Microsoft working on touchscreens with one millisecond lag

The responsiveness of touchscreen displays nowadays is pretty good but Microsoft's Applied Sciences Group has a different vision of where touch input delay needs to be. Microsoft wants touch responsiveness to take no more than one millisecond to respond to a tap or gesture.

Paul Dietz is part of the research group that's been looking into the technology which could potentially be used to bring this vision to existence. Dietz believes that 1ms is a great benchmark to aim for, as it reduces the lag between touch input and display response to a pretty much imperceptible level.

Current touchscreens have a lag of around 100 milliseconds, according to Microsoft's data, which results in a recognizable delay between your touch and the onscreen interaction you're trying to activate. Check out the video below and let us know your thoughts - will a reduction from 100ms to 1ms really make a big difference in practice?

Report a problem with article
Previous Story

Anonymous targets Vatican, says “a storm is coming”

Next Story

Internet Explorer 9 flaws found in Pwn2Own event

54 Comments

Commenting is disabled on this article.

I thought Microsoft was dependant on Samsung for their touchscreen panels. You know Surface, and I think the 82 inch screen they used at W8CP launch was by Samsung too?

gzAsher said,
I thought Microsoft was dependant on Samsung for their touchscreen panels. You know Surface, and I think the 82 inch screen they used at W8CP launch was by Samsung too?

MS most likely wouldn't manufacture this large scale, but rather license out the tech to companies like Samsung.

bleh, this is the problem with microsoft, they have all these great stuff that they work on but none of them ever seem to get released.

If they could put this out in the market right now, it will make the Retina Display on the new iPad not matter at all.

FalseAgent said,
bleh, this is the problem with microsoft, they have all these great stuff that they work on but none of them ever seem to get released.

If they could put this out in the market right now, it will make the Retina Display on the new iPad not matter at all.

I thought AMOLEDs were already better than the Retina Display?

THIS is why I HATE typing on touchscreens. By the time I press the second letter, the first letter presses down. It's very annoying. I imagined a future where touchscreen response didn't improve and my kids would be calling me old for not using touchscreens, but it appears this won't be a reality at all. I might come to love touchscreens just as much as (or more than) a physical keyboard.

Cool. This could definitely be nice if it were cost effective. It's really cool all of the stuff Microsoft is working on.

dotf said,
So once Apple become the first licensee of this technology, the world will forget this video and appreciate how much Apple innovates /S

Sadly, it would seem that way.

dotf said,
So once Apple become the first licensee of this technology, the world will forget this video and appreciate how much Apple innovates /S

sounds about right. And apple will start suing for responsive touchscreens.

Dead'Soul said,
100ms is too slow. my phone seems like 10 ms or better? microsoft: please update your software

how are you people testing this?

Dead'Soul said,
100ms is too slow. my phone seems like 10 ms or better? microsoft: please update your software

Microsoft Research aren't testing against their own software - they're using real devices liike Pads and such. They're a research division of computer scientists, not Microsoft's marketing department. It's in their interest to use real devices to get the best results. There's no way your phone is reacting appropriately within 10 ms, it's not realistically technologically possible with the current market tech

~Johnny said,

Microsoft Research aren't testing against their own software - they're using real devices liike Pads and such. They're a research division of computer scientists, not Microsoft's marketing department. It's in their interest to use real devices to get the best results. There's no way your phone is reacting appropriately within 10 ms, it's not realistically technologically possible with the current market tech

yeah, id love for people to show how they are getting these results in the above comments, and explain what tools they are using to test.

I am Reid said,

yeah, id love for people to show how they are getting these results in the above comments, and explain what tools they are using to test.

They are using the Eyeometer and STD (Subjective Time Differential) Scanner.

I am Reid said,

how are you people testing this?


i'm wondering the same thing, how do you guys know that it's 'more like 10ms' and not '100ms'?

A 1ms latency is going to be bounded by screen refresh rate. If your refresh is 120hz, the screen will only update every 8.3 ms. If the interrupt that drives the touch even happens instantaneously, you're still looking at an average 4ms lag (which still isn't too bad). Throw in software handling of the event, which can take a 3-4 ms, you're lucky if you can come in under 10ms no matter how good the hardware is. So, part of the equation is to push for 240hz refresh rates and to optimize the software stack to favor input events over other processing. With all that in place, you may be able to attain 3 ms response times.

Kevin Hsu said,
A 1ms latency is going to be bounded by screen refresh rate. If your refresh is 120hz, the screen will only update every 8.3 ms. If the interrupt that drives the touch even happens instantaneously, you're still looking at an average 4ms lag (which still isn't too bad). Throw in software handling of the event, which can take a 3-4 ms, you're lucky if you can come in under 10ms no matter how good the hardware is. So, part of the equation is to push for 240hz refresh rates and to optimize the software stack to favor input events over other processing. With all that in place, you may be able to attain 3 ms response times.

1KHz refresh rates anyone?

GS:mac

Charles Keledjian said,
Great. Now I'm going to be aware of the incredible lag all touchscreens have and will be bothered by it until I can get that 1 ms.

hahahahahaa

Testing with an HTC Flyer and an HP Touchpad, they both certainly seem to be inline with 100ms. It's easier to get a better idea if you use a thin stylus or something rather than your finger. They not mention how long they expect before this kind of research begins to influence real world products?

Same here , with touch screen test on android i'd say its more like between 10-25 ms delay.
No way near as bad as 100ms

Tested on Galaxy tab 7.7 and Samsung galaxy s2

cRuNcHiE said,
Same here , with touch screen test on android i'd say its more like between 10-25 ms delay.
No way near as bad as 100ms

Tested on Galaxy tab 7.7 and Samsung galaxy s2

Yeah my iPad 2, Galaxy Nexus, Motorola Xoom and Galaxy Player are not that bad at all.

UndergroundWire said,

Yeah my iPad 2, Galaxy Nexus, Motorola Xoom and Galaxy Player are not that bad at all.

Same with my Optimus Slider. Stock, it might be that bad, but after rooting, etc., there's hardly any lag.

cRuNcHiE said,
Same here , with touch screen test on android i'd say its more like between 10-25 ms delay.
No way near as bad as 100ms

Tested on Galaxy tab 7.7 and Samsung galaxy s2

Sir, what kind of experience do you have counting milliseconds?

satukoro said,
Same with my Optimus Slider. Stock, it might be that bad, but after rooting, etc., there's hardly any lag.

Rooting doesn't do anything but give you "admin" rights. But just for the record My Xoom and Samsung Galaxy Player are not rooted and perform very well. My Galaxy Nexus is rooted with no modifications. It also performs really well.

UndergroundWire said,

Rooting doesn't do anything but give you "admin" rights. But just for the record My Xoom and Samsung Galaxy Player are not rooted and perform very well. My Galaxy Nexus is rooted with no modifications. It also performs really well.

By "rooted, etc.," I meant I rooted my phone, installed a custom rom, and optimized performance.

satukoro said,
By "rooted, etc.," I meant I rooted my phone, installed a custom rom, and optimized performance.

Which is why I had to be clear. For any Android hater out there, they will say that you had to install the custom ROM to make it smooth. I root my phone to have Titanium Backup, AdFree and Autostarts installed and running. I choose not to install custom ROMs because something always doesn't work for me. Especially CyanogenMod.

UndergroundWire said,

Which is why I had to be clear. For any Android hater out there, they will say that you had to install the custom ROM to make it smooth. I root my phone to have Titanium Backup, AdFree and Autostarts installed and running. I choose not to install custom ROMs because something always doesn't work for me. Especially CyanogenMod.

I tend to have good luck with CM. However, since Gingerbread became more prominent than Froyo, the interface lag has been drastically reduced on stock consumer devices.

satukoro said,
I tend to have good luck with CM. However, since Gingerbread became more prominent than Froyo, the interface lag has been drastically reduced on stock consumer devices.

Nothing to do with the features of it. My company pays for Verizon Visual Voice Mail, however Visual Voice Mail does not work on ANY ASOP ROMs unfortunately.

UndergroundWire said,

Nothing to do with the features of it. My company pays for Verizon Visual Voice Mail, however Visual Voice Mail does not work on ANY ASOP ROMs unfortunately.

How come it doesn't work? Is it system-file dependent?

My tablet reacts how the '10ms' example reacts.

I don't think there's any touchscreen device that is as slow as the '100ms' example.

Sacha said,
My tablet reacts how the '10ms' example reacts.

I don't think there's any touchscreen device that is as slow as the '100ms' example.


Well, I saw some really bad example on 50" screen. Also on my HP Touchsmart while on Windows 7 there was pretty huge lag too (and again it's more easily noticable on larger than tablet 23" screen), but then I installed Windows 8 CP and latency improved a lot, seems like 20ms or so. which becomes pretty usable for MIDI touch interfaces.

I think they were using an iPad. Open up a painting program, and move your paintbrush across the canvas very fast. I think their test device was a 15" screen, based off of the guy's hand size. Now imagine that your paintbrush has to move even faster (to cover a 15" screen in the same time) and think about the lag.

Then *poof* it's gone.

It's much less noticeable on smaller screens since we move our fingers slower. The larger screens make it apparent.

For some reason I always assumed the latency was a hardware limitation, and as the hardware gets better the latency w/ touchscreens would go away.

EDIT: I just tried it out on the iPhone 4S and the latency seems to be around 10-25ms. Is the 100ms latency for Windows Mobile devices?

Are they not using an iPad in the video? >.< Wasnt entirely sure what the other "test" device was, but it looks like a white iPad 2.

KeR said,
For some reason I always assumed the latency was a hardware limitation, and as the hardware gets better the latency w/ touchscreens would go away.

EDIT: I just tried it out on the iPhone 4S and the latency seems to be around 10-25ms. Is the 100ms latency for Windows Mobile devices?

how exactly did you test this?

Wow, we can send data to a server thousands of KMs away with less of a delay than it takes a touchscreen to respond.

I'm surprised the lag is that high though, most of my touch devices seem pretty much instant.

giantpotato said,
Wow, we can send data to a server thousands of KMs away with less of a delay than it takes a touchscreen to respond.

I'm surprised the lag is that high though, most of my touch devices seem pretty much instant.

im thinking its mostly because the touchscreen needs to go through a software driver that is shown afterwards.. also to probably to keep power consumption down.

giantpotato said,
Wow, we can send data to a server thousands of KMs away with less of a delay than it takes a touchscreen to respond.

I'm surprised the lag is that high though, most of my touch devices seem pretty much instant.

Most devices are smaller so you don't notice much but at 1m/s it's very noticeable. As touch monitors and displays become more common, this will be more important.

laserfloyd said,

Most devices are smaller so you don't notice much but at 1m/s it's very noticeable. As touch monitors and displays become more common, this will be more important.

More common and MUCH larger. With cheap OLED wall-sized screens on the horizon, we're going to need better LARGE touch and wave-gesture device sensitivities.

still1 said,
is that really necessary?? you cant possibly do anything on that 99 millisecond.

I would guess it would depend on the size of the display and the speed of the gesture being acted on the display. For example, you'll be able to notice the lag of the display when moving an object quickly across a touch enabled table. On smaller devices the difference would be noticeably when you move stuff quickly across a touch screen.

As for being necessary. Most technology isn't, but it's nice to have.

still1 said,
is that really necessary?? you cant possibly do anything on that 99 millisecond.

The video shows you, current ones don't exactly feel like your moving something. It's more like a gesture, with the 1ms touch it's instant so it would feel a lot better, especially on big displays. Also it would make things like writing with touch and drawing with touch a lot easier.

I hate when I see those ipad adverts and the guy draws a brilliant picture. It's simply miles away from real drawing, until the lag is fixed.

Xerax said,

Did you even watch the video? Go be an idiot somewhere else.

That wasn't really necessary. It was a valid question. But hey I guess you proved who the real idiot is.

still1 said,
is that really necessary?? you cant possibly do anything on that 99 millisecond.

Even in the case where the user wouldn't notice the difference, that's 99 additional milliseconds of processing for the app, which can be that much more reactive.

still1 said,
is that really necessary?? you cant possibly do anything on that 99 millisecond.

You probably don't realize how "fast" a real button is, this is aiming for that speed.

still1 said,
is that really necessary?? you cant possibly do anything on that 99 millisecond.

It really adds up. I hate delay on touch devices. You can do more precision type stuff. It's very cool tech.