Understandable. But putting it in the same application, ie, pressing a button with your toe or finger (the same button, mind you), won't be a huge difference in time (though significant enough to fail at many games). If you were pushing a pedal down with your hand, then your hand would be slightly faster (unless it takes more force than your used to exerting from your hand/muscles, then it would be slower). Comparing pushing a nes button to a gas pedal (or any other application with a different part of your body with different requirements) is silly. Compare it to a drummer who has to be quicker with his feet than his hand, depending on the style of music, where your feet have to have a significantly higher response time than your arms (also a major displacement difference from your foot to a kick pedal and your arm+stick from a high hat )

You wouldn't notice the difference between light being emitted from 5 feet to 100 feet as light travels much quicker than anyone can think.

As for games, any rockband and guitar hero game will be next to impossible to play if there's a 10 ms lag in either audio or video.

I'm not familiar enough with the entirety of the library to know how much cue-ing time you get in most games.


But cropping the cueing time on Tyson, for instance definitely matters to a lot of people.

From my own experience with L19+ on Tetris, I'd say takes 2 frames, or more to be truly detrimental, but you would have to ask Jonas or Harry to characterize how much individual frames of latency impact their play.




You would need to know the actual cue-length of specific events though to say what percentage of people are impacted by each additional frame of latency based on average response times.




But your light vs distance argument is many orders of magnitude different from the per-frame timescales we are discussing it shouldn't even bear mentioning.

He is just trying to disqualify my statement of the times not being relevant to my stated formula, by picking the most extreme case possible.

damnit I have too many posts to quote here and quoting all in one post is a pain.



That was 90% sarcasam.  I mean sure, adding an exentsion cable would add a bit of lag, but I can't see it even being a MS. 



I'll back down on the 400 MS for feet, me saying that is based mostly on anecdotal evidence from drag racing.  You have two main types of trees, a normal one that lights up at 500 ms intravels and a pro that lights at 400 ms.  Everyone's always told me the pro tree was designed because that's how long it takes to react and press the foot down.




I'd say you should know to give me more credit than that, but then again I'm quoting anecdotal evidence from drag racers with 3-4 beers in them for human reaction times.

And yes, I had to look up how to spell anecdotal.

Ozzy, it probably does take 400ms for drag racers and they're focusing on more than one thing with more than one application. You need to compare reaction times that are actually comparable. Catching a dollar bill between your fingers and press a button on a controller would be similar applications, same with kicking a football vs kicking a board during a Kung fu demonstration.

The light thing would have no effect unless you 1.) physically can't see the screen clearly anymore, or 2.) using highly sensitive photoelectric applications.

Hey um, guys? We've been wondering around for a while now I'm lost. Anyone remember where the topic is?

Someday, Nintendo Age will have a thread that stays on topic for more than 3 pages.

I think it is much more likely that 100 ms "relative" intervals (i.e. 500 vs 400) may have just been the convenient tolerance of inexpensive light-control equipment when the sport was formalized.

And from the days with full-throw pedals it was probably the case that 300ms was too short for most people.

Thanks to both Ozzy and arch for commenting on this. There seems to be wide disagreement among experts when it comes to input lag. Hopefully between you two we will be able to reach a conclusion.

Played with an emulator called übernes, and it's got some really cool features - online scoreboard, nes database with game descriptions, etc. and it seems to not have any lag in full screen mode. Unfortunately you have to use a dumb NTSC filter, I couldn't get decent full screen otherwise. So I'll be sticking with Retroarch.

I believe at this point that Retroarch with the posted settings will result in the least possible input lag of any nes emulator running on Windows. I'm using the nestopia core, but "hard gpu sync" at 0 frames seems to do some kind of magic that results in less input lag than nestopia running by itself.

To get a truly definitive answer, we would need to look at all of the ultra-tight-timing situations we can think of and bench them in an emulator that was somehow running frame-by-frame (so latency doesn't factor in, at all).

Then we could look at the actual exact frame-duration between an on-screen cue and the absolute latest that the player can successfully respond.

Adding a few feet of length to a controller cable would increase the controller latency by only a few nanoseconds. Sitting a few feet farther back from the TV would also create a few nanoseconds of lag because the signals travel at the speed of light, which is 186,000 miles/sec, or 297,000 km/sec. So if you want a few precious extra nanoseconds of display lag, keep all cable leads as short as possible and scoot up really close to the TV. In practice, make the leads as long as you want because those all extra nanoseconds are statisically insignificant.

Display lags are measured in the tens of milliseconds. For the record, 10ms is ten million ns. So yes, maybe ten million controller extension cables would provide a percievable difference in input lag. But your controller would have long ceased to function due to the resistance and impedance of the copper conductors causing massive signal degradation.

I think I'm going to go ahead and get this:

http://www.leobodnar.com/shop/ind...

It's not as expensive as I thought it would be and should give a definitive answer to the lag question. At least the question of how much lag I'm actually experiencing using an emulator on an LED HDTV.

^^ A bit late to the party, don't you think?

That is only going to show the processing and pixel response time of the TV, so it will miss huge parts of the chain from controller/emulator/os/graphics card. Definitely well designed for what it does but you won't be able to compare anything besides brand/model of TVs.

I think you missed a couple layers of sarcasm and facetiousness.


5 ns/m is the propagation delay for CAT5E cable.  That is 5 billionths of a second per meter of cable, roughtly 3 million times smaller than a single frame.

Agreed.


Didn't your testing involve a special ROM and a high speed camera?

That setup would let you test emulators.  Did you ever get to try it on a Retron5?

So is there any way to test the total latency that doesn't involve spending hundreds of dollars?

High speed camera + controller with LED, tested on Balloon Fight: http://www.nintendoage.com/forum/... Has it really been that long since I did more tests?

He also missed the post where I pointed out once before this is sarcastic.  We now must point and laugh at him.

Just stop it guys. I was illustrating a point that adding lenth to a controller cable is moot, unless those lengths are astronomical. As usual, I took way to many words to explain myself.

Ignoring signal loss and degradation, the controller would fail once the propagation delay reaches a couple microseconds (I forget the exact period of the clock signal), mainly due to the fact the controller I/O signals need to be synced with the console and the signals are sent at a very rapid rate. I forget off the top of my head but I did an experiment with NES and SNES Retrobit extention cables daidychained together. Both the NES and SNES controllers stopped working after about seven extensions. 6 feet x 7 extensions + 8 foot controller cord = 50 feet. Go much beyond this length and the NES/SNES controllers quit working on original hardware.

Retrobit knows this as they advertise you can join up to six cables together. Add a seventh extension and the controller mysteriously quits reading.

The first gsync enabled monitors are now available, although they are really expensive - $1,200 at amazon.

http://www.amazon.com/PG278Q-27-I...

These displays have NO INPUT LAG.

We are on the verge of having an hdtv technology that can replace CRTs.

Display Port only? NO DVI/HDMI? So this will not work with the HDMI NES?

Can't you get an adapter dongle?

Better to wait a few years anyway. I just think it's exciting that they're now putting gsync in monitors. Input lag will soon be a thing of the past. One less barrier to near-perfect console experience via emulation.

But CRTs always had zero latency with analog signal, and HD over VGA. Maybe the digital revolution had to take a huge step backwards before it could catch up to analog standards?

Pipe dream I know, but:

IF they release a G-sync monitor that works with legacy analog inputs like Composite, S-Video, RBG, Component, and VGA, then it could translate the signal using an analog-to-digital converter for true realtime lagless display. Better yet, since the G-sync monitor could theoretically operate at any clock rate, consoles with wierd or non-standard NTSC timing would be no problem. I doubt it but it may even be possible for them to work with a modified zapper. We just need to rethink the way the display drivers operate.

Or a person could spend 10x less and play the real hardware on a CRT....