• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

How do LCDs handle high input lag?

ChadVader

n00b
Joined
Oct 21, 2007
Messages
59
As people have seen, Input Lag can get way up there, well over 80ms. For an example let's say we bought a monitor and it has 66ms input lag. This is enough time for 4 frames to be displayed (60 frames per second, 16.6ms per frame). As far as I can tell there are only three ways to handle this situation.

1. It stores all 4 frames at once in a huge frame buffer, and processes all 4 frames simultaneously. It processes 25% of each frame every 16ms. Here's an example timeline:

0ms - Frame1 enters queue
16ms - Frame1 (25%), Frame2 enters queue
33ms - Frame1 (50%), Frame2 (25%), Frame3 enters queue
50ms - Frame1 (75%), Frame2 (50%), Frame3 (25%), Frame4 enters queue
66ms - Frame1 (100% and is displayed on screen), Frame2 (75%), Frame3 (50%), Frame4 (25%)

The problem I have with this is, of course, why the hell would you process JUST 25% of each frame every 16ms? Why not put all the processing power into a single frame at a time?

2. The monitor has a frame buffer to hold 4 frames at once, and 4 processors for each of the 4 frame buffers. Each of the processors are only fast enough to finish an entire frame in 64ms. This is the only solution that makes sense. However..... FOUR frame buffers and FOUR processors???? Come on!

3. It takes 64ms to display the 1st frame, so the 2nd, 3rd and 4th frames are dropped.

Do any of these scenarios make sense or am I missing the mark completely? Do we have any experts on input lag around here?
 
However you do it, you end up with the same delay and can only introduce more delay.

If you wait for a buffer to fill up with 4 frames before processing, they still need to go through the processing so it will take longer.
If the buffer is after the processing, waiting for it to fill up will extend the time it already takes too.

The fastest route is to immediately process with no extra delays and to immediately display after processing with no delays.


Dropping frames isnt acceptable.
This does happen on most LCDs if you use a refresh rate higher than 60Hz.
Some of the extra frames are dropped to maintain 60Hz output and it can look jerky.


To answer the first question, LCDs often have different modes of operation to reduce processing and therefore reduce lag.
If you cannot change the processing, you are stuck with the full lag.

Sometimes its a guessing game.
I've just purchased an Atec (Philips/LG) 37" 1080P LCD but I dont know how much lag it introduces without processing enabled, hopefully next to none.
Before buying, I did check that it has numerous processing modes so the lag can be improved or processing disabled.
It was a real bargain and I can sell it without much loss so fingers crossed :)
 
It isn't necessary to process the frames at all. Simply take what's given and stuff it into correct pixel position. That's what the very lowest sub-frame rate input lag monitors do. So, why do so many monitors go to the extra trouble to buffer and process up to 4 frames? I suspect that they're playing tricks by employing temporal filtering. Doing that, they can compare on a pixel by pixel basis, the brightness on successive frames and if a big jump in brightness is detected between frames 2 and 3, they can start changing the brightness on frame 1 ahead of time and they can even "overdrive" the pixel over the course of the 4 frames arriving at the final brightness value on the 4th frame. Example; a given pixel is supposed to change brightness from 30 to 200 from frame 2 to 3. So, to overcome the poor response speed of the display, start driving the pixel to 255 while displaying frame 1. The display won't actually change much in brightness due to it's slow response. Continue overdriving the pixel for the duration of frame 2 and still the pixel has only changed brightness a little. For frame 3, knock it back from 255 to 100 or so and the display will have probably reached the desired 200 level, but you need to keep it from overshooting the mark to 255. Then finally on frame 4 drive that pixel to the actual 200 level. All this is done in an effort to reduce the apparent response time. People are so dead set on getting ultra-fast response time, that many of them don't realize the tricks mfgrs play, to get their monitors to measure fast response time. In the case of a 4 frame buffer, input lag is sacrificed. Not that any mfgr specs input lag. So, most people don't even know it.

O.O.
 
It isn't necessary to process the frames at all.

Thats why many LCDs allow you to disable processing so it is more of a straight through process.
however this wont help change the response and visual artifacts of some LCD panels.
For example, TN's can respond very fast (but have poor colour quality), VA and IPS type screens have been improving but are slower and therefore introduce more ghosting.

Some of the processing is designed to mask the ghosting or lack of colours so while the processing lag is undesirable, the on screen effect of not using the processing can be that it doesnt look so nice.
Always best to choose a panel that will suit your style of gameplay (and dont get one that really is crap).

S-IPS displays are down to 5mS grey to grey which is much better now and suitable for gaming.
This is what I have bought as S-IPS has great viewing angles and very good colour rendition.
The faster panel means I should be able to disable any processing without it looking arse.
 
outerorbit, do some homework before blaming Response Time Compensation for lag.
http://www.national.com/appinfo/fpd/files/48_03.pdf

It is TNs which employ temporal dithering (FRC) and it's used for emulating missing 16½M colours lacked by 6-bit TN panels.

Perhaps temporal dithering for the purpose of increasing color depth, is not the only form of temporal filtering and perhaps another form of it is used to reduce response time and introduces lag due to frame buffering? The cited reference is just ONE method of improving response time that happens to not introduce lag. It may very well be, that is not the only method in common use.

O.O.
 
Back
Top