240hz Acer KG251Q Input Lag

[Chief Blur Buster]

Gotcha -- a regional variant.

You literally need a geek decoder ring for decoding those Acer "bmiipx" suffixes.

[adam10603]

Right now I think I'll send it back, and go for an Alienware AW2518HF.

Btw, do you think there could be a difference in total input lag between the AW2518H and AW2518HF (G-Sync/Freesync variant)? I found a test on Tom's Hardware that shows a really low input latency for the G-Sync one, but I can only get the Freesync version within my budget.

[Malinkadink]

Right now I think I'll send it back, and go for an Alienware AW2518HF.

Btw, do you think there could be a difference in total input lag between the AW2518H and AW2518HF (G-Sync/Freesync variant)? I found a test on Tom's Hardware that shows a really low input latency for the G-Sync one, but I can only get the Freesync version within my budget.

https://www.youtube.com/watch?v=9VbVWY1-FPc

They're both basically the same but the freesync one has more options for calibrating. It really depends on whether you want to have gsync or not and are willing to pay for it.

[adam10603]

Right now I think I'll send it back, and go for an Alienware AW2518HF.

Btw, do you think there could be a difference in total input lag between the AW2518H and AW2518HF (G-Sync/Freesync variant)? I found a test on Tom's Hardware that shows a really low input latency for the G-Sync one, but I can only get the Freesync version within my budget.

https://www.youtube.com/watch?v=9VbVWY1-FPc

They're both basically the same but the freesync one has more options for calibrating. It really depends on whether you want to have gsync or not and are willing to pay for it.

Thanks. I can only afford the freesync one, that's why I was asking. I don't mind not having adaptive sync, but I really want low input lag after being unlucky with this Acer.

[phatty]

Your test was simple but there's no proof that actuation is occurring at the exact same time. Position of the finger seems to be the same, but there's no way to tell since they are in the air and 4 ms mind you is faster than trigger time in the olympics which is 10 ms.

Your range of error is too high imho to make a determination on the input lag definitively.

[Chief Blur Buster]

Your test was simple but there's no proof that actuation is occurring at the exact same time. Position of the finger seems to be the same, but there's no way to tell since they are in the air and 4 ms mind you is faster than trigger time in the olympics which is 10 ms.

First, welcome to Blur Busters!

Actually, I have something to say about that. It's apples versus bananas.

I paid a peer-reviewed researcher to do some research on eSports reaction times, and here's Blur Busters Human Reflex Article

Now, I have further things to add.

(1) The "Pass-the-finish-line" Effect
An Olympics athlete still wins passing the finish line 2ms before his competitor, even if they don't know it until they see scoreboard. Two persons have the same reaction time. They start at the same time. They finish at the same time. It's a tie (which occured in the early 20th century until stopwatches and finish-line cameras became accurate enough to break ties). By the straightshoot math, assuming no variance in reaction time: If one person starts 2ms sooner because they heard the starting pistol 2ms sooner than their competitor, they'll win the race. Now reaction times do vary, but it still averages out over time. Now, the reaction time spread is extremely tight in eSports; Championship Player #1 through Player #4 may have a reaction-time spread of only 20ms apart, with many situations of reactions only mere milliseconds apart, often less than 4ms for certain kinds of stimuli, like turn-corner, see-each-other, shoot-simultaneously, the display with the less lag will be the statistical tiebreaker in equal-reaction. While tickrate granularity (e.g. 128tick CS:GO servers) muddies this, it still statistically averages over time, 4ms less lag in a display means 50% chance the frag event occurs in a earlier 1/128sec window. There's no way for a game to pre-handicap a player that has a less laggy display (of the same refresh rate), so a lower-lag display is a sure way to tilt tiebreaker reactiontimes into your favour in those eSports-tight reactiontime spreads. That's "Crossing The Finish Line Effect"!
CONCLUSION: You do not need to see/feel the millisecond to win the millisecond

(2) The Lag-Training Effect
At 4000 pixels/second, shooting 4ms later to a stimulus means 16 pixels overshoot. 4000(pixels/sec) / 1000(ms in a second) x 4(ms) = 16 pixels. You're always trained to a specific lag of a specific display, so when latency changes even by a mere 4ms, a seasoned player can definitely feel things are off -- they're making fewer long-distance hits, and wonder why, they can't feel the lag, they can't see the lag, they just score less because lag is not the same as the lag they pre-trained at. A flick-turn immediately followed by a frag, the screen horizontally pans at many thousands pixels per second. One screen width per second is 2000 pixels/sec on a 1080p display -- see TestUFO Photo Example at 1920 pixels/sec. Look at that animation full screen on your 1080p monitor. That's slower than a turn left/right in a FPS shooter. Many eSports players shoot without stopping their turn. Many are trained to be able to precisely shoot while turning. It's like trying to shoot an archery arrow on a motorized horizontally-moving archery target. Now if you're doing reliable bulls eyes on a specific display. Then you switch to a display with 1ms more lag than your original display at 4000pix/sec, your arrow will hit an average of 4 pixels to the left of the bulls-eye region. The entire scattergraph shifts leftwards by 1ms worth of movement (8 pixels at 8000 pixels/sec, or 4 pixels at 4000 pixels/sec, or 2 pixels at 2000 pixels/sec). I have confirmed eSports players genuinely feel performance changes from the lag-training effect even if they cannot detect the lag change directly. Now, look, the average human have widely varying reaction times. But a pro player can have consistent reaction time precision. That are really tight (varies only 10ms-20ms from a specific stimuli). Now if the bellcurve of the reaction time precision is a tight 4ms spread, such elite seasoned players definitely feel misaiming problems from a 4ms lag change even if they cannot directly detect the lag change.
CONCLUSION: You do not need to see/feel the millisecond to feel your performance has improved/degraded from mere milliseconds lag changes

I haven't even gone into spinoff things of why milliseconds matters; some much more human-detectable -- such as frame visibility effects. 240Hz displays can display refresh cycles in only 4.166ms while 120Hz displays can display refresh cycles in 8.333ms. The difference between 120Hz and 240Hz is human visible. There can be different signal processing lag and other factors so the worst 240Hz display can be worse than the best 144Hz displays. But clearly, 1/120sec duration refresh cycles versus 1/240sec-duration refresh cycles, are clearly human-visible -- the time differential of 1/120sec and 1/240sec is only 4.1666ms! Yet nearly everyone can tell apart the motion clarity of the two in TestUFO -- and often competitive players notice this in ultra-high-framerate games.

That's only three or four nuances so far. I have more to go on.
Shall I keep going on why milliseconds matters here at Blur Busters?
Sure they matter less for some, and more for others, but you have to dig the dozens to hundreds of other nuances of the milliseconds -- including direct & indirect benefits -- other than the nuance you're speaking off.

Most light scientific studies of players can't punch through all the noise of non-professional non-olympic reaction times. They often pull students off the street to play a game and record their reaction times. But the calculus changes when you recruit real seasoned eSports players (especially paid champions) and try to measure their reaction times more closely to a real olympics athletes. Surprising data starts to show up when Player A and Player B have near-glassfloor reaction time spreads AND fairly equally matched.

P.S. I am recruiting peer-reviewed researchers for additional study/research on these topics. We're revealing unexpected human-relevant surprises lurking in the humble millisecond -- whether it be GtG or MPRT or reaction times or lag -- Contact me [email protected]

[phatty]

This is fascinating stuff.

I had always assumed that the false start in olympics of 100 ms was because it was physiologically impossible to push off from hearing a gunshot at 100 ms. While there is some lead time issues with the speed of sound and the actual speed of the neurons passing down the twitch response, I imagine that this number was something solid.

Granted APMS in starcraft players show that they can have sub 5 ms clicks.

I think that with regards to point 1, a person that obtains information faster by all means will have more lead time to react to it. All things being equal 4 ms additional processing time is always an absolute headstart of 4 ms.

With regards to point 2, I don't doubt that a person can "feel" they are playing worse, but I think that there are many reasons to play worse when you get a new display - to assume that it is input lag as opposed to something else, as subtle size change from 24 inch to 24.5, distance from the monitor that a person is sitting, etc.

In this specific case, testing the input lag with a hand in the air pushing a button doesn't ensure accurate actuation so a mere 4 to 8 ms can be within the range of expected variance from physiologic response times. Fingerpad pressure / depth / etc

Some of the prior models of this monitor have had low input lag results, but I'm not sure to their validity.

For example the 144hz version of this line has a input lag of ~ 5ms.
https://www.displayninja.com/acer-kg251qf-review/

[Chief Blur Buster]

Indeed, so many lag nuances like key travel depths, mouse poll rate (and mouse antibounce filtering lag), game engine variability, and all. They still build up.

But what happened is a lot of weak links disappeared, to the point where a single 4ms single-device change, actually matters a lot in the paid six-figure-money-prize eSports leagues....

-- we now got 1000Hz mice
-- we now got games with 10ms full-chain "buttons-to-photons" times (CS:GO on some 240Hz displays)
-- we now got games with ultrahigh framerates (300fps = 3.3ms frame granularity) such as CS:GO
-- we now got 240Hz displays

Our 2017 GSYNC 101 tests, done by Jorim Tapley, had 12ms button-to-photons results, but we know some testers who achieved 8ms-9ms from the time a mouse button is pressed (LED illumination in high speed camera of a modified computer mouse designed to illuminate a LED, followed by actual screen reaction like a gunshot muzzle flash -- with a high speed camera, the spread is now single-digit milliseconds on the fastest computers on some popular games). That is punching the full chain of ALL OF THE NOISE -- including software & hardware -- from mouse button to display. Such tight button-to-photons lag pushes other lag factors well above the noisefloor.

We were the world's first website to discover a way to measure the input lag of GSYNC (GSYNC preview in 2013), and the test method is now being used to measure button-to-photons latency via a high speed video camera. Newer tests by multiple sites show the fullchain achieved in single-digit milliseconds -- that's all the mechanicals, all the software, all the drivers, all the cable lag, all the CPU/GPU/monitor processing lag -- button to photon time is determined already -- So we know how tight the noisefloor is already. Button-to-photons is a catchall, and it's already very tight on 240Hz CS:GO running at >300fps with a 1000Hz mouse, to the point where 4ms is a big statistical hole that you can drive a truck through if you're trying to win a $100,000 eSports trophy with competitors of nearly-matched reaction times. Eyeroll to many, but important to others. The millisecond has earned my 100% respect, however.

The noisefloor is always falling, falling, falling. So new surprises of formerly "lost-in-noise" milliseconds are becoming visible as tech progress continues. With a fast moving noisefloor, I'm losing my balance... PUN! Ha! So, the rule of Blur Busters nowadays, is to keep an open mind about the various future surprises of the humble millisecond.

Keep an eye on my Holiday 2018 follow-up to my Holiday 2017 article, Blur Busters Law And The Amazing Journey To Future 1000Hz Displays.

[adam10603]

Your test was simple but there's no proof that actuation is occurring at the exact same time. Position of the finger seems to be the same, but there's no way to tell since they are in the air and 4 ms mind you is faster than trigger time in the olympics which is 10 ms.

Your range of error is too high imho to make a determination on the input lag definitively.

Sorry for the late response, haven't checked this thread in a while.
I agree that I have a pretty high margin of error, since I only had 240fps footage, but that's why I did dozens of clicks with both monitors, and dozens of mouse bumps, and the results were pretty much the same every time. Not once did this Acer come out on top, or even tied with my old monitor, so I didn't just base this off a sample size of 1. That's why I'm pretty confident saying it had more input lag.
Also there's the fact that I could tell right away that something was wrong with it. Before you say placebo though, this happened before I had done any testing, or even gave it any thought. I never would have expected above average input lag from it, so I wasn't even looking for any difference. I just fired up CS:GO at ~400fps to see what 240Hz is like, and I could immediately tell there was sort of a disconnect between my mouse and the screen when doing fast flick shots, even though it looked very smooth. That's the very reason I started testing it, which just confirmed what I felt.

On a different note, I got a refund, and I had picked up an Alienware AW2518HF since then, and I couldn't be happier. Did all the same tests, and the input lag on this one is definitely lower.
Between bumping the mouse and first change on screen (on 240fps slow mo) it's sometimes 2, sometimes 1 frame, whereas that Acer was 3 to 4 frames in the exact same test (same mouse, same everything).
This Alienware also has a much more rigid stand with a lot more adjustments, a better OSD imo, a USB 3.0 hub, so it's just a nicer monitor all around, and of course very low input lag. Definitely worth it over that Acer, I'm very happy with this.

[phatty]

I have both monitors actually(gsync varient of the 2518), the Predator IPS 165(oc) 27 inch 1440p, and the dell 2716dg

I don't actually perceive the difference in the input lag as you do subjectively, but I am over 40, which probably physiologically makes a difference.

The testing to me is just too inaccurate as 1 frame has a physiologic variance that objectively can be within the range of error.

Just not precise enough, but it may be accurate (as it has been reproducible).

There are are issues as optimal settings, the correct drivers as each monitor has their own driver specifically, different games, etc - video card - freesync / gsync / etc.

While all things being equal may introduce the extra 4 to 8 ms on the 240 hz monitor, then we come down to the cost, as the Acer is available at 229 at micro center, the comparable alienware freesynce is usually at around 270 to 280 on sale. The Gsync variant - which I have - is actually 350 when on the best sale.

The stand on the alienware is much much better. I agree it is overall the better monitor.

I'm not trying to discredit blurbusters but to give some context:

https://displaylag.com/testing-method/

Excellent
This rating indicates the absolute best of the best; that a display exhibits unparalleled performance for input lag and response time. If you are the type of gamer that demands the fastest display possible, this is what you should look for. Due to the nature of this requirement, only displays ranging from 0-20ms are applicable to fit this category. Please note that there can be a 1 frame difference at 60hz between displays in this category. Stick to 10ms or lower if you want the best performance at 60hz.

This data base is a bit outdated:

https://displaylag.com/display-database/

If you select for 144 and 165 hz monitors, you get the gaming monitors from 2014 and onwards, you'll see most rest around 10 ms to 12 ms. If the acer is in fact 1-2 frames slower, we're looking at 20 ms in reference (upper range) this these averages.

Another reference:
https://www.rtings.com/monitor/tests/inputs/input-lag

If your input lag from the gsync variant is 1-2 frames, it would suggest to be around 6 ms, and then acer is running at 3 to 4, which would imply around 14 ms.

One could generally extrapolate the input lag to probably be between 14 ms to 20 ms, which I agree would be a bit slow for a pro gamer series monitor.

It is a budget 240. I wouldn't necessarily throw this out for at $229.

As a caveat:

https://www.displayninja.com/acer-kg251qf-review/

This site cites the lower model at 144 hz to be 5 ms input lag, which while the 240 hz panels are different and the internal design may be different as well, it does lead me to believe the the input lag would be comparable to most gaming grade monitors.

Would I buy the freesync Alienware @ 279 over the acer? Yes (mainly for the stand). If you don't care about the stand, then this is a decent choice if you are really strapped for cash.