Additional input lag with new Geforce driver 337.50?

[war1]

ok, so im off course here? im on winxp though, so no aero. When i drag an icon, the cursor lags behind the icon? is that what you mean?

the window desktop's cursor position is determined at the start of each refresh cycle. therefore the cursor has no tearing AND minimal lag at the top of the screen.
by minimal lag, I mean that when the screen updates, it is displaying the latest possible information.

compared to the top, at the bottom of the screen the cursor has slightly less than one frame of lag. that is why I asked you whether you could feel the difference.

is that basic gui, or was it done with tearing earlier in time(driver)/without driver(no driver, no lag on winxp)? (it might be what i feel different, if it was done with tearing ealier, bottom of screen about one frame lag)

true fullscreen games, such as those which can show tearing when vsync is off, render directly to the display and bypass the aero compositor.

Does that include older games which uses mouse positioning from desktop cursor(sorry if thats out of place to ask about)

That osx drawing of mouse with lag does not seem like a good idea imo, but i might be the minority here who will take tearing and stutter over smoothed and delayed.

I greatly appreciate you having the patience with me flood, and taking the time to engage in the conversation with me.

[flood]

you're welcome; I was very interested in this myself a few years ago

afaik: if you have any drivers installed, the cursor position is determined at the start of each refresh cycle (minimal lag at top of screen, 1frame lag at bottom).

if you have no drivers installed, the cursor position is determined as everything is drawn. (minimal lag at top and bottom of screen).

on xp yes there is no aero or vsync on the desktop. actually in xp, dragging icons and stuff happens with minimal lag, assuming the system isn't running anything heavy. so the cursor will follow the icon perfectly at the top of the screen, but lag behind the icon at the bottom by one frame.

I have no idea about older games; do you have any one in particular in mind?

[war1]

I think we've hit something on the head here. If older drivers and no drivers acted the same way = no noticable lag, and now new driver updates with refresh rate, resulting in a 1frame lag near the bottom of the screen, that might be the input lag im feeling.

Id games prior to quakelive use get_cursor position afaik, plus all the games that use id tech engines, basically a lot of fps games. I believe a lot of games prior to windows xp which need acceleration fix are all affected by this.

Assuming no other buffer or scaling delay from nvidia driver, you have single handedly solved the nvidia input lag mystery.

Even if the game can get tearing from vsync off, but gets it's mouse input from windows cursor, it will be affected by the drawing method, but theoretically only half a frame(centers cursor in middle of screen)? or is it completely in my head, and the moment the game orders nvidia driver to go fullscreen with tearing, cursor is only affected by fps instead of refresh?

EDIT: I just did the icon dragging, top of screen, it follows the icon perfectly, bottom of screen the mouse cursor lags behind icon! is that effect only when dragging the icon, or is that exactly what i feel all the time when just moving cursor?

[flood]

I think we've hit something on the head here. If older drivers and no drivers acted the same way = no noticable lag, and now new driver updates with refresh rate, resulting in a 1frame lag near the bottom of the screen, that might be the input lag im feeling.

i don't think older drivers act the same way as no drivers. but i dont have any card that supports the older drivers so i can't check.

Id games prior to quakelive use get_cursor position afaik, plus all the games that use id tech engines, basically a lot of fps games. I believe a lot of games prior to windows xp which need acceleration fix are all affected by this.

Assuming no other buffer or scaling delay from nvidia driver, you have single handedly solved the nvidia input lag mystery.

Even if the game can get tearing from vsync off, but gets it's mouse input from windows cursor, it will be affected by the drawing method, but theoretically only half a frame(centers cursor in middle of screen)? or is it completely in my head, and the moment the game orders nvidia driver to go fullscreen with tearing, cursor is only affected by fps instead of refresh?

not sure... if getcursor really is the problem by working in the way you suggest, well it should be a REALLY bad problem because the cursor is effectively updating at 60hz...........

EDIT: I just did the icon dragging, top of screen, it follows the icon perfectly, bottom of screen the mouse cursor lags behind icon! is that effect only when dragging the icon, or is that exactly what i feel all the time when just moving cursor?

it's what is there all the time.

[flood]

btw something else weird with nvidia:

when you have vsync off, and you drag something around with the mouse, you get tearing. if you have a 500 or 1000hz mouse, there are many many tears every frame because the mouse position updates several time over a refresh period.

the expected behavior is that the tear lines are all over the place. indeed, this is what i've seen with my gtx 460 and my laptop with integrated intel graphics. but for some reason, in my newer gtx 750ti, when I run windows 7 with aero disabled, that isn't the behavior. basically, at 60hz, the top half of the screen doesn't tear, but the tear lines are all concentrated at the bottom of the screen. i have no idea wtf is going on with that, but fortunately the tear lines behave normally in fullscreen games.

[war1]

i don't think older drivers act the same way as no drivers. but i dont have any card that supports the older drivers so i can't check.

I might have to go check older drivers now with the icon behavior, and report back.

not sure... if getcursor really is the problem by working in the way you suggest, well it should be a REALLY bad problem because the cursor is effectively updating at 60hz...........

It does some kind of in between updates calculation, uses how far cursor moved, then resets to middle of screen. Some older games where you alt-tab out of, but still have the cursor captured, you'll see that constant recenter of cursor, but now that you mention it, it's affected by upping polling rate on mouse, but so are the desktop cursor? I'm not too sure about the works of it, and my explanation is just roughly how it works.

it's what is there all the time.

then it's definitely what i feel when moving the mouse on the desktop compared to no driver.

btw something else weird with nvidia:

when you have vsync off, and you drag something around with the mouse, you get tearing. if you have a 500 or 1000hz mouse, there are many many tears every frame because the mouse position updates several time over a refresh period.

the expected behavior is that the tear lines are all over the place. indeed, this is what i've seen with my gtx 460 and my laptop with integrated intel graphics. but for some reason, in my newer gtx 750ti, when I run windows 7 with aero disabled, that isn't the behavior. basically, at 60hz, the top half of the screen doesn't tear, but the tear lines are all concentrated at the bottom of the screen. i have no idea wtf is going on with that, but fortunately the tear lines behave normally in fullscreen games.

it sounds more complicated now. I'm not particularly sensitive to tear lines, so seeing them on anything where its not a running tearline e.g a 60fps game on 60hz display, is very hard.

if you have a 500 or 1000hz mouse, there are many many tears every frame because the mouse position updates several time over a refresh period.

so the dragged content is following the mouse update and making tears, but cursor is still on refresh rate cycle and lags? is that correctly understood?

I have to have a look at how older drivers behave with icons, because i still think that lagging cursor is only on newer drivers, i'll report back.

[flood]

so the dragged content is following the mouse update and making tears, but cursor is still on refresh rate cycle and lags? is that correctly understood?

yup

It does some kind of in between updates calculation, uses how far cursor moved, then resets to middle of screen. Some older games where you alt-tab out of, but still have the cursor captured, you'll see that constant recenter of cursor, but now that you mention it, it's affected by upping polling rate on mouse, but so are the desktop cursor? I'm not too sure about the works of it, and my explanation is just roughly how it works.

oh man yup I remember that from cs1.6

[war1]

ok, DAMN! bad news! I really thought you had cracked the problem here, unfortunately old driver has exactly the same lagging behind cursor, but still feels way more crisp for some reason(175.19) it seems less smoothed, like it moves free and faster.(it's still related imo, even though they look to behave the same)

Driverless is pretty funny though, the cursor is faster than the content, and is going ahead of the dragged content.

I'm really bummed out about it though, i really though i would see the same behavior as driverless rolling the driver back and that would be definitive proof of the different feeling. BUT what about the idea that the lagging behind is ONLY when dragging on the old driver, because if i focus on cursor when dragging(old driver). it feels like it's more lagged than when not dragging. I'm convincing myself for now that's the difference.

driverless = no lagging of cursor, even when dragging.
old driver = lagging of cursor, but ONLY when dragging.
current drivers = lagging of cursor, even when not dragging.

[war1]

I thought about it some more (slept on it) and I think i can describe it even more clearly now.

The fact that mouse is now drawn with the refresh rate makes it delayed one frame at the bottom of the screen = "input lag"

The different feeling of the mouse is due to this also, on top of the delay. If position is only drawn with the refresh rate, you will loose some movement, or speed, on top of the delay(waiting to draw maybe?). The look of the the cursor is still bound to polling rate though(smoothness), you still see a difference from 125hz to 500hz polling.

This explains what i've seen on other gaming forums. Some members in the cs1.6 community are complaining about a different feel, completely ruining the game, and i found some guy having made a raw input hook(Rinput.exe) because of this. I(we) at the quake community describe the input lag(or call it that), i've seen ppl call it about 10ms(which makes sense about half a frame when drawing is done in the middle of the screen)

How is that sounding, could this be the explanation of what's going on?

edit: a bit more reading around it's games that use WM_MOUSEMOVE(cursor), instead of WM_INPUT(rawinput) which was only available after windows xp, that are affected. It goes a lot deeper though, ppl are blaming microsoft(maybe rightly so) on neowin forums(win8), and the CS:GO crowd are having huge troubles with cs:go's rawinput and blaming valve.

Back in the day, the only trouble we had with mouse input was acceleration(easy fix), now, it's a whole lot more complicated, when it shouldn't be.

[flood]

it would be really nice if somebody with a mouse+led rig could test this

personally i cant tell the difference between rawinput on vs off in csgo. does that only affect mouse motion or also clicking?