pre-flash edits. It explicitly excludes live-target tuning and nvpmodel edits.
How it works
- Resolve active target profile and ensure BSP image is extracted with an initialized overlay tracker.
- Choose an operation: BPMP DTB edit (max-rate-custom), EMC DVFS disable/enable, or nvpower.sh edits; combine as a single workflow when doing the MAXN recipe.
- Operation 1 (BPMP DTB edits): follow the BPMP DTB customization protocol to modify the clock node's max-rate-custom, and optionally disable EMC DVFS by setting bwmgr.enabled and cactmon.enabled (and possibly removing osp-controller on T26x). Recompile and commit via the overlay tracker.
- Operation 2 (nvpower.sh edits): edit Linux_for_Tegra/rootfs/etc/systemd/nvpower.sh in the overlay tracker; apply substitutions through the canonical edit recipes and commit; deploy via /jetson-promote-image and /jetson-flash-image.
- For MAXN stress runs, apply both operations in one protocol cycle: remove max-rate-custom lines, enable EMC DVFS disable edits, and set both cpufreq and devfreq governors to performance in nvpower.sh, then set boot-default nvpmodel to MAXN via the sibling skill.
- Deployment proceeds through /jetson-promote-image and /jetson-flash-image to take effect on next boot.
When to use it
Use when you need to lock/cap specific Jetson clock rates or disable EMC/CPU DVFS before flashing, or when you want to pin to MAXN for stress tests. Do not use for live tuning or nvpmodel edits.
What it can touch
Touchpoints include:
- BPMP DTB edits under the Linux_for_Tegra/ path (content edits to max-rate-custom, EMC DVFS nodes, and osp-controller deletion on T26x).
- nvpower.sh edits located at Linux_for_Tegra/rootfs/etc/systemd/nvpower.sh in the overlay tracker.
Caveats
Edits are image-building, not live-tuning. max-rate-custom only lowers ceilings below max-rate-maxn. The effective runtime ceiling is the minimum of BPMP cap and active nvpmodel cap. EMC DVFS changes depend on SoC (T23x vs T26x) and may require additional edits; ODMDATA tokens may override BPMP edits at flash time. nvpower.sh edits can be clobbered by package upgrades; use a systemd drop-in or sibling helper for persistence.
# Customize Clocks ## Purpose Customize CPU, GPU, and EMC clock behavior on a Jetson target by editing files under `Linux_for_Tegra/` before flashing the image. Two layers are in scope: - The **BPMP DTB** at `Linux_for_Tegra/bootloader/<BPFDTB_FILE>` — per-clock `max-rate-custom` ceilings, plus the EMC DVFS gate (bwmgr + cactmon on all SoCs; osp-controller on T26x only). - **nvpower.sh** at `Linux
What does the jetson-customize-clocks skill do?
Use to lock/cap Jetson CPU/GPU/EMC clocks, toggle EMC/CPU DVFS, or change cpufreq governors by editing BPMP DTB and nvpower.sh pre-flash. Do NOT use for live tuning or nvpmodel edits.
How do I install it?
Run `npx skills add NVIDIA/skills --skill jetson-customize-clocks --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From NVIDIA/skills, a repository with 2,789 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.