Hypertune FPS test method (E4-8 draft) · 28 October 2026
For: Vin (draft), Travis (review 3 Nov, E4-13), Austin (sign-off, E4-9). From: Som. Sits with the internal facts page. Shared with outlets that retest Hypertune (E4-9). Not public.
What this sheet is for
Every FPS figure Hypertune publishes should say which test produced it. This method splits the gain in two, because they come from different places:
- Hardware and system tuning: CPU Overclocking, GPU Overclocking, Memory Optimizations, Windows Debloating, Network Optimizations. Same in-game settings before and after.
- Game-profile settings: Game Hub only. The game's own settings change; the hardware stays at stock.
A buyer, an outlet or an AI answer can then tell how much came from tuning the PC and how much from changing the game's settings.
Test PC (record every field)
| Field | Example of what to write |
|---|---|
| Processor | Intel Core i7-14700K |
| Graphics card | NVIDIA GeForce RTX 3080 10 GB |
| Memory | 2 x 16 GB DDR5-6000, XMP on or off |
| Motherboard and BIOS version | |
| Storage | NVMe SSD model |
| Cooling | CPU cooler model |
| Windows | Windows 11 version and build |
| Graphics driver | Version number |
| Hypertune | App version |
| Game | Title and patch version |
| Resolution and preset | 1920 x 1080, High preset |
| Scene | Map or benchmark scene, and how long it runs |
| Capture tool | The frame-time tool used [Austin: which tool Hypertune uses today] |
| Tester and date |
The four runs
Run each configuration three times in the same scene, after one warm-up run that is not counted. Restart the game between configurations. Reset Hypertune with the Restore tab before each new configuration.
| Config | Hypertune features on | In-game settings | What it measures |
|---|---|---|---|
| A. Baseline | None (stock hardware, Windows as installed) | The fixed preset above | Starting point |
| B. Hardware and system | CPU Overclocking, GPU Overclocking, Memory Optimizations, Windows Debloating, Network Optimizations | Same fixed preset | Gain from tuning the PC |
| C. Game profile | Game Hub only | Whatever the Game Hub profile sets | Gain from the game's settings |
| D. Everything | All six features | Game Hub profile | Combined gain |
Record for each run: average FPS, 1% low FPS, and run time. Keep every run's numbers, not only the average.
How to report the result
- Gain = (config FPS minus baseline FPS) divided by baseline FPS, for average FPS and 1% lows separately.
- Report B, C and D against A. Do not publish D alone without B and C.
- For C, list each setting the Game Hub profile changed (for example shadows from High to Low), so readers can see what they trade for the frames.
- Show the run-to-run spread. If the three runs of a config differ by more than the gain, say the result is within noise.
- A published figure always names its game, test PC and date. "Up to" figures name the single test they come from (V8 claims rule; E3-1).
- Overclock results depend on the chip and cooling. Write "on this PC", not "on every PC".
Figures already public that need this method behind them
| Figure | Where it appears | What is known | Missing |
|---|---|---|---|
| "Users see up to 60% higher FPS" | Approved V8 line; launch press says "in internal testing" | Nothing about game or PC | Game, PC, configs, date (E4-17, asked 20 Oct) |
| 18.9% to 28.2% | Tom's Hardware and TechPowerUp, 9 Sep | Core Ultra 9 285K with RTX 5090; Homeworld 3 at 18.9%, a Tomb Raider title at 28.2% | Exact titles, settings, which features were on, runs, date |
| 4.3% to 9.8% | Same articles | Core i7-14700K with RTX 3080; Rainbow Six and Marvel Rivals named | Which figure is which game, settings, features on, runs, date |
If Austin's team still has these runs, re-record them in the table above. Otherwise, rerun them on these two PCs with this method before the results page (E4-5).
For outlets that retest (E4-9)
We send: this sheet, a licence, and our table for the same game. They use their own PC, record the same fields, and publish whatever they find. We ask only that they run B and C separately and name their test PC.
Open questions for the call
- Austin: which capture tool and scene Hypertune used for the launch figures.
- Should Windows Debloating and Network Optimizations be their own config, split from the hardware overclocks? This adds two runs per game but shows the Windows gain on its own.
- Does the Game Hub profile lower visual settings in some games? If so, config C must list them (see reporting rules).