Your FPS counter
is not telling
the whole truth.

1. Restart Windows so an old game session or memory leak does not contaminate the test.
2. Close other games, downloads, unnecessary browser tabs, overlays, and recording tools.
3. Launch the game and warm up the test area once. Do not count the warm-up as a measured run.
4. Choose one 60–90 second route you can repeat closely: the same save, path, camera direction, and actions.
5. Write down the game version, display resolution, graphics preset, frame-generation setting, FPS cap, and driver version.

Why: If the route or conditions change, Frame Truth cannot tell whether your setting helped or the test simply became easier.

Press Run PC audit now. Frame Truth reads the Windows information it supports at that moment, including memory pressure, large running processes, supported games, GPU/display details when exposed, drive free space, power plan, WHEA hardware events, and application crashes.

Commit percentage

How much committed virtual memory Windows is using compared with its limit. Very high commit can signal memory pressure, but it does not prove that RAM is defective.

Available memory

Physical RAM Windows could provide at the audit time. A low value is a warning to retest after a restart with unnecessary apps closed.

WHEA event

A Windows Hardware Error Architecture record. It can support a stability investigation, but one event alone does not name the failed part.

Application crash

A Windows reliability event showing that an application failed. Match its timestamp and application name to the problem you experienced.

Next: Correct obvious contamination—such as another game running—then capture gameplay. Do not buy hardware from an audit warning alone.

Press Start live memory to refresh physical-memory readings every two seconds. Available shows RAM currently available, Used shows the percentage in use, and Total shows installed usable physical memory.

This panel does not measure live CPU load, GPU load, FPS, temperatures, clocks, or latency. Stop it when you are finished. A low-memory reading is evidence to repeat the test cleanly—not an automatic recommendation to purchase more RAM.

A CSV is a text table containing measured frame rows. Frame Truth accepts supported columns including DisplayedTime, MsBetweenDisplayChange, MsBetweenPresents, or FrameTime.

1. Create a capture using a tool you already own and trust.
2. Launch the game and perform the same 60–90 second route.
3. Choose the resulting CSV in Frame Truth.
4. Name the case and label three identical runs Baseline.
5. Change one variable and import three Comparison runs.
6. Review repeatability, P99 frametime, and severe hitches together.

Privacy: Frame Truth reads the CSV on this PC. It does not upload the capture.

Frame-time reference: 60 FPS equals about 16.67 ms per frame, 120 FPS equals 8.33 ms, and 144 FPS equals 6.94 ms. Consistency matters more than briefly reaching a high FPS.

1. Set the Test group to Baseline and use a clear label such as “High preset · FG off.”
2. Capture the same warmed route three times. Three runs reduce the risk of trusting a lucky or unusually bad pass.
3. Change exactly one variable—for example, frame generation off to on, one graphics setting, or one FPS cap.
4. Set the group to Comparison and label the change.
5. Capture the same route three more times under the same conditions.
6. Review P99 and hitch-count direction together. Check the repeatability label before calling the result proven.

Measured improvement: P99 decreases and hitch count does not increase across complete comparable groups. Not confirmed: the result is mixed, worse, incomplete, or too inconsistent. Restore or run a cleaner test instead of forcing a conclusion.

Diagnose ranks possible fault domains from the symptoms you select. A ranked hypothesis means “test this first,” not “this part is broken.” Test Plan creates a controlled recipe from your selected game, goal, refresh rate, VRR choice, and documented context.

Read the words Preview, Documented context, and Verify on this PC literally. The app does not claim that every listed fact applies to your current machine until it measures or you confirm it.

Most Frame Truth recommendations are instructions only. The current build supports one documented Crimson Desert configuration action. For a supported action:

1. Press Preview supported file change and review the exact target and proposed value.
2. Press Apply only if the target is correct. Frame Truth creates a backup first.
3. Repeat the six-run evidence workflow.
4. If performance does not improve or the game behaves incorrectly, press Restore latest backup.
5. Check the Change Ledger for the action time and result.

Never edit or delete the backup while you are evaluating the change.

The JSON report is structured evidence for software or technical review. The technician report is a readable HTML summary. Reports contain summaries—not every raw frame row. Keep both redaction boxes selected unless a trusted recipient needs your labels or filenames.

Open the exported report before sending it. Check names, paths, labels, timestamps, and action history. A report supports troubleshooting; it does not guarantee a repair or establish that hardware is defective.

App will not open

Restart the app. On a blank window, repair or install Microsoft Edge WebView2 Runtime.

CSV cannot be analyzed

Confirm it ends in .csv, is under 25 MB, contains at least ten valid rows, and includes DisplayedTime, MsBetweenDisplayChange, MsBetweenPresents, or FrameTime.

Capture has no frames

Create a new measurement with the game running, then import the resulting CSV.

Results change wildly

Warm the scene, close background interference, repeat the exact route, and avoid mixing first-run shader compilation with warm runs.

Audit looks old

Run a new audit. Every imported audit keeps its original capture timestamp.