Paul wrote:Especially when we're entering the 240Hz zone, <1 frame (0 ~ 4ms) of input lag isn't detectable by human senses by any means and if someone claims they can detect it, they're sitting too high on their MLG-branded horse and need to get down ASAP.
There are subtle effects to consider:
Don't forget the "
cross-the-finish-line-first" effect / "
race-to-the-trigger" effect in simultaneous-draw situations (e.g. shooting each other at the same time).
In Olympics sprints, beating the finish line 1ms earlier can mean you win the race.
So you don't need to feel the 1ms difference in order to win with 1ms.
In this situation, you've got skilled humans, some with almost the same reaction times (e.g. let's say, humans with average "X" ms reaction time competing against a human with a (X+1) ms reaction time).
If you can save 2 or 3ms of input lag in your hardware, that can give the slightly slower reacting human ("X" + 1ms) a statistical leg up on the faster reacting humans ("X" ms reaction time). Less input lag in your hardware subtracts from your human reaction time which are often well-matched in the rareified heights of eSports gaming. It will all vary a lot and skills play a huge role, so there's lots of fuzz factor, but over time, many matches and many competitions later, that few milliseconds lag advantage can eventually build up statistically.
Now, if we're talking about old engines like CS:GO and Quake, we run extreme framerates at extreme refresh rates, on a very tight button-to-pixels cycle which can be only a few tens milliseconds long totalled, in high-speed-camera tests. Shaving 2ms-3ms off that can mean more than 10% shorter button-to-pixels cycle in some overkill-systems-with-old-engines situations. That's quite a leg-up on your competitors in well-matched reaction-time situations, even if you cannot feel the 1ms.
Let's say, 5 champion players. Let's say on a specific reaction-time-tester (they vary a lot, due to various added lag) -- imagine these 5 champion players have average reaction times of 160ms, 161ms, 162ms, 162ms, 163ms in a specific reaction time tester. Now with this, 2ms can really give you a leg-up on this very tight spread of human-reaction times of the top leagues. Just like the top 5 Olympics sprinters can have a spread just milliseconds apart, crossing that finish line.
So even if you can't "feel" that 1ms, it can help you "race-to-the-trigger" first in simultaneous-draw situations.
In normal casual FPS gaming, you're competing with all kinds of people with all kinds of reaction times, so that lag advantage does not matter nearly as much.
Scientific studies and random tests on casual/random gamers have shown that already, 2ms lag usually doesn't matter for casual FPS.
But when it comes to the rareified heights of the upper leagues, well-matched reaction times, well-matched elite skills, and very low-lag pipelines (e.g. GTX1080's on CS:GO), that 2ms can begin to make a statistical difference in beat-to-finish-line-situations -- the milliseconds doesn't matter for a high school gym class race, but matters a lot for a 100 meter Olympics race.
If you're the one seeing that single frame sooner, your finger is hitting the fire button equally sooner in identical reaction time situations of the top professional leagues. Two competitors in an FPS game. They run into each other around a corner. They suddenly see each other simultaneously. They shoot at the same time. Both persons have exactly the same reaction time. Who wins? Yup -- input lag begins to statistically matter in these specific situations. Even 2ms.
race-to-the-trigger, just like a 100 meter Olympics race.
You don't have to feel the 2ms to win with 2ms.
Limitations in games (game tick rates, microstutters, mouse pollrate limitations, pauses from disk access, rendertime hiccups, etc) start to fudge around all the lag statistics quite a lot at the millisecond levels. Sure, there's rounding effects due to game tick rates and frame refresh cycles, but, your human eye retinas are still seeing refresh cycles 2ms sooner and your button press transmitted to the server -- might end up rounding-off to the previous tick cycle. Potentially fragging your enemy one full tick cycle before them, even if the tick cycle is closer to 10ms. (e.g. 64Hz or 128Hz tick rate). And depending on engine, that may mean sudden rounding-off to identical (0ms difference, unseen) versus occasional huge difference (1/64sec difference, visible to human eye). Which means the 2ms occasionally amplifies to human-visible differences based on rounding-off effects, depending on how certain games handles it. Humankind is full of unexpected surprises including the ability to tell apart 1ms strobing from 2ms strobing (
LightBoost 10% vs 100%) and ability to
tell apart a 500Hz vs 1000Hz mouse poll rate -- there's possibly hidden effects that occasionally amplifies 2ms input lag into something human-visible -- such as noticing muzzle flashes becoming visible 15ms earlier about 20%(ish) more frequently because you had 2ms less lag. But we're not talking about that here, really (it might not be happening, but again -- there are many 'unexpected effect' factors that are always continually being discovered) -- milliseconds can matter anyway even if not perceived due to the finish-line effect.
In modern graphics heavy games on midrange games, 2ms is easily totally lost in the noise of a very jittery 100ms buttons-to-pixel pipeline. Especially when matched against plainly casual gamers with highly variable reaction times. Over packet-jittery Internet connections?
But
CS:GO with 4Ghz + 240Hz + GTX1080 + VSYNC OFF + SSD + LAN + >500fps ? And playing in the top leagues (few variances in reaction times)? This is
when 2ms start statistically affecting results. Well-matched skill/reaction situations for simultaneous-draw situations, even if you cannot feel the 2ms.
Sure, most of the time, 2ms doesn't matter. But that neglects the fact 2ms is important in the 100 meter Olympics sprint race.
And also in eSports situations where there are
race-to-the-trigger situations of simultaneous-draw. Even if you can't feel the 2ms.
That said, many pro/eSports leagues does require identical hardware to prevent input-lag differences from giving anybody an advantage.