USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Everything about latency. Tips, testing methods, mouse lag, display lag, game engine lag, network lag, whole input lag chain, VSYNC OFF vs VSYNC ON, and more! Input Lag Articles on Blur Busters.
Futuretech
Posts: 35
Joined: 11 Oct 2020, 23:52

USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Futuretech » 12 Oct 2020, 00:49

I'd like to inquire why USB 3 and the future spec 4 have been so neglected by manufacturers of mice and keyboard for the purpose of bumping up polling rates.

I'm aware no one cares about anything past USB 2.0 really USB 1/1.1 for the information sent by keyboard and mice. I mean hell some people even use old PS/2 ports for their keyboard particularly the NKRO and other low-level usages PS/2 still leverages.

Many years ago when USB3.0 came out it was touted as a great thing by some for speeding up transfer rate. But completely ignored by all manufacturers of keyboards and mice. They thought it was useless much like PCIe 3.0 and even current 4.0. While the fools believe it's just the transfer bandwidth improvement behind the scenes there are countless updates to the protocol. Hell boggles the mind they can keep backwards compatibility on their PCIe ports.

One thing that was discussed and only recently as of the last few months discussed even though I was thinking about this YEARS ago. Is the usage of high polling rate or to be more specific in USB3.0+4 terms simulated polling rates.

While polling rates have been eliminated to a psuedo- if not outright interrupt system in USB3.0. I fail to understand why developers of mice and keyboard didn't jump on the bandwagon and used the higher simulated polling rate of the interrupt mechanism on the USB3.0 protocol back when USB3.0 came out. The fact that it works at the 8,000Hz -to- 24,000Hz level is proof enough we could have for years now 24,000Hz simulated polling rates for mice and keyboard. Even making PS/2s low-level functions obsolete as the USB3.0+ port would have out muscled the polling rate simulation of PS/2s interrupt style.

According to the article on Overclock.net which was released sometime in 2012/2013 and promoted by people like Manyak and other prominent multi-forum gamers hell even a few Hard[OCP] members used the guide.

[paraphrasing the guide as I can't find it in search]

USB 3.0 protocol is not polling based rather it is simulated to polling rates to give an example. The micro-frame system or 125µs or about 8,000Hz polling rate. The maximum for USB 3/4 would be about 0.04166666666666666666666666666667 or about 41.7µs or 24,000Hz. The USB is "simulated" to respond at those levels but it's based on an interrupt or psuedo-interrupt system. The USB protocol shows full duplex so a non-series system so parallel firing. It would also hit the CPU quicker as there is no polling so the system is interrupted or psuedo-interrupted and thus gain an advantage to move when you decide to move. Not miss a poll and perform various details to continue moving or stopping the mouse or keyboard for that matter.

{The only reasonable search quote found on Overclock.net apparently this person knew perfectly well what I've known since about the same time frame circa 2013 or so when I investigated said USB 3.0 benefits. The guide is from 2012/2013 right when USB 3.0 was mass known and coming out as the future. Note: This isn't USB-Type-C plug but rather the very protocol in which Type-C was based on and future revisions of USB and their whole "Gen" nomenclature.}
wo1fwood wrote:Anyway, from my understanding the theoretical maximum speed for USB 3.0 mice would be approximately a 24,000Hz polling rate, but I would doubt that speed would be achievable due to other considerations (125µs is more than achievable however). That being said, USB 3.0 is also non-serial, so it can send and receive packets at the same time. Also remember that transmission packet sizes are even smaller than that the standard 125µs window, so there's still some sort of transmission bottleneck with 3.0 if talking about getting a single packet of data from the mouse to the host (that and packet sizes aren't always the same iirc). This is not my strongest area though as there's a lot about USB 3.0 that I never ended following up on in detail.
Is there a specific reason why manufacturers didn't jump on USB3.0 benefits to mice and keyboard?

(I'm aware they didn't want to retool or recreate their contracts for USB3.0 and even USB3.0 was a headache coming out with various implementations from PCIe slots AiCs to final implementation of USB directly to mobos and hell even now a days we are still dealing with USB-Type-C trying to replace every port and the whole "Gen" USB nomenclature(mentioned before) along with USB 4 coming out and the whole thunderbolt mixing and whatnot.)

It's a big headache only recently have I seen some mice using USB3.0+ and Type-C but mostly cheapy desktop/laptop mice from Microsoft and a few other manufacturers using USB-Type-C as a one size fits all solution to USB costs.

Anyways any idea besides what I mentioned why people didn't jump to the better standard of USB3+ for mice and keyboard increased performance of "simulated" polling rates?

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Chief Blur Buster » 12 Oct 2020, 01:57

Moved to Latency forum, as not a display related topic.

8000Hz kind of already pushes limits — given the stall of performance increases per core. We need 8000 Hz out in the open first to allow software to optimize first.

Further precision improvements are likely welcome but better yet is timestamping the polls mouse-side instead with microsecond accurate timestamps from the mouse’s microprocessor/microcontroller. This will help de-jitter polls and reduce rounding effects too.

A new optional game-accessible mouse protocol, in a manner of speaking. Longshot to see it happen soon, but the discussion could begin as a potential method of improving 8000Hz mice, and letting polls become more asynchronous without jitter or performance degradations.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

User avatar
schizobeyondpills
Posts: 103
Joined: 06 Jun 2020, 04:00

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by schizobeyondpills » 17 Oct 2020, 08:32

Windows kernel timer has a limit of 0.5 ms interrupt intervals. the source of time itself within os. why do you believe you can plug anything of 8000hz or more in 99.99999% of systems and expect no perf degradation? and think it will work properly.

pixart sensors per their specsheets have a temperature operating temp of 40deg c, thats junction inside not on top where its cooler. heres a pic of xm1 on 1000hz from being moved with OPEN shell for 5min. scale up for closed shell usage and being used with a sweating palm and ofc working sensor in a fps game. just moving it causes spikes of mcu sensor to go up to 36-40 deg C

Image

Alpha
Posts: 132
Joined: 09 Jul 2020, 17:58

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Alpha » 18 Oct 2020, 20:05

schizobeyondpills wrote:
17 Oct 2020, 08:32
Windows kernel timer has a limit of 0.5 ms interrupt intervals. the source of time itself within os. why do you believe you can plug anything of 8000hz or more in 99.99999% of systems and expect no perf degradation? and think it will work properly.

pixart sensors per their specsheets have a temperature operating temp of 40deg c, thats junction inside not on top where its cooler. heres a pic of xm1 on 1000hz from being moved with OPEN shell for 5min. scale up for closed shell usage and being used with a sweating palm and ofc working sensor in a fps game. just moving it causes spikes of mcu sensor to go up to 36-40 deg C

Image
Does USB port matter as far as latency for XM1 or really any mouse? I've read on AMD machines they should be plugged into the port that directly ties into the CPU instead of the Chipset but I don't recall.

Meowchan
Posts: 40
Joined: 17 Jun 2020, 02:06

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Meowchan » 19 Oct 2020, 06:36

40 Celsius is very cool operating temperature for electronics. I believe the reason Pixart specs their sensors to those temperatures is because that is what they currently tend to run at so they design around it. If higher hz sensors started producing more heat and even if better cooling was inadequate then Pixart and others can design around the new higher operating temperatures.

It is my understanding that 8khz mice already work in Windows, either via hacked driver or the new gen that is in the works. The 0.5ms Windows timer is not a hard cap of 2khz on hardware.

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Chief Blur Buster » 19 Oct 2020, 16:46

Timer accuracy varies between systems. There are new sub-0.5ms high precision timer workflows available nowadays; I found some of them worked reliably enough for beam racing purposes (1 pixel tearline movement in ~10 milliseconds) during Tearline Jedi experiments (treating VSYNC OFF tearline as rasters). System A had the 0.5ms granularity but System B had near-microsecond granularity. I was able to eliminate busywaits and still retain near busywait-precision with certain types of timer events. I wouldn't trust this to be universal, but it already exists out there, which disproves the point that all system timers are limited to 0.5ms timer precision.

Though it degrades performance of some systems more than others (like early 1000Hz mice on older systems), as these mice come out, many systems will become more optimized for 8000 Hz poll operation. The mouse testing utilities show that systems can keep up provided optimizations are made to all weak links.

I also did not notice any heat difference between the 1000Hz mouse and 8000Hz mouse; the sensor readrate on the 1000Hz mice is approximately 12-16KHz already, and the sensor readrate on the 8000Hz mouse is 20KHz and actually seems slightly cooler, probably due to improved electronics. I wouldn't worry about the mouse-heat aspect.

But yes, peformance can degrade on some systems because not all systems can handle a 8KHz poll rate.

The release of a new accessory has a way of forcing manufacturers to re-optimize around it. What's true is that many systems are suboptimal, but is definitely not the raison d'etre NOT to release a device. It's also why Razer released a number of prototypes so that the industry can educate itself around 8000 Hz, and optimize systems. Throw them out into the open and see what sticks.

As we speak, I am building mouse-benchmarking software.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

User avatar
axaro1
Posts: 627
Joined: 23 Apr 2020, 12:00
Location: Milan, Italy

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by axaro1 » 19 Oct 2020, 16:57

Chief Blur Buster wrote:
19 Oct 2020, 16:46

The release of a new accessory has a way of forcing manufacturers to re-optimize around it. What's true is that many systems are suboptimal, but is definitely not the raison d'etre NOT to release a device. It's also why Razer released a number of prototypes so that the industry can educate itself around 8000 Hz, and optimize systems. Throw them out into the open and see what sticks.

As we speak, I am building mouse-benchmarking software.
Is this one of the rare cases where using USB 3.1 Gen 1/2 can actually be beneficial over USB 2.0 / 3.0? Is the increase in bandwith capability an indicator of better polling rate behaviour from the USB controller or is it mainly an issue fixable by drivers/changing the way the Windows ecosystem handles USB peripherals?
XL2566K* | XV252QF* | LG C1* | HP OMEN X 25 | XL2546K | VG259QM | XG2402 | LS24F350[RIP]
*= currently owned



MONITOR: XL2566K custom VT: https://i.imgur.com/ylYkuLf.png
CPU: 5800x3d 102mhz BCLK
GPU: 3080FE undervolted
RAM: https://i.imgur.com/iwmraZB.png
MOUSE: Endgame Gear OP1 8k
KEYBOARD: Wooting 60he

User avatar
Chief Blur Buster
Site Admin
Posts: 11647
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by Chief Blur Buster » 19 Oct 2020, 17:57

axaro1 wrote:
19 Oct 2020, 16:57
Is this one of the rare cases where using USB 3.1 Gen 1/2 can actually be beneficial over USB 2.0 / 3.0? Is the increase in bandwith capability an indicator of better polling rate behaviour from the USB controller or is it mainly an issue fixable by drivers/changing the way the Windows ecosystem handles USB peripherals?
USB 2.0 is already overkill bandwidth for mice which technically don't need 480 Mbps of bandwidth. No performance improvements were observed for USB 3.0 ports.

Contention (interference between high-Hz poll devices) is more important than whether port is USB 2 or USB 3. Give each high-Hz device its separate trunk, a completely separate USB hub tree -- which may mean ports far away from each other, and/or using a USB PCI Express card for a separate high-Hz poll device. The total bandwidth is low.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

User avatar
axaro1
Posts: 627
Joined: 23 Apr 2020, 12:00
Location: Milan, Italy

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by axaro1 » 20 Oct 2020, 03:16

Chief Blur Buster wrote:
19 Oct 2020, 17:57
USB 2.0 is already overkill bandwidth for mice which technically don't need 480 Mbps of bandwidth. No performance improvements were observed for USB 3.0 ports.

Contention (interference between high-Hz poll devices) is more important than whether port is USB 2 or USB 3. Give each high-Hz device its separate trunk, a completely separate USB hub tree -- which may mean ports far away from each other, and/or using a USB PCI Express card for a separate high-Hz poll device. The total bandwidth is low.
Ok, thanks explaining
XL2566K* | XV252QF* | LG C1* | HP OMEN X 25 | XL2546K | VG259QM | XG2402 | LS24F350[RIP]
*= currently owned



MONITOR: XL2566K custom VT: https://i.imgur.com/ylYkuLf.png
CPU: 5800x3d 102mhz BCLK
GPU: 3080FE undervolted
RAM: https://i.imgur.com/iwmraZB.png
MOUSE: Endgame Gear OP1 8k
KEYBOARD: Wooting 60he

User avatar
schizobeyondpills
Posts: 103
Joined: 06 Jun 2020, 04:00

Re: USb 3/4+ Future USB 8,000Hz -to- 24,000Hz polling rate

Post by schizobeyondpills » 25 Oct 2020, 14:51

Chief Blur Buster wrote:
19 Oct 2020, 17:57
USB 2.0 is already overkill bandwidth for mice which technically don't need 480 Mbps of bandwidth. No performance improvements were observed for USB 3.0 ports.
Hello friend, its not about bandwidth, its about latency, latency is absolute measure of time it takes for any action to complete aka change of state, meaning its value applies to 1 packet as much as to 10 packets. unlike bandwidth which is useless.

Bandwidth = load of bridge per two lanes
Latency = time it takes for 1 car to cross the bridge

Meowchan wrote:
19 Oct 2020, 06:36
40 Celsius is very cool operating temperature for electronics. I believe the reason Pixart specs their sensors to those temperatures is because that is what they currently tend to run at so they design around it. If higher hz sensors started producing more heat and even if better cooling was inadequate then Pixart and others can design around the new higher operating temperatures.

It is my understanding that 8khz mice already work in Windows, either via hacked driver or the new gen that is in the works. The 0.5ms Windows timer is not a hard cap of 2khz on hardware.
40 celsius is nowhere near very cool for electronics. temperature scaling is exponentional with frequency of those electronic devices you think about. just because something works doesnt mean its working at optimal performance, and it needs to work with optimal performance 8000 times a second.



Pixar doesnt even give you full documentation of the registers/firmware impl if you are their customer, so no, their specs are adjusted for marketing just like all others are ( TiN from evga who designed top extreme OC mobos warns about spec sheets being marketing material here https://xdevs.com/article/tcr_test/ )
Chief Blur Buster wrote:
19 Oct 2020, 16:46
Timer accuracy varies between systems.
correct, however my post was meant to say that source of tracking time within the whole OS is done with interrupts at 0.5ms intervals.
having a mouse do 8000 interrupts a second means that same functions of input pipeline are called 8000 times in time slices of 0.125ms intervals which under system load playing a game will cause massive issues of microstutter, especially for eSports optimized setups that remove alot of smoothing and compensation. as you optimize for performance you do so by stripping away compensation, this reveals the microstutters more and more.
Chief Blur Buster wrote:
19 Oct 2020, 16:46
which disproves the point that all system timers are limited to 0.5ms timer precision.
windows time (not timer) source occurs at 0.5ms intervals. https://randomascii.wordpress.com/2020/ ... le-change/ __rdtsc() or __rdtscp() works in cycles, ofc.
Chief Blur Buster wrote:
19 Oct 2020, 16:46
Though it degrades performance of some systems more than others (like early 1000Hz mice on older systems), as these mice come out, many systems will become more optimized for 8000 Hz poll operation.
i hope so, but i doubt it. there's a limit to how much you can push a general purpose OS. windows 10 already struggless with 1000Hz (under load).
Chief Blur Buster wrote:
19 Oct 2020, 16:46
the mouse testing utilities show that systems can keep up provided optimizations are made to all weak links.
mouse tester is a flawed outdated tool. another thing to keep in mind is mouse testing utilities are not ran under load and under 3h+ gaming sessions.
Chief Blur Buster wrote:
19 Oct 2020, 16:46
I also did not notice any heat difference between the 1000Hz mouse and 8000Hz mouse; the sensor readrate on the 1000Hz mice is approximately 12-16KHz already, and the sensor readrate on the 8000Hz mouse is 20KHz and actually seems slightly cooler, probably due to improved electronics.
waiting for Razer The_Fiend to provide some measures(under load with full system specs).
Chief Blur Buster wrote:
19 Oct 2020, 16:46
It's also why Razer released a number of prototypes so that the industry can educate itself around 8000 Hz, and optimize systems.
should've sent me one since i have lowest ram latency in the world and without doubt most stable setup ( 10k euros + in low latency equipment ;) )
Chief Blur Buster wrote:
19 Oct 2020, 16:46
As we speak, I am building mouse-benchmarking software.
Great, please make it in C++ and post source code online, MouseTester is very bad and outdated using C# on 32bit platform(WoW64)

also can we finally have inline new post quote selected text? ty <3

Post Reply