GPU Undervolt Settings and Reported Results

An open dataset of community setting references and attributed external reports. Each row identifies its source and evidence category. A setting reference is not a completed experiment.

The dataset is rendered as a browsable table here: https://techfuelhq.com/articles/gpu-undervolt-settings-database/

Use the method notes to understand what a source reports before borrowing any settings.

---

Status: eight entries; corrected September 8, 2026

The dataset contains 8 rows compiled by TechFuelHQ:

Source type	Rows	What it means
editorial-firstparty	0	No logged TechFuelHQ before/after test is present
review-measured	1	Alva Jonathan's RX 9070 result as reported by Notebookcheck
community-verified	2	Attributed XDA and Esports Tales reports; not independently reproduced here
community-consensus	5	Setting references with unestablished outcome fields set to null

The category names are retained for schema compatibility. In this dataset, “verified” in a category name does not mean that TechFuelHQ reproduced the test. Read each row's method and source.

null means no configuration-specific outcome is established; zero would mean a reported zero change. The RTX 5070 Ti's 50 W saving belongs to the source's Steel Nomad test. The RX 9070's rounded -7% records a loss. Different workloads make these reports unsuitable for claiming a typical saving or guaranteed stability across cards.

Cards covered so far: RTX 5090, RTX 5080, RTX 4090, RTX 4080 Super, RTX 5070 Ti (NVIDIA) and RX 9070 XT, RX 9070, RX 7900 XTX (AMD).

Every row carries a source_url. Contributions should add a traceable stock-versus-tuned report, with unmeasured fields left null.

---

How to contribute

1. Fork this repo.
2. Add one object to the results array in data/undervolt_results.json.
3. Open a pull request. The submission template fills in automatically — complete the field guide and checklist in it.

Opening a pull request requires a GitHub account. Email hello@techfuelhq.com with the same information if you prefer not to use GitHub. Maintainers review submissions through pull requests.

What the automated check does

scripts/undervolt_outlier_flag.py runs on pull requests touching the dataset or checker (workflow). It checks schema, numeric bounds and required source fields, and warns about statistical outliers. These checks cannot establish that an experiment happened, authenticate a measurement, or prove stability. The site table excludes rows marked flagged; maintainers must review the evidence.

Run it yourself before opening the PR:

python scripts/undervolt_outlier_flag.py --self-test   # proves the checker works
python scripts/undervolt_outlier_flag.py --check       # validates the dataset

The rules that get a row rejected

• An outcome without an attributable measurement. Leave it null. Do not estimate. A value from a published report must name that source and workload; never present it as your own test.
• No resolvable source. source_url is required — a review, a forum or Reddit thread, or your own screenshot or log.
• An untested "stable" claim. Say how long you tested and in what. FurMark does not validate an NVIDIA curve undervolt; it needs a real load that holds the card at the locked clock.
• Setting flagged or flag_reason yourself. Those belong to the maintainer and the script.

---

Schema

Top level is an object with two keys: meta (dataset metadata, including a field dictionary) and results (the array of rows). Each row:

Field	Type	Notes
id	string	Unique slug, gpu-board-setting
gpu	string	Canonical model, e.g. RTX 5080, RX 9070 XT
vendor	string	NVIDIA or AMD
board	string	AIB/board variant, or "" if generic/consensus
tbp_watts	number	Stock total board power, watts
method	string	MSI Afterburner V/F curve (NVIDIA), or Adrenalin offset / max voltage (AMD)
setting	string	Human-readable summary of what was applied
voltage_mv	number|null	NVIDIA curve target, or AMD RDNA3 absolute max voltage
offset_mv	number|null	AMD negative voltage offset (RDNA4)
clock_mhz	number|null	Target/held core clock
power_limit_pct	number|null	Percent of stock (NVIDIA, e.g. 90) or delta (AMD, e.g. -10); null if untouched
power_saved_w	number|null	Measured reduction in board power
temp_drop_c	number|null	Measured hotspot/junction drop, °C
fps_delta_pct	number|null	+ gain, 0 parity, - loss
stability_hours	number|null	How long it was tested
test_method	string	Workload, attribution and limits; reference instructions for untested rows
source_type	string	editorial-firstparty, review-measured, community-verified, community-consensus, or community-submitted
source_url	string	Required. Source for the setting reference or reported outcome
source_url_2	string	Optional second source
submitter	string	Who submitted it
date	string	YYYY-MM-DD added/corrected date, not necessarily a test date
flagged	boolean	Maintainer/script owned — leave false
flag_reason	string	Maintainer/script owned — leave ""

meta.fields in the JSON carries the same dictionary, so the file documents itself.

Correction history

September 8, 2026: removed outcome estimates from five consensus references and synchronized the public download with the website. The original public RTX 5080 row incorrectly claimed first-party provenance; a July local correction had not reached the public repository. No first-party before/after result is established for that row. The README's former “eight tested configurations” framing and “no account” contribution instruction were also corrected.

---

License and attribution

The dataset is licensed CC BY 4.0 (Creative Commons Attribution 4.0 International). Use it, remix it, publish it commercially — just credit it:

TechFuelHQ Community GPU Undervolt Results (CC BY 4.0). https://techfuelhq.com/articles/gpu-undervolt-settings-database/

Contributions are accepted under the same license.

The scripts and workflow in this repo exist to keep the data honest, and are provided as-is.