Looking for People to Test a Possible Fix or atleast find a starting point — Controlled Experiment

[MontyTheAverage]

There was a CPU intensity setting in older OCCT builds. I've switched to using VRAM and 3D adaptive after that.. Muscle memory, mb. I made changes in the main thread.

I tried the CPU setting but it just freezes the PC and game doesn't start. VRAM and 3D works. And I do think it improves the game. Not fully but I can tell the hit reg and desync issues are little less

[Ven]

There was a CPU intensity setting in older OCCT builds. I've switched to using VRAM and 3D adaptive after that.. Muscle memory, mb. I made changes in the main thread.

I tried the CPU setting but it just freezes the PC and game doesn't start. VRAM and 3D works. And I do think it improves the game. Not fully but I can tell the hit reg and desync issues are little less

That's a good start! If you're already noticing some improvement in hit registration and desync, try experimenting a bit more to see if you can find the sweet spot. Adjust the OCCT intensity and your in-game graphics settings and see how the results change. The required intensity seems to vary depending on the game and settings, so try different combinations and compare. Let me know if you find a combination that works better!

[ablemor]

The problem is often that the placebo effect plays tricks on us—for instance, if the opponent is weaker and doesn't react the same way, our brain expects whatever we're trying to work. This needs to be tested further no matter what we do.

Anyway, there would be a much easier way to do all of this, we just need a few dollars.

[Ven]

The problem is often that the placebo effect plays tricks on us—for instance, if the opponent is weaker and doesn't react the same way, our brain expects whatever we're trying to work. This needs to be tested further no matter what we do.

Anyway, there would be a much easier way to do all of this, we just need a few dollars.

I completely understand the placebo concern. When we're dealing with issues like these, it's easy to mistake expectations for actual improvements. I've also seen people suggest all kinds of solutions and troubleshooting methods over the years, only for them to eventually fade away without any conclusive results. I'm fully aware of that, which is exactly why I spent around 3–4 months testing this before posting here.

Over that time, I've noticed fairly consistent results, particularly a reduction in desync and some improvement in hit registration. That doesn't mean placebo isn't a possibility, though.

That's precisely why I made this thread. I want multiple people to try the experiment so we can compare results across different systems and see whether the effect is actually reproducible. We need more testing before drawing any conclusions, and hopefully we can establish that together.

[MontyTheAverage]

The problem is often that the placebo effect plays tricks on us—for instance, if the opponent is weaker and doesn't react the same way, our brain expects whatever we're trying to work. This needs to be tested further no matter what we do.

Anyway, there would be a much easier way to do all of this, we just need a few dollars.

No placebo here. Something is definitely happening here and it is very easy to tell after you deal with the desync and input lag for so long and you can notice every noticeable difference. The difference in hitreg, incoming damage and TTK and TTD are all noticeable.

[MontyTheAverage]

It definitely does something with the desync. It is same as when I click record with NVIDIA recorder or changing resolution or something tweak, same thing happens where it minimizes the desync but for some time like a match or two. It seems to always come back though. I also notice it coming back running this test. It worked for a match and a half but I felt it return in the second match. I was getting lasered, dying fast and hit reg being shitty once again in the middle of the second match. Now, why does this happen? If we can figure it out, would really make some advancement to figure this whole thing out. And also the problem is it is never consistent. Something works to mitigate the desync and other issues, but it doesn't seem to work again or as consistently. It's like playing tricks on us.

[Ven]

This is exactly the kind of feedback I was hoping to get. One thing I've noticed with this method is that I can reproduce the effect almost every time I run the test. It's not like some tweaks where the desync temporarily improves, fades away, and you can't seem to get the same result again. Even if the effect wears off after a few matches, running the test again seems to bring the improvement back.

Maybe try adjusting the OCCT intensity if you notice the desync returning after a few games. The sweet spot might change depending on the game and its settings, so it's worth experimenting with that too.

I agree that the temporary nature of these effects is interesting. The main question is why they happen in the first place and what exactly is changing when we run these tests. If we can establish that the effect is reproducible across different systems, hopefully it'll give us a starting point for investigating the underlying cause.

[MontyTheAverage]

Tried the OCCT + Nvidia recording test on a BF6 KotH match.
Before hitting record I'm 6-11 and team losing bad.
After started recording, 22-17 and we made a comeback and won. Input lag didn't go away but desync decreased, wasn't dying so fast and hitreg improved. Why this causes this change I don't know.

[Ven]

Tried the OCCT + Nvidia recording test on a BF6 KotH match.
Before hitting record I'm 6-11 and team losing bad.
After started recording, 22-17 and we made a comeback and won. Input lag didn't go away but desync decreased, wasn't dying so fast and hitreg improved. Why this causes this change I don't know.

I have a working theory, although I'm not entirely sure about it yet. Around a year ago, we did some extensive CPU/GPU tracing using Windows Performance Analyzer (WPA), and from what I vaguely remember, we noticed some irregularities where the CPU seemed to finish processing game logic, but there was a noticeable gap before the GPU started drawing. I need to revisit those traces to confirm the details.

I'm wondering whether inducing GPU load with OCCT temporarily changes the pacing between CPU-side game processing and GPU rendering, potentially bringing them into better alignment. Intel PresentMon's interaction with the graphics presentation pipeline might also play a role, although I don't know whether its overlay or monitoring is actually responsible.

This is just a working theory, but considering you notice similar temporary changes when starting NVIDIA recording or changing resolution, I think it's worth investigating whether these actions have something in common.

[Apexer15]

Tried to test it out but my handheld PC couldn't handle it it started freezing during the test in game. Set the intensity from 40 to 20 and still same thing. That's unfortunate because I wanted to help and feel some improvement...