ThrottlePoint

Energy Optimization Governor for HPC systems.

ThrottlePoint dynamically caps CPU frequency and NVIDIA GPU power based on electricity price, grid carbon intensity, and temperature. Which signals drive the decision is fully configurable, and you can plug in your own temperature source.

It runs from cron as root, with no daemon and no batch-system integration. Each run reads the signal its mode is configured for — a 1–10 rating from price and carbon intensity, or a temperature band — and sets the CPU's maximum scaling frequency accordingly. Running jobs are never signalled, paused, or killed; they run under a lower clock ceiling while conditions are bad.

All the functionality and code logic located in src/hpc_eff stem from https://gitlab.cesnet.cz/dexter/hpc_eff.

---

Documentation

docs/install.md	Start here. Prerequisites, building the RPM or DEB.
docs/deployment.md	Configure, first run, verify, tune — one node, either mode.
docs/cluster-rollout.md	Many nodes, upgrades, uninstall.
docs/configuration.md	Every key of /etc/hpc_eff/config.ini.
docs/regulation-modes.md	What each mode does with its input: the rating maths, the thermal bands.
docs/monitoring.md	Database schema, state.json, useful queries.
docs/troubleshooting.md	Symptom-first fault finding.

---

Use cases

• Driver: carbon intensity of electricity
  • Primary GreenDIGIT scenario
  • Lowering carbon impact
• Driver: electricity cost or consumption profile
  • Lowering operational costs
• Driver: datacenter temperature
  • Optimize datacenter usage
  • Continuity of services
• Driver: power-grid or local heating signals
  • Lowering operational costs, carbon impact

---

Quick start

git clone git@github.com:CESNET/hpc_eff.git
cd hpc_eff

RHEL / AlmaLinux / Rocky

sudo dnf install -y ipmitool make kernel-tools rpm-build rpmdevtools
make                                   # build + install the RPM

Debian / Ubuntu

sudo apt install -y build-essential devscripts debhelper dh-python \
                    python3-all python3-setuptools fakeroot
sudo apt install -y ipmitool cpufrequtils python3-numpy python3-requests
make deb
sudo dpkg -i ../hpc-eff_*.deb          # this also enables the cron job

Then configure and start:

sudo vi /etc/hpc_eff/config.ini        # pick control_mode, then its section (see below)
sudo hpc-eff                           # one real run: applies a cap immediately
sudo hpc-eff --enable                  # install /etc/cron.d/hpc-eff (every 10 min)

What to set depends on the mode: co2 needs [CO2_API] (an API key) and [frequency_tables]; temperature needs [TEMPERATURE_SOURCE] and [CPU_THERMO], and calls no external API at all.

Full prerequisites and build detail: docs/install.md. Configuration and verification: docs/deployment.md. Fleet rollout and uninstall: docs/cluster-rollout.md.

---

Regulation modes

[MODE] control_mode is the single switch per node. It is required.

Mode	CPU	GPU	Inputs
co2	rating 1–10 → max-frequency cap	untouched	spot price + grid carbon intensity
temperature	hysteresis thermal bands	NVIDIA power limiting	one temperature sensor

CPU regulation is mutually exclusive: a node is driven either by carbon/price or by temperature, never both. Details: docs/regulation-modes.md.

---

Data sources

Signal	Source	Notes
Electricity price	spotovaelektrina.cz (Czech OTE spot market)	Current price via API; the 12-month baseline is scraped from HTML
Carbon intensity	Nowtricity (default) or Wattnet	Selectable via [CO2_API] TYPE; Nowtricity needs an API key
Temperature	IPMI sensor, HTTP API, or your own Python module	"Bring your own reader": see docs/configuration.md

---

Logging

Every run writes one row to /var/lib/hpc_eff/history.db (SQLite) and rewrites /var/lib/hpc_eff/state.json with the current evaluation plus a rolling history. Schema and queries: docs/monitoring.md.

---

License

BSD-3-Clause. See LICENSE.