I Measured My Whole Input Chain. The Slowest Part Was Free to Fix.
Part 3 of 3. How to actually measure end-to-end latency at home without a high-speed camera, what the numbers mean, and why the expensive links are almost never the problem.
Two articles of theory are worth exactly nothing if you cannot measure your own machine. So this one is the method.
You do not need lab equipment. You need a phone, a stopwatch mentality and about twenty minutes.
Method 1: the software readout (easiest, and enough)
Modern GPU drivers and several games expose an end-to-end system latency metric in their performance overlay. It is not a lab number, but it is measured on your machine, with your settings, which is what you actually care about.
What to do:
- Turn the readout on and note the number in a quiet menu.
- Note it again mid-fight, when the GPU is under real load.
- The gap between those two numbers is your render-queue problem. A small gap means your settings are healthy. A large gap means Part 2, item 1.
That single comparison is worth more than any spec sheet.
Method 2: the phone camera (slower, more honest)
If you want an absolute number rather than a relative one, film it.
- Set your phone camera to its highest frame rate — 240fps on most phones now.
- Film your mouse hand and the screen in the same frame.
- Click, then count video frames between the physical click and the first pixel change.
- Each frame at 240fps is 4.17 ms. Multiply.
Repeat ten times and take the median, not the best one. The best one is luck.
What the numbers actually mean
Rough interpretation for a competitive shooter, measured end to end:
| Measured total | Verdict |
|---|---|
| Under 25 ms | Excellent. Stop optimising and go practise. |
| 25-40 ms | Normal for a healthy setup. Small wins left. |
| 40-60 ms | Something in the chain is misconfigured — usually the render queue. |
| Over 60 ms | A setting is wrong, not a component. Work Part 2 top to bottom. |
The thing worth internalising: the range from "misconfigured" to "healthy" is bigger than the range from "healthy" to "expensive". Configuration beats hardware, and it is free.
If your measured latency drops 20 ms from a settings change, no monitor upgrade on the market was going to give you that.
The isolation trick
If the total is bad and you cannot tell which link is responsible, remove links one at a time:
- Windowed borderless vs exclusive fullscreen. Different compositor paths, sometimes a large difference.
- Overlays off, everything. Then back on one at a time.
- Frame cap on and off at several values.
- Different USB port, direct to the rear panel, no hub.
- A second mouse, if you have one, to rule out debounce.
Change one thing, re-measure, write it down. This is boring and it is the only method that produces an answer.
If the isolation pass never converges — the numbers move without you changing anything — stop tuning and check what else is running. Our sister site's how to tell if your PC has malware covers the signs, and unexplained frame-time variance is one of them.
What I actually found
The pattern that comes up over and over in this kind of testing is the same one: the slowest link is a setting, not a component. Frame cap and render queue first, USB and polling second, overlays third.
The peripherals — the part with the marketing budget — are usually somewhere near the bottom of the list of things costing you time. They matter for fit and comfort, which affects your aim through your body rather than through the chain, and that is a real effect worth spending on. Just not for the milliseconds.
If your latency is fine and your route is bad, that is the other chain: GearUp Booster ping review, and GearUp Booster if it helps on yours. If your problem is comfort over a four-hour session, that is the desk setup that ruins your aim and a chair like the AutoFull M6 Ultra 2.0.
Related
Best gaming monitors 2026 · best gaming mouse 2026 · optimise Windows for gaming · PC bottleneck guide · complete guide to monitor settings
Q: How do I measure input lag without special equipment? A: Film your hand and screen together with a phone at 240fps, click, and count video frames until the screen changes. Each frame is about 4.17 ms. Take the median of ten attempts.
Q: What is a good end-to-end latency for competitive shooters? A: Under about 25 ms is excellent and 25-40 ms is a healthy setup. Above 40 ms usually indicates a misconfiguration rather than inadequate hardware.
Q: Does exclusive fullscreen reduce latency compared with borderless? A: It can, because it uses a different presentation path with less compositor involvement. The size of the difference varies by game and by Windows version, so measure both on your own machine.
Back to Part 1.
*Some links on this page are affiliate links, marked as sponsored. Where a partner pays us, our view of the product does not change.*