Microsoft says today's touch screens have latency of 100 milliseconds between a finger hitting the glass and the image on the screen changing. At that speed, says Paul Dietz, Assistant Director of Microsoft Applied Sciences, fingers have moved ten centimetres and the “analogy of moving a real physical object breaks down.” At …
This depends entirely on graphical complexity. Try running Google's GDC11 test app. It's a little globe that feels natural enough. I mean you can debug it yourself - touch events are sampled at 60 fps and frames are drawn at 30 fps. 100ms is a glaring, painful delay - you'd notice it. It's not there.
Now, many apps run that slowly, especially on single-core processors, but that's not the touchscreen's fault imo.
On the other hand, getting the lag time down to 1 ms would be pretty impressive, but when was the last time you saw a 1000hz display, even without a touchscreen?
RE: Whuh? or Latency != graphics or frame rate
To the chap(ess) saying it's all about graphical complexity:
Touch latency (which is what this article is about) has nothing to do with graphical complexity or the fps a device acheives.
Consider applications such as rhythm devices (drum machines / audio applications). These applications and others like games require accurate temporal activation. Delays > 10ms are easily detectable when using such a device.
To see the problems of latency with regard to audio see http://www.youtube.com/watch?v=G5XhzwZ78Sc
Re: RE: Whuh? or Latency != graphics or frame rate
Yeah, agreed - hadn't even considered audio.
My point was mostly that 100ms is an absurd latency time... things aren't THAT bad right now, at least not on my 18-month-old Android phone. Maybe it was a typo in the original article? ¯\(°_o)/¯
30-10ms is more realistic. I get the distinction you're making between input latency and output latency and you're definitely right, but the article was talking about a hyper-responsive graphical interface, and I just don't see how you do that without a much higher refresh rate. I suppose the trick is to draw that very first frame as soon as possible after the touch event, but now you're talking about synchronizing the asynchronous...
"goal to shoot for over the next decade"...
I'd be happier if MS would address the problems they are going to have with the technology of the next couple of years rather than dreaming what they would like to do in 10 years. While they seem to have some capability to predict trends (eg: tablet computing) their actual response didn't seem to have much imagination. They did recognise the existence of the Internet fairly early, but again the response was pretty poor.
Perhaps this is part of their corporate culture. Wait until a technology trend becomes unavoidable, then buy or copy the best they can at that time.
Re: "goal to shoot for over the next decade"...
> They did recognise the existence of the Internet fairly early
The first edition of 'The Road Ahead' made no mention of the internet at all. Later printings did add this.
With Windows 95, MS had the original MSN which was a closed environment for Win95 users only and was isolated from the internet. Their response was not just 'poor' it was trying to shut it off.
> some capability to predict trends (eg: tablet computing)
MS is very poor at predicting trends. What it tries to do is to _control_ trends. The tablet concept was around before Microsoft existed (see Dynabook). While MS may have been putting Windows on Pen and Tablet devices for quite a while, this obviously was _not_ the trend. It was only when something completely different came out, ie _not_ Windows, that it became successful.
Microsoft corporate culture is to control with an iron fist. By using contracts, 'partnerships' and discounts. When cheap netbooks with Linux started arriving MS killed these off by wielding discounts and reviving the old XP. With ARM tablets (such as WebOS and Android) MS has brought out WOA in order to be able to wield discounts to have these forced into using Windows/Metro, or killed off. It has already worked for WebOS, HP could not afford to lose the discount on all its Windows PCs.
I weep for MS Research
I work on exactly that problem and all MS have done is paired away any processing, created a small mono display and said in 10 years this is were we'll be at.
Anybody could produce that, the hard part is adding colours, converting the path into a vector, grouping, handwriting recognition and other actual useful things in that 100ms that is the big problem.
In 10 years time processors and buses will have improved enough to make existing software do all that processing in 10ms.
I tested a drawing package on an iPhone and iPad to test/observe the 'latency problem' first-hand. I can't say I've actually noticed a problem in the past, so I was intrigued. While I'm certain there was some kind of delay, the response was so fast I couldn't be certain I wasn't just imagining it. Even trying to see a problem, it felt instantaneous, fluid.
I could see the delay in the example video, but I couldn't relate it to real-world experience on a touch device. It seemed as if someone was intent on creating a problem in order to have something to solve. It was telling that they had to use a high-speed camera (and slow down the playback) in order for you to actually see the latency in the sketching example. If you have to slow it down to clearly see the problem, is it a genuine problem? And the demo system seemed sluggish compared to contemporary Apple kit. I can't speak for Android - I don't have a current system to hand. Anyone able to provide first-hand experience on that front?
Re: Meaningful problem?
100ms sounds short, but its a meaningful time to a fast typist, say 120wpm. Based on a standardised 5 character word thats 600 characters per minute, or 10 characters per second, i.e. 100 ms per character. That'd mean they weren't getting any feedback on whether a keypress registered until they'd normally already be typing the next one.
Now whether anyone can actually type that fast without a proper keyboard with moving parts is another matter. However part of the problem might be that what little feedback mechanism there is can't keep up.
Isn't this Microsofts standard model? Bloated code and wait for the hardware to catch up?
Wintel was a great little civil partnership, each drove the others business. However now Microsoft are also seeing other people, they (Microsoft) can't drive the show so much anymore. Their new Beaus have other things to consider like power draw etc. So Microsoft has to flirt again, not be the demanding Alpha partner.
I just don't see Ballmer or any of his team as the chocolates and flowers gang. I think they are more of the "Right, bend over..." mindset.
-- "However now Microsoft are also seeing other people, they (Microsoft) can't drive the show so much anymore. Their new Beaus have other things to consider like power draw etc."
There's a large irony here with the whole Nokia business. Microsoft thought - or assumed - that Nokia understood about power consumption issues on smartphones. They've now discovered that Nokia don't understand. It was *Symbian* who understood. And Nokia and Symbian were not actually the same company.
The last time I tried typing while somewhat drunk, it felt as though the keys weren't getting hit when I wanted them to be. It came out all right, though. (What I was writing may have been less than ideal, however.)
I'm now using a desktop touchscreen, and if I stroke a stylus across the screen, the on-screen pointer follows about 2 cm behind. If I stop, it catches up. There is a lag, and it makes using the screen a less natural act - like being drunk - but I can live with it.
They've got it all wrong
Forget latency on touch screens - I want a tablet that does what I want it to before I've fully comprehended the thought that came into my head.
"Ah that's just what I was about to want it to do."
MS drops Metro and goes with the interface in this clip. You can still call up the Desktop by drawing it all with your finger.
looking forward to this happening
And even with the super fast response time when your windows bloated tablet needs a fan and is still hot to the touch in your lap it will fail ala the Kin, Zune, WP7, etc.
I havent read theRegister much over the last couple of years, mainly i've been to busy good to see the anti microsoft troll are still trolling around.. no doubt if this was research by apple or google you'd all be going ooooh wow this is so clever OMG and other girlie things like that. you sound like a bunch of bitter old women!
"Microsoft simulates one millisecond latency in the following video, but doesn't have the technology to make it happen".
So, nothing changes at Microsoft.
Too much caffeine this morning?
Tech company research department does tech research, identifies problem that might need solving, but will take years. Seems nothing wrong with that, so why all the negative comments?
As happens, I doubted this 100ms figure (as did some above), so dug out my iPad 1. Admittedly not the fastest now, but good enough to check with.
What I found is that I don't move my finger that fast. If I do, then I get that kind of lag. So do I not move it that fast because I don't want to, or because lag has trained me not to?
Anyway faster would obviously be better, but screen refresh rates are going to have to improve an awful lot. Plus pushing all that extra data from GPU to screen isn't trivial.