Before/After benchmark

The Benchmark page lets you prove an optimization actually did something. You take a Before snapshot of your system, run your tweaks, take an After snapshot, and BRUTAL shows the exact delta — real numbers, not vague promises. It's the honest counterpart to one-click optimization.

What a snapshot measures

Each snapshot samples real Windows performance counters and process data — it takes a few seconds because CPU and disk are sampled over a one-second window for accuracy:

• Available RAM (MB) — free physical memory.
• CPU usage (%) — current load.
• Running processes and running services.
• Threads and handles — total across all processes.
• Disk read / write (MB/s) — system-wide I/O at sample time.
• Context switches/sec — how hard the scheduler is working.

Every snapshot is timestamped, and the live log on the page shows each measurement as it's captured.

Running a Before/After comparison

1. Open the Benchmark page.
2. Click Capture Before and wait a few seconds for the snapshot.
3. Go run your changes — Boost Now, Smart Optimize, Clean Memory, or whatever you're testing.
4. Come back and click Capture After.
5. The comparison card appears with the deltas, and the result is saved automatically.

The comparison reads in plain language, for example RAM freed: +512 MB (3200 → 3712), with the same before→after arrows for CPU, processes, services, threads, handles and context switches. A higher available RAM and lower process/service counts after optimization are the typical wins.

Reading the results honestly

A couple of things to keep in mind so you interpret the numbers fairly:

• CPU and disk are point-in-time. A background app waking up during a sample can move them — for those two, trends across several runs are more reliable than a single delta.
• RAM freed and process/service counts are the headline wins. These reflect what optimization genuinely closed or trimmed.
• Same state, both snapshots. For a fair test, capture Before and After under similar conditions — don't open a browser between them, or you'll "undo" the gains on paper.

Where results are saved

Each comparison is written as a timestamped JSON file under %LOCALAPPDATA%\BRUTAL\benchmarks\. When you reopen the page, the most recent result loads automatically so you can see your last run without re-measuring. Older files stay on disk if you want to keep a history.

Best practice

• Close anything you don't want counted before the Before snapshot, then leave the system alone between Before and After.
• Run the comparison right after a Boost or Clean Memory to see that specific action's effect.
• For CPU/disk claims, run the pair two or three times and look at the pattern rather than one number.
• Pair the benchmark with the Performance Score: the score shows live headroom, the benchmark shows what changed.

Common issues / troubleshooting

• "Capture After" is disabled. You need a Before snapshot first — capture Before (or run a Quick benchmark) to enable it.
• RAM "freed" is negative. Something started using memory between snapshots (a browser, a game launcher, Windows itself). Re-run with a quiet system.
• A line says "skipped". A performance counter wasn't available on your system for that metric; the rest of the snapshot is still valid.
• Numbers barely moved. Your system may already be lean, or the optimization you ran targets things a snapshot doesn't capture (like latency tweaks). That's the benchmark being honest, not broken.

The optimization receipt

Alongside the before/after snapshot, the Benchmark page now shows a receipt: for every change, what was asked for, what your machine reports back now, and where in Windows you can go and check it yourself. It scores only values it can read back — registry write-back, timer resolution, the active power plan, available RAM against a stated noise floor. Frame rate is shown but never scored, because a frame-capped game on a matching monitor pins both captures at the cap and the “delta” would be zero by construction. “No measurable change” is one of the answers it is allowed to give, because sometimes that is the true one.