GPU scaling and input lag?

[Chief Blur Buster]

Side note though the Asus has noticeable input lag where the BenQ did not.

It may be also possible the ASUS VG248QE used frame-buffered scaling, and the BENQ used real-time scaling. Also, at least when not doing scaling, I know the VG248QE and the XL2411T has the same input lag.

Scaling is something that can be done in virtually real time (single scan-line buffered) and thus should not add a single millisecond of input lag at all.

[calcifer]

Someone know how to force display scaling with a nvidia GPU (GTX 680), a BenQ xl2420T and windows 8.1? I can only choose GPU scaling in nvidia driver (i can choose display scaling with windows 7)

[calcifer]

Side note though the Asus has noticeable input lag where the BenQ did not.

It may be also possible the ASUS VG248QE used frame-buffered scaling, and the BENQ used real-time scaling. Also, at least when not doing scaling, I know the VG248QE and the XL2411T has the same input lag.

Scaling is something that can be done in virtually real time (single scan-line buffered) and thus should not add a single millisecond of input lag at all.

Thank you for your response.

Maybe you could do some input lag measurement, like you have done for g-sync, with your led/mouse method?

It would be really nice to know how much input lag GPU scaling is adding (lot of competitive gamers use 4:3 or lesser resolution than native in games like CS:GO or quake live) and if it is worthy to return to windows 7 because nvidia cards force GPU scaling with all BenQ gamer monitors, and maybe asus too, with windows 8...

Maybe you could also compare input lag between AMD cards and nvidia cards! This would be really interesting because lot of people say that there is more input lag with nvidia cards than with AMD. This could be a really important information for competive gamers!

[sharknice]

Side note though the Asus has noticeable input lag where the BenQ did not.

It may be also possible the ASUS VG248QE used frame-buffered scaling, and the BENQ used real-time scaling. Also, at least when not doing scaling, I know the VG248QE and the XL2411T has the same input lag.

Scaling is something that can be done in virtually real time (single scan-line buffered) and thus should not add a single millisecond of input lag at all.

Thank you for your response.

Maybe you could do some input lag measurement, like you have done for g-sync, with your led/mouse method?

It would be really nice to know how much input lag GPU scaling is adding (lot of competitive gamers use 4:3 or lesser resolution than native in games like CS:GO or quake live) and if it is worthy to return to windows 7 because nvidia cards force GPU scaling with all BenQ gamer monitors, and maybe asus too, with windows 8...

Yeah I find it funny they do that, I can't stand looking at the distortion, but I totally understand how it is easier to aim with a 70 FOV instead of 90. They should just let you change the FOV between a set range like every other game does.

[RealNC]

Yeah I find it funny they do that, I can't stand looking at the distortion, but I totally understand how it is easier to aim with a 70 FOV instead of 90. They should just let you change the FOV between a set range like every other game does.

Create a custom resolution that keeps the native resolution of the monitor. For 1080p monitors, you need 1440x1080. This will give you a perfect 4:3 ratio 1080p mode. There will be no scaling at all, just cropping of the left and right sides.

I tried this with CSGO and it works. 4:3 is unplayable for me, but if you actually like horse blinders, then this will help

[Berserker]

How I feel:
GPU - Aspect ratio - most latency, harder to aimbot between targets, fast flickshots in QL connect really good, on asus vg248qe it feels fine, like + one refresh latency.
GPU - No scaling - less latency, but flickshots dont connect
Display - Aspect ratio - dont remember exactly, got tired of installing/uninstalling refreshes with strobelight.
Display - No scaling - least latency, easy to aimbot between targets, but fast flickshots dont seem to connect, always miss by few %.

When mouse reacts too fast I miss, maybe due to how I aim and high static friction of my mousepad, I also snipe much better with 3 frames to render ahead than 1 in TF2, or with mouse with more latency from sensor, but not from high graphical settings that have more latency. Kinda weird.

[klexen]

Is there anyway to force or hack the driver to keep the display scaling option?

There are a few other people with this issue as well. I posted about it on Nvidia forums, but that go me nowhere.

https://forums.geforce.com/default/topi ... 1/#4116986

I figure this would be the best place for a solution.

I was able to somehow trick the driver into giving me the option again by lowering my resolution and refresh rate, then unplugging and plugging my monitor power back in. When I did that it made the driver think my native resolution was 1280x960 with 85hz refresh.

After updating my video driver though it reset it, and now I can't get it to think my native resolution is lower, so I can't get the display scaling option again.

[RealNC]

harder to aimbot between targets

Are you a damn cheater? Freudian slip?

[HeLLoWorld]

A modern gpu can do a bilinear blit unbelievabling fast, no number right now but I believe it is orders of magnitudes less than 1ms, and if done properly it would only affect latency not frame throughput. Someone thinking it would add time due to proceesing power has no idea of the amount of computations going on in the 8 or 16ms it takes to render a modern game frame.
Besides I see zero valid reason why it would add one frame or one refresh of latency.
If gpu scaling indeeds measurably affects latency then it is due to suboptimal implementation.

[klexen]

I would love to know if and how much input lag is caused by GPU scaling vs Display scaling. I can't tell you how much time I've wasted trying to make it work.