Razer Says Its New Mechanical Keyboards Have 'Near-Zero' Input Latency (theverge.com) 172
Razer has announced an update to its popular Huntsman lineup of mechanical keyboards that reduces input latency to "near-zero," the company claims. The Verge reports: [T]he newly announced Huntsman V2 and Huntsman V2 Tenkeyless (which omits the numpad, volume wheel, and media controls for a more compact board) both have a polling rate of 8,000Hz, meaning they can theoretically detect key presses eight times faster than the original Huntsman keyboards. Combined with the keyboards' optical switches, which use an infrared beam of light to sense when they've been pressed rather than metal contact points, Razer reckons the two new Huntsman keyboards will feel more responsive for gaming, especially when combined with a high-refresh rate monitor. In contrast, standard mechanical switches can suffer from what's known as a "debounce delay," when the keyboard has to take a moment to work out if a key has actually been pressed or not.
Other improvements introduced with the V2 keyboards include new doubleshot PBT keycaps, which have a more durable design with legends that shouldn't wear away over time. The doubleshot design also allows the keyboard's programmable RGB backlighting to shine through the caps. There are seven preset lighting effects built into the keyboard, and you can customize them via Razer's software and save them to the board's firmware. Both keyboards are available with either Razer's clicky or linear optical switches. The linear switches have also seen improvements since the keyboard's first iteration, with the addition of a silicon sound dampener inside, and more lubricant to make them feel smoother to press. Razer also says it's improved the acoustics of the keyboards, with the addition of a new layer of sound dampening foam, and there's now a wrist rest included in the box with both keyboards. The full-size Huntsman V2 features a volume wheel and media controls on its top right, but only the smaller tenkeyless model has a detachable USB-C cable.
Other improvements introduced with the V2 keyboards include new doubleshot PBT keycaps, which have a more durable design with legends that shouldn't wear away over time. The doubleshot design also allows the keyboard's programmable RGB backlighting to shine through the caps. There are seven preset lighting effects built into the keyboard, and you can customize them via Razer's software and save them to the board's firmware. Both keyboards are available with either Razer's clicky or linear optical switches. The linear switches have also seen improvements since the keyboard's first iteration, with the addition of a silicon sound dampener inside, and more lubricant to make them feel smoother to press. Razer also says it's improved the acoustics of the keyboards, with the addition of a new layer of sound dampening foam, and there's now a wrist rest included in the box with both keyboards. The full-size Huntsman V2 features a volume wheel and media controls on its top right, but only the smaller tenkeyless model has a detachable USB-C cable.
now this is news for nerds (Score:4, Informative)
Not so sure if it's stuff that matters or just an ad, though.
Re: (Score:2)
Indeed. And a Razer ad to boot. 8000Hz is all fine while it works, but it's kind of pointless when half the keys inevitably stop working. The bigger news would be Razer making any hardware that lasted more than a couple months.
Re: (Score:2)
Re: (Score:2)
A pretty blatant ad for snake oil.
Could not have said it better.
Re: (Score:2)
Not so sure if it's stuff that matters or just an ad, though.
Does not matter (except the sheer stupidity of people buying such a thing because they do not get this gives zero actual advantage), it is a blatant ad. The whole "argumentation" about the speed being better is so close to a series of lies that the difference hardly matters.
Re: (Score:2)
Normally USB has a limit for keyboard (and other HID (Human Interface Device) gear) of 1000Hz, because the lowest polling rate you can specify is 1ms.
It's possible to go higher with a special driver, which is what Razor are doing.
If it makes any difference is debatable. The fastest monitors are about 250Hz so the standard 1000Hz rate is already 4x that.
Some people like short throw keyboards because they are faster for gaming. Some keyboards that use exotic key detection mechanisms let you program the trigge
Re: (Score:2)
Fastest monitors for fast response have a nanosecond-grade delay. To them, 8kHz input is actually meaningful. I.e. CS pros playing on CRT monitors.
But even LCDs have been clocking to 360Hz for a few years. But there, input latency is pretty fucked up due to all the processing that is needed. It's still better to get you input into the previous frame rather than wait for the next in a twitch game for young professional gamers with appropriate reflexes however.
Re: (Score:2)
These metrics are irrelevant. Atomic clocks are precise, too. So what. Instantaneous response time doesn't alter the fact refresh rate is what it is.
What matters is the speed at which events can be received and processed in the inner loop. A keyboard that substantially outruns that does not matter. You would have to argue that games process events at least an order of magnitude faster than the highest monitor refresh rates for there even to be an opportunity. That is nonsense.
"But even LCDs have been c
Re: (Score:2)
Do you know what a nanosecond is? One billionth of a second, or one tick at 1GHz. No monitor comes close to that level of delay, either for refreshing the LCD panel with an image, or for the pixels transitioning state, or even for transferring the image from the computer to the monitor over a cable.
The fastest monitors aren't even a million times slower than that.
Re: now this is news for nerds (Score:2)
Re: (Score:2)
The ones that have to sweep an electron gun all the way across for each scanline? What about them?
Re: (Score:2)
So young'un, have you ever heard of CRT monitors?
Are high-end gamers now using vector CRT displays again, so they don't have to wait the full frame refresh time for the magnetically-swept beam of electrons to plod its way across the screen 768 times (or 1080 or 1440 if you get a real nice one) and then zip back to the top to refresh the pixel that just changed there? An LCD that can *deliver* true 300Hz pixel update rate from GPU to screen is going to beat any reasonable CRT in performance, though perhaps not in aesthetics.
Processing time does suck on mo
Re: (Score:2)
A human takes 250 ms to react to anything if their brain has to be involved typically, at the edge of the best of the best, you will never do better than 150ms. Increasing the polling from 1khz to 8khz is on average making input reaction about 440 ns faster. This is 0.3% of the most insane possible human reaction time for gaming. It's snake oil.
Re: (Score:2)
It's just so funny how so many people are desperately grasping onto their version of "you can't see more than 24 frames per second anyway" year after year, after year.
It's usually the same people who scream "Cheater!!!!oneoneone, no way you had time to see me" at people like Shroud.
Re: (Score:2)
Anyone who's ever said 24 frames is insane. Even TV sets the bar at 60 fields per second for better motion smoothness. Film is a certain look that lets your brain fill in the gaps pretty well.
I'd put a reasonable estimate of the eye at closer to 150Hz, but since you can't sync your eye with a screen, you might have to hit 300Hz to hit the Nyquist-equivalent limit.
Re: (Score:2)
I didn't say you can't see more than 24 frames per second, I said from the time that input hits the retina, then get processed by the brain to formulate a response, and getting the hand to start to react is generally 250ms, no better than 150ms.
Higher frame rates are more pleasant and can skip motion blur to look nice and crisp and be convincing. Particularly relevant to VR which can't use motion blur since the head is moving in ways that would change the blur, so 72hz is a needed, and 90hz is noticably cr
Just FYI (Score:2)
Studies have shown that air force pilots could identify aircraft after seeing it for only 1/220th of a second.
I, personally, can attest that it is possible to see an LED flash of only 10 microseconds in duration (i.e. 100kHz).
While the "persistence" of vision for most people is 20 Hz, there's more to it than that. The 20 Hz value is true under what you might call "typical" indoor lighting scenarios involving typical lighting levels. The actual phenomena is not driven by the time exposed to a given li
Re: (Score:2)
Re: (Score:2)
I never spoke to refresh (though I would be shocked if any real gaming performance difference was detected beyond 72hz, though maybe more comfortable to play up to 120 hz).
The 'bad keyboard polling rate' means a superhuman that reaction takes 151 ms instead of 150.6ms to register. This doesn't matter.
Re: now this is news for nerds (Score:2)
Re: (Score:2)
For those who want the short of it, the article at *best* shows 12.7ms latency from the computer. Meaning that the ~400us average latency improvement would represent about 4 percent chance of being 'the culprit' for hitting a different tick in what is presumably a highly synthetic utility to try to maximize nVidia marketing collateral, even if you ignored the human reflex component. Add in human component and the best of the best would see it being the culprit about 0.3% of the time. Add in internet lag an
Re: (Score:2)
I mean sure, you'll need a high refresh rate monitor to externally measure things, but from the software side, a game is able to sample and process and input for each frame that it can process.
Hence if your game runs at 500 FPS, while your monitor is only 60Hz, your input latency is technically determined by those 500 FPS (as well as all the tech between the
Re: (Score:2)
Unless you lock the frame rate of the game to the refresh rate of the monitor, what the monitor can display isn't that important to the input latency.
Games literally do this. The whole logic loop of the game has to complete before the next frame is drawn. Frames don't matter for trajectories and inputs. You don't need a picture to determine collisions. You do that with math before drawing the next frame.
Re: (Score:2)
Within that frame you can have a pretty much arbitrary number of 'sub frames' if you use loops or timers where you can process stuff to get a more precise time stepping for your physics for example. But you won't get any other value for the input until the next frame happens.
If that frame can be displayed on the monitor is irrelevant. All that matters is what is getting polled and processed by
Re: (Score:2)
Wait...people are enabling wildly inconsistent frame rates just so they can get around an engine bug? Polling inputs can happen arbitrarily often and just hold the timestamp of the input until the engine needs it.
Re: (Score:2)
Real time processing with dynamic inputs like in games and especially multiplayer is a bitch. I suppose the main problem is that all this happens on computers, with tons of other stuff going on in the background. When I work with micro controllers, where I can count the cycles and use interrupts, things is a lot more straight forward and less uncertain.
And yes, that is basically why a lot of players un
Re: (Score:2)
Still completely meaningless. The switches alone add a few whole ms of uncertainty per key-press. So does the biological apparatus pressing them. Sure, the usual "tech-affine" nil-whits will think this is somehow an improvement and waste their money. It is not.
The fact of the matter is that around 100Hz polling rate, the whole human-switch-electronics system peaks (or rather has peaked a while before already) and any increase is no improvement anymore. Except it does give marketing higher numbers. And that
Re: (Score:3)
Looks like would-be audiophiles have picked gaming as their money sink. They'll probably get oxygen-free keyboard cables too.
Re: (Score:3)
Oxygen-free cables? Amateurs! Only electron-tube driven keyboard controllers will do! Sure, may be as large as a living room, but the clarity of the keystrokes is unsurpassed!
Re: (Score:2)
With audio it's just for your own enjoyment. With gaming it potentially gives you an edge against other people.
A lot of the top players use what would seem to be the least suitable choice, e.g. some prefer the analogue stick even for 2D games designed for a d-pad. That suggests skill is far more important than the controller.
On the other hand things like auto-fire have an obvious advantage. The most interesting one I've seen is replacing the joystick with buttons. You have an up, right, left and down button
Re: (Score:2)
Normally USB has a limit for keyboard (and other HID (Human Interface Device) gear) of 1000Hz, because the lowest polling rate you can specify is 1ms.
It's possible to go higher with a special driver, which is what Razor are doing.
And that's really the root of the problem, having to poll the device because, last I checked, USB doesn't do interrupts (maybe a new standard does, IDK). If gaming freaks want immediate response to their key and mouse button presses, they should use PS/2 ports.
Re: (Score:2)
It could be argued that the keyboard polling rate does not match the screen, so with a higher polling rate, you can "brute force" having the most current key state matching whatever the game inner loop is doing.
But you could also just syncronize the keyboard with the game code and poll the keyboard when the games ask it to
Too bad your hands still do. (Score:2)
What was it? 20ms or 200ms? :)
Definitely not zero.
And your eyes too.
Really, we're permanently living in the past.
But acting in the future.
Re: (Score:2)
The point is to get your input into a previous frame rather than have it wait for the next one. In this one, it matters what kind of rendering pipeline you have, and what it's outputting on.
This will actually matter for CS pros playing on CRTs for minimal latency. It might still have some relevant to same pros playing on 360Hz monitors. It probably won't matter to a rando playing on their office grade 60Hz monitor.
Re: (Score:2)
"The point is to get your input into a previous frame rather than have it wait for the next one. In this one, it matters what kind of rendering pipeline you have, and what it's outputting on."
Why not get the input into the frame before that? Or 10 frames before that?
What does a keyboard have to do with a "rendering pipeline"? Nothing. Absolutely nothing. But at least you can sound knowledgable.
"This will actually matter for CS pros playing on CRTs for minimal latency."
Prove it. All you can offer is bla
Re: (Score:2)
>Why not get the input into the frame before that? Or 10 frames before that?
That's what we did when we went from 125Hz polling to current standard of 1000Hz, and from 30FPS over 60Hz to 360FPS over 360Hz.
>Prove it. All you can offer is blah blah having NOTHING to do with input to the game.
You don't know how inputs into the game work, and want a free lecture. Ok.
Good luck with that.
>An LCD panel with a native 360 Hz update rate is not a "360Hz monitor". You understand far less than you think.
As far
Re: (Score:2)
The small things add up though:
- debounce delay (typically 10 ms* on only rising edge, only falling edge or both -- or avoided on both with optical, hall-effect or capacitive switches)
- max polling rate the keyboard supports (peripheral MCU/interface bound)
- how often the host actually polls (bus contention)
- processing time in the game (CPU bound)
- how the game's logic delays player movements (Software)
- rendering time (GPU bound)
- display refresh rate (Display, affected by display protocol)
*: Mechanical s
Re: (Score:2)
The human operator is imposing a 150ms lag on response in the most insanely responsive hypothetical. With that general timescale, an improvement of about 440ns of average delay due to usb polling is nothing.
Re: (Score:2)
That's how I write debounce code these days, but it seems that some people do debouncing differently, and wait for the noise to settle down before accepting the change (e.g., here: https://dygma.com/blogs/storie... [dygma.com]). I have no idea why that's done, but maybe there is some good engineering reason?
Re: (Score:2)
useless (Score:2)
Re: (Score:2)
Re: (Score:2)
And the maximum speed that the game can receive this information? "Think it through"
If a game processes keyboard events at 200 Hz and every keyboard can do 1000 Hz, how much improvement is offered by increasing that to 8000 Hz. None.
Re: (Score:2)
Not a big issue to me, but it may have a mesurable impact.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
The chance that the 500ns window falls across the tick of a game engine is negligible, and is in the statistical noise of any attempt to measure that whole scenario.
If you wanted to compare people by income, the fact that one person bought two lottery tickets and another bought one doesn't matter, even though *technically* the purchase of two lottery tickets represents an increased chance.
Re: Think it through (Score:2)
To run at 100 FPS, the computer has 100ms to turn your input and the game logic into new pixels.
No, itâ(TM)s not going to mean a huge difference in FPS, but even a couple can matter, especially at the low end.
Re: (Score:2)
USB supports device-initiated transfers, just like CPUs have interrupt lines to tell them that something external has happened without requiring the software to poll for events.
Re: (Score:2)
Yes, and that is putting it kindly.
USB was originally polling only and still can be operated that way. This mode was intentionally simple specifically to make keyboard and mouse support easy. That's not to say that keyboards and mice are hobbled by USB 1.0, it's to say that the assumptions made then are still valid and those operating modes still exist. You can't assume that a device supports advanced features just because it is possible that they do.
What happens in an individual USB packet is irrelevant
Re: (Score:2)
Scanning a keyboard matrix at 8KHz isn't difficult. You get 125us per scan which is plenty of time.
Re: (Score:2)
It wasn't even challenging to do this 40+ years ago.
PS/2 keyboard? (Score:3)
Why are gamers not using PS/2 keyboards? They work by interrupt? No polling then!
Re: (Score:2)
Re: (Score:2)
more the opposite. cheap keyboards all have issues with combo presses because of shortcuts, but it took a while to work around ghosting on USB while ps/2 doesn't have a problem with it.
at this point, though, USB keyboards with n-key rollover are pretty easy to find.
Re:PS/2 keyboard? (Score:5, Informative)
Some gamers actually do, because they have been convinced that it would have less latency.
PS/2 is not faster than "Full Speed" USB that most gaming keyboards use. In theory PS/2 could be just as fast as USB in most circumstances for single key presses, but in reality there are long pauses between bytes over PS/2 that the proponents of PS/2 for gaming don't take into account.
To learn more, I'd recommend: Ben Eater [youtube.com]: he goes into depth on both protocols, studying them in oscilloscopes.
That Razer keyboard (and others offering >1kHz polling rate) use High Speed USB, BTW. As to how fast it really is is anyone's guess until an independent party actually measures it. Keyboard latency has many factors, and Razer is infamous for having made outrageous claims right on the edge between truth and lie.
Re: (Score:2)
Re: (Score:2)
Re:PS/2 keyboard? (Score:5, Funny)
they don't have rgb led lighting. everyone knows rgb lights make things faster.
Re: (Score:2)
Re: (Score:2)
This was a thing, and then it wasn't as of high-speed usb.
True, the keyboard is *technically* polling instead of interrupt driven, but the clock to transfer the data is so slow in PS/2 that it takes longer to transfer the keypress than high-speed usb would poll.
they have to sell stuff (Score:2)
I have been playing computer games since 1976.
Setting aside actual system hitches where everything locks up for a moment or more (Windows updates, I'm looking at you) I have never in four decades thought "if /only/ my input device updated at 8000Hz instead of this paltry 300!"
And believe me, if I could possibly have used such an excuse I would have.
Note that afaik to even register anything other 1000hz (the USB polling rate spec) I guess you'd need to overclock your USB controller? Want more heat in unanti
Re: (Score:2)
You do not need to overclock your USB controller to get more than 1000Hz. As far as I know, theoretical maximum for USB3 is closer to 24000Hz, but there are considerations that are likely going to make that impossible. That said, 8000Hz input has been a thing for at least half a decade over USB3 because that one is realistically achievable.
In case you don't know, SweetLow allowed you to basically hack your mouse driver to force it to work in 8000Hz polling rate mode over USB3 for at least half a decade. The
Re: (Score:2)
While potentially more relevant for mice, I'm still pretty sure that they would have had identical game result if whatever utility they used to check the polling interval just placebo-style faked output to say 8,000 hz. At a 1ms sampling interval, the mouse motion is very well tracked. By comparison, 250 Hz was enough to be pretty convincing, but vaguely wrong at high motion for VR headset tracking (and 250 Hz was serviceable after adding motion interpolation to predict position and orientation based on t
Problem? (Score:4, Insightful)
Because so many people suffer from keyboard latency issues?
Re: (Score:2)
Because so many people suffer from keyboard latency issues?
Some will after they have read this AD. Psychosomatic effect only, but many people are clueless enough as to how things actually work to they will feel their perception is real.
Re: (Score:2)
Re: (Score:2)
Because so many people suffer from keyboard latency issues?
The fact that you don't doesn't mean professional gamers don't exist and there isn't a market of companies catering to them.
Me? It doesn't matter how long the latency of my keyboard is, it's a lower latency than the very frequency warranty claims that would need to happen for common razer products.
Re: (Score:2)
Unless the professional gamers are now androids, this doesn't help do anything but make them *feel* like something is different.
Even then if you had a theoretical zero-latency android playing a synthetic nvidia measurement utility as 'the game', it still wouldn't make much of a difference.
Razer deserves no marketing value for any market segment due to this claim as it is utterly pointless.
Re: (Score:3)
People apparently do.
Gamers in the 2000s often sought out PS/2 keyboards and mice because USB was too slow. This was understandable as USB keyboards and mice, besides only having 6 key rollover, were polled at astoundingly slow rates (around 40Hz on USB1, 100Hz for USB2). This was fine for ordinary users, but caused a lot of potential latency while gaming.
PS/2 worked on demand - if you pushed a key, the keyboard would immediately report the key even
Cloud account and Synapse required (Score:5, Informative)
Re: (Score:2)
Latency (Score:5, Interesting)
Did they measure the latency of having to log into a fucking cloud service in order to have all the functions on your keyboard actually function?
Did they measure the latency of having to clean off all the spyware and bullshit that can be installed via security escalation and arbitrary code execution flaws in their software?
I can't use it (Score:2)
Wait, doubleshot is new to them? (Score:2)
I have a red-backlit keyboard with mechanical keyswitches (presumably much cheaper ones -- mine are outemu red IIRC) which cost $35 and it has doubleshot keycaps.
Razer is grossly late to the party.
Re: (Score:2)
Doubleshot key caps ones are _old_. I think the original model-m already had them, so more than 30 years old.
Is that importat? (Score:2)
Re: (Score:2)
Under what kind of conditions would this make a significant different when compared with run-of-the-mill, inexpensive keyboards?
When you have a fast machine press the keys. Note that even fast key switches need considerable activation time if you do not want to damage them, so even then the difference would be small. For humans: No discernible effect.
Completely irrelevant (Score:2)
Normal keyboards poll around 70 times per second. That is already massively faster than any human can react. That advertised latency decrease serves only one purpose: To separate gullible techno-affine morons from their money.
Re: (Score:2)
Monster Cable strategy at play, works on audiophiles, also works on a lot of pro-gamer culture.
who cares if the USB-C cable is detachable? (Score:2)
Why the hell would you choose a keyboard -- or increase its score in your multifactor comparison of keyboards -- because its USB-C cable is detachable?
Why is a detachable cable of any significance whatsoever for a keyboard?
If you're using a non-wireless keyboard, you need the cable attached.
If you're not using the keyboard right now, you don't care whether it has a cable attached.
The only possible reason to want a detachable cable is if you are critically short on cables and need to plug in a scanner or som
Re: (Score:2)
Re: (Score:2)
All right, I guess that makes sense, if Razer stuff is that fragile. I have been using the same Dell keyboard since about 1997 and no part of it has ever broken.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Why stop at 8000Hz? (Score:2)
Why not boost it up to 2.4Mhz and make each button wireless?
Re: (Score:2)
And you'd have to type in the WiFi password on each one of them using only that key.
"Near Zero" latency (Score:2)
8000 Hz - Bah, that's nothing :) (Score:2)
Back in the ZX Spectrum days (bless your soul Clive) you could poll the keyboard as often as you could issue an IN instruction IIRC. 3 cycles @ 3.5 MHz = ~857 ns latency.
Of course, it also simplified that we didn't have triple-buffering so we could just check the state in real-time whenever it was required.
Modern problem with mechanical keyboards (Score:2)
I really like mechanical keyboards with some travel to them.
However, over the years I've run into one problem with them - they are just too loud for anywhere you are working around other people.
Before that was an issue in offices, these days it's been a problem every since my wife started working at home full time as well when the pandemic started...
I wonder if this Razer, or any mechanical keyboard can get close to the sound levels of laptop keyboards...
Re: (Score:2)
Re: (Score:2)
actually, it's quite fast, key debouncing tends to use much longer intervals.
Re: (Score:2)
...and the earth takes 24 hours to spin once on its axis. So what?
If you want faster, don't use switches that require slow debounce techniques. 8000 Hz is slow and trivially achievable. Apple 2's had keyboard response rates within an order of magnitude of 8000 Hz back in the 70's. There was no reason to be faster because they had a 1MHz CPU!
Re: (Score:2)
Only crappy debounce. Correct debounce limits repeats, rather than delaying detection.
Re: (Score:2)
so you don't repeat in some interval, I don't see how that contradicts anything I said, or even where I said you had to wait first!
Re: (Score:2)
They probably mention 8 kHz because it is the maximum USB2 High Speed polling rate. Notice that polling rate is not the same as the serial interface baud rate you mentioned. E.g. USB2 High Speed bit rate is 480 Mbits/s. So USB is doing better than USART in the speed category. Latency is worse since 8 kHz polling rate puts lower limit of 125 us for latency. If they would use USB3 Super Speed then they could use asynchronous interrupts and achieve much better latency than 125 us. They would not be limited by
Re: (Score:2)
"Conclusion: their advance is not in the communication interface with PC but in the switches."
The proper conclusion is that there is no advance at all, only marketing.
"They probably mention 8 kHz because it is the maximum USB2 High Speed polling rate."
Sure, and that would be expected if this were marketing rather than technical. Internal polling rate of switches does not need to be tied to USB polling rate.
"Latency is worse since 8 kHz polling rate puts lower limit of 125 us for latency."
Worse than what?
Re: (Score:3)
The 8000 Hz figure is not the transfer rate -- It is the polling rate: how often the host asks the peripheral if any key has changed.
The whole keyboard's state is transferred up to 8000 times a second.
If we're going to discuss improvements to USB HID communication, I'd like to suggest that the OS/driver should model the period of the application's need for key input and poll only right before expected demand -- thus polling at frame rate and making the 8000 Hz figure into how granular the timing of the prot
Re: (Score:2)
Nobody sane ever went above the usual 70Hz polling in keyboards (my own measurement on 3 different ones, other rates may exist) because there is absolutely no need and absolutely no noticeable benefit to polling faster. This "innovation" is like investing a ton of money to make the toilet flush faster.