USB 3.0 - interrupt protocol, lower latency for gamers?

[IanM]

Strictly hypothetical as I don't think there are any HID peripherals yet that use USB 3.0, and there may never be as the controllers are expensive, and bandwidth is not needed. However:

Increasing the USB polling rate for lower latency can lead to erratic performance, so would it be better if future gaming mice adopted USB 3.0 in order to use the interrupt protocol? Would it be genuinely lower latency than 250Hz-1000Hz USB, or at least more consistent? would CPU overhead be reduced or increased?

Also, I think some people prefer PS/2 keyboards, but the legacy socket is being phased out, so would keyboard users also prefer USB 3.0?

[daedalus]

Haters gonna hate when I can beam characters from my keyboard to my PC at 5Gbps. Fuck yeah.

[ripster]

USB 3.0 is definitely the future. But mainly it's the greater amperage capacity that would make it useful for keyboards. Like REAL USB hubs, not the wimpy 100ma HHKB/Das/gamerz versions.

PS/2 is the past. Just not room for it any more.

[NewGuy]

Sounds like a pretty interesting idea. With all the companies out there selling gaming peripherals, you have to imagine that at least one of them must be working on something using this.

That said, at 500Hz you're down to 2ms latency, 1ms at 1000Hz. With my mice I have felt a difference going from 250Hz to 500Hz but not 500Hz to 1000Hz. I do wonder if it's more than just the latency that I'm noticing though, the times are so small that you wouldn't expect it to be a perceptible difference.

[strike015]

i agree with what newguy saids, i dont think a latency <1ms would really be detectable, and i think that people prefer ps/2 because they need nkro for whatever reason or the fact that they want the keyboard to work should the usb port become unstable due to overclocking.

[ripster]

They use PS/2 because Steelseries and OCN have told them to for ages.

Stupid Steelseries and Overclock.net.

[BababooeyHTJ]

I assume that this would require a gold plated connector?

[sixty]

Don't get ripster started on goldplated connectors. I think he has rageful nightmares about those every night. That is probably why he is so grumpy all the time!

[webwit]

Rapster snapped when I told him about gold plated dip switches. Or was it when he got into the ALPS?

[Daniel Beaver]

I hear Monster is having a sale on USB cables. Only $49.95!

The braids in the cable make it electricity go faster.

[False_Dmitry_II]

Yeah, well, I build all my computers. So for as long as possible I'll keep buying mobos that have PS/2 on it, and continue to use it. There may be no real reason to use it, but there's no real reason not to.

Even if the interrupt advantage is a strictly academic one, it still is an advantage. On top of that I just generally run into problems trying to use USB keyboards. Sometimes they'll just sit and stare at me without working, even with me plugging them into different ports and rebooting, etc. PS/2 just works.

Granted, my only times trying/using USB are with rubber domes so that could be part of the reason I was having difficulty getting them to work. On top of that, all my mech boards are old so I'd have to buy a USB adapter to even use them on that. Why bother?

As far as mice go, interrupt based would sound like a cool idea. Except for the fact that they send much more data than 'this switch is on, now off' like keyboards do. Keep in mind PS/2 mice even poll, so I'd guess that it isn't really feasible.

Though keyboard interrupt on USB3 might be cool.

[ripster]

I build my own computers too. Never had a problem with USB connections.

And my DealExtreme HHKB/Poker Killer works fine with a PS/2 dongle.

[NewGuy]

Those keys look horrible, like a Mac keyboard clone. (I hate the new Mac keyboards)

PS2 generally is more stable than USB when doing extreme overclocking—that is, people going for world records and running systems on liquid nitrogen at 8GHz.

With the latest Sandy Bridge processors, overclocking is all done on the CPU multiplier though, so USB will be unaffected, and the new 1366 overclocker boards all have direct controls for clockspeed/multiplier on the board themselves now:

[ripster]

I haven't started to OC my Asus P8P67 Deluxe yet. It's already tearing through my AVCHD videos and the Newegg replacement one should arrive tomorrow. I'm tempted to fry the old one before sticking it in the box since Intel is just gonna collect it and put it under a bulldozer anyway.

[webwit]

None of that is gonna fit in your iPad.

[Rotlaus]

I build my own computers too. Never had a problem with USB connections.

And my DealExtreme HHKB/Poker Killer works fine with a PS/2 dongle.

Have you reviewed this yet? Is this comparable to my Thinkpad Edge Keyboard? Looks interesting.

[eclairz]

Thinkpad edge keyboards have a more curvier keycap, chiclet keys are usually flat, mechanical keyboards are often curved so the finger presses the key in the centre to reduce the resistance. So by combining the two keycaps you'll get the edge keyboard. Otherwise, the Thinkpad keyboard uses a special version of the scissor switch, as they often purchase more "clicky"? switches from companies like Chicony, so the travel and resistence maybe different than your standard chiclet keyboard.

[daedalus]

I'd say that there's various grades and tolerances of scissor switch and rubber dome, so that you can either very cheap mushy ones, or crisp ones that are a bit more expensive.

[Minskleip]

I can play twice as fast with goldplated connectors

[IanM]

this thread got a bit derailed and I forgot about it until now...

That said, at 500Hz you're down to 2ms latency, 1ms at 1000Hz.

i agree with what newguy saids, i dont think a latency <1ms would really be detectable

I agree with these sentiments. I always set my polling at 500Hz, never 1000Hz, and now there are a lot of mice that allow 333Hz which may be a better compromise than 500Hz.

I remember that when 1000Hz polling was a relatively new option, various technical reviews tested the performance. In practice 1000Hz was a worse option than 250Hz due to erratic latency - very big dips in performance and overall 250Hz and 500Hz were much more consistent. I don't know how true this would be today if the same type of testing were conducted with contemporary motherboards and chipsets, but it doesn't matter - 1ms v. 2ms is nothing to worry about, there is much more latency elsewhere in the PC system. That said, I firmly believe that 8ms-9ms is detrimental (half a frame at 60Hz) and so anyone concerned about latency while gaming should seek to avoid adding more than this (either because of the mouse, or the display) There were some old mouse reviews from Japan that tested latency. The links from this old forum thread appear to be dead now, but some mice were tested at over 25ms latency.

I'd like to see all future gaming mice have consistent, guaranteed performance. It doesn't matter if it's in the same amount of average latency we get now from 500hz, if using interrupts can ensure that there are no erratic dips and spikes in the latency, I think I would be willing to spend a little extra to upgrade the mouse.

[ripster]

I think all mice should be bulldozed and gamers get back to the Joystick.

And that idiotic OCN Mechanical Keyboard Guide STILL talks about PS/2 being the preferred choice for keyboards. They'd still be using PS/2 mice if Steelseries and Razer told them to.

[IanM]

I think all mice should be bulldozed and gamers get back to the Joystick.

with a single button? VCS style is awesome:

And that idiotic OCN Mechanical Keyboard Guide STILL talks about PS/2 being the preferred choice for keyboards.

I'm just amused that the site is supposed to be primarily about overclocking, they give multiple reasons to use PS/2 but not not the best one i.e. USB keyboards often fail to register at POST after a sudden power loss, reset or BSOD

[Crazy9000]

Slip someone at Razer a few bucks to release this and we can test that theory.

[IanM]

Slip someone at Razer a few bucks to release this and we can test that theory.

3Dconnexion might be a little peeved if they did!

However, while looking for the space explorer mouse I did find ripster's ideal mouse

[Findecanor]

3Dconnexion are not the first to release such a controller. There is the Space Ball, of which there has been many versions since the early '90s I think. Then there was the Logitech Magellan.

Btw. All leet Mac gamers connect their keyboards with Thunderbolt.

[ripster]

I'm still waiting for USB 3.0 card readers. Some Japanese company announced one and I haven't been able to find it.

Otherwise eSata on a ThermalTake BlacX is fine for me.

Virgin Ports above slutty well used dongles.

[IanM]

I'm still waiting for USB 3.0 card readers. Some Japanese company announced one and I haven't been able to find it.

PhotoFast CR-8600 or Pretec P240 ? the 2nd one looks like it's available

[Telstar]

Don't touch ps/2!
The only interface that always works since my first computer.
Long live ps/2.

[ripster]

My Filco Gen 2 Cherry Reds and Camo Blue runs at 1000hz (1 ms max delay for up to 7 character packet).

PS/2 is for old folks. Doesn't even work half the time with modern motherboards.

http://geekhack.org/showwiki.php?title= ... ility+List