Does scanline sync really have more input lag than Adaptive Sync?

[betonKruglosuTotchno]

Here's what I've seen in two tests.

https://youtu.be/fAEBVloCV_0?t=2175 - test for 360Hz LCD, frame pacing is 2,7 ms . Scanline sync is almost 5 ms worse than adaptive sync.

https://youtu.be/8ZRuFaFZh5M?t=476 - test for 60Hz LCD, frame pacing is 16 ms . Scanline sync is 30 ms worse than adaptive sync.

Meanwhile, I have recorded a trace of scanline-synced Half Life 1 and I do not see how it's possible that scanline sync lags so much. Can you please explain why these two benchmarks show such a huge disadvantage?

[Chief Blur Buster]

1. Did you use VSYNC OFF with scanline sync?

2. Did you verify that tearline appears when you recalibrate the scanline sync scanline number?

Incorrect calibration of RTSS Scanline Sync will result in terrible latency, so it's critical to watch a stationary tearline appear and then adjust its position to almost exactly right above the top edge of the screen (not right below bottom edge of screen), usually using negative indexes.

FACTS:
1. Lag is always highest right above a tearline
2. Lag is always lowest right below a tearline

The latency difference above/below a tearline is always one full frametime, so at framerate=Hz, you've got a 1/60sec latency difference on both sides of a VSYNC OFF tearline.

So you must always use VSYNC OFF with Scanline Sync, and position the tearline to the top edge of the screen (bottom-of-VBI syncing).



If you've never seen tearing with Scanline Sync, always intentionally make tearing appear with Scanline Sync, then calibrate the tearing to disappear above the top edge of screen.

Some reviewers don't test this properly, I haven't seen the videos, but this is a common mistake with Scanline Sync.

[betonKruglosuTotchno]

I fully understand what you say. As for me, I do not have a measurement device and an adaptively syncing monitor to verify those findings those people present in benchmark.

>you've got a 1/60sec latency difference on both sides of a VSYNC OFF tearline.

Yep, and those benchmarks clearly show almost double the latency difference, so it does not fully explain this.

[betonKruglosuTotchno]

>but this is a common mistake with Scanline Sync.

Can you point to some correct comparison with measurements?

[Chief Blur Buster]

I fully understand what you say. As for me, I do not have a measurement device and an adaptively syncing monitor to verify those findings those people present in benchmark.

>you've got a 1/60sec latency difference on both sides of a VSYNC OFF tearline.

Yep, and those benchmarks clearly show almost double the latency difference, so it does not fully explain this.

There can be other factors that add another frame. Using a non-VSYNC ON technology with certain settings with Scanline Sync can add multiple frames sometimes. Also, tester location can affect things. The bottom edge of screen is usually 1 refresh cycle lag to top edge of screen, so position of photodiode tester can have a big effect on VSYNC ON latency differences, as not all pixels refresh at the same time, high speed videos at www.blurbusters.com/scanout

I'll let others chime in for sources -- Although I'm the one who triggered Guru3D to add Scanline Sync to RTSS, this is not my specialty this year as I'm focussed on other things at the moment.

[betonKruglosuTotchno]

In short though, should Scanline Sync have almost same input lag as adaptive sync or should it be at least one frame interval worse? This forum post says that scanline sync has inherent lag in it.

viewtopic.php?p=56699#p56699

[betonKruglosuTotchno]

This post claims that scanline sync adds zero latency compared to capped unsynced 60fps on 60 Hz monior:

viewtopic.php?p=37558#p37558

[betonKruglosuTotchno]

SpecialK devs told me their Latent Sync is already better than RTSS Scanline Sync.
https://github.com/SpecialKO/SpecialK/issues/57

[betonKruglosuTotchno]

Devs made new page stating that Latent Sync is almost equal to adaptive sync.
https://wiki.special-k.info/en/Advanced ... atent-sync