Me too, I was worried maybe a USB 2.0 hub with my keyboard and mouse connected would be worse than connecting both to a dedicated USB 3.0 port, even though both are USB 2.0 devices of course. But it seems like USB hubs don't add any big amount of lag and it's barely even measurable.
Brilliant test, exactly the scenarios I was looking for! And the results are so surprisingly positive! Thanks for tackling this question the right way and absolving me of my latency concerns. Subbed!
Thanks for taking the time to conduct and publish these test. I'm sure we've all seen hundreds of posts with anecdotes or theoretically arguments over whether hubs affects input lag. But it takes controlled testing to get a definitive answer. This makes me feel better as my keyboard, mouse and sound-card are connected through a KVM switch (acts like a hub but lets me switch between two systems). I guess I'll have to go back to blaming internet lag and bad hit-boxes when someone beats me in CS:GO.
In theory, it's not surprising at all that the latency does not increase even during a bulk transfer, because that's the whole idea behind the USB polling scheme -- since the traffic is completely shaped by the host controller, it can bound the latency for interrupt-class endpoints and the bulk transfer cannot interfere. It's very nice to see that being confirmed in practice by testing, however! If anything, it might be interesting to test whether the latency is affected by having more devices on the same USB root hub, since that means more devices that need to be polled, which might affect the overall polling rate.
Well it does add additional lag to the raw signal, since speed of electricity through copper = speed of light. It's just that the timing of the USB protocol itself forces the USB controller to wait for the next packet anyway, to keep everything at a very predictable and steady update rate, much higher than what the extra copper length could introduce.
Well, by definition, electricity can't travel at the speed of light due to it being dependent to some degree on baryonic interactions. If it were solely dependent on bosons (like the photon) it would be able to move at the speed of light. I'm not a nerd; _you're_ a nerd!
Great video. Curious why click latency isn't effected by hubs but sometimes double clicks can be registered because of hubs. I've experienced this several times and have read other cases of it happening online.
thank you for this awesome test. i always hesitated to use my monitors usb hub for my input devices because of being worried that it might cause "problems", but now i have to rethink my cable management thanks to you!
Thank you for yet another great test! I haven't been able to use one of my USB hub inputs because I feared I would get additional input delay, so I had to plug a larger connector in which blocked one of the inputs. Now I can swap that and use all of the inputs :)
From the moment I saw the title, I was prepared to answer "no"....the reason: USB is a polling communication, with the controller as master (why you can't just connect 2 PCs with a simple USB cable). While it's communicating with a device, that is the only time the device can give back data (unlike ethernet where packets can come and go independently). So all actual communication is directly between USB controller and device. All that USB hubs do is identify to which port they have to make ELECTRICAL connection. If that would be delayed, USB controller would close the communication window and "disconnect" the device. The only delay a hub could do is the "repeating" of the electrical signal, but even that should be done shorter than the USB clock signal (I assume; even if longer, we are talking nano/micro-seconds, not milliseconds).
@@Tufet Delay, no. The only worry with extenders is signal degradation and wire resistance, but 1m should be negligible. I personally use 3m USB3 extender for my simracing setup, which goes to hub with KB, mouse and wheel. The wheel has it's own power, so it's a small load on the hub. But I would avoid charging a phone, even with USB3.
@@mateiberatco500 i will use for a headset with rgb and condesator microphone fifine k669, should it be ok? My USB hub has a type c power supply port as well
@@Tufet I don''t know how much power they need, but if you have usb3 up to the hub, it should be fine. USB2 hub/extender might be problematic, but I guess you should see that immediately when listening at high levels. My biggest worry is HDDs, as that could screw their content on writing.
Thanks for doing this video. Just thought of this question this morning when I was considering getting a USB switch to quickly swap my mouse and keyboard between my gaming computer and work laptop. Didn't want my work life to potentially slow my gamer reflexes!
This is a cool video. Wondering if there’s an update. I noticed if results had been on a line graph showing normalized data (perhaps a simple % increase/decrease) then there are a number of places where you said “no change” that was actually a 5% change. Perhaps you could argue that it’s negligible, but it’s not zero.
Finally found someone covering a related topic! I was wondering if I introduce any sensor error or input latency while using an USB A adapter for USB c gaming mouse or keyboard. I pick them to have a 5 ms advantage, but if it's diminished by the adapter and not having a USB c port in my pc, it doesn't make sense of course. Thanks!
Great video! (I was planning the very details of my new PC setup when this question came up.) I expected to only find the usual "guesswork / non technically founded" discussions, but your video really blew my mind. Physically _measuring_ and then _veryfiying_ the results is a very scientific approach, I really appreciate this :).
I genuinely think you are the best tech UA-camr on the planet. You have answered all the tough gaming questions with objective test results on so many topics I have needed answers from. We all really appreciate your hard work.
I never actually thought about it. I had a bit thought, about that distance to motherboard a bit longer when connecting mouse/keyboard to front panel instead of rear, but it is like 0.5 or less meters, what is like maybe 0.01ms delay. So, thanks for such informative video!
This makes me wonder about raw PS/2 mouse vs USB->PS/2 converted mouse vs raw USB mouse. I've heard that PS/2 has its benefits because it sends interrupts to the CPU but I've never looked into if it actually works.
You earned yourself a sub! This nice, good to understand and accurate testing is what i personally prefer to some "5 min production" videos. Keep the great work up!
I love how every website out there swears that usb hubs gives you a shit ton of input lag and you should never use one for gaming, but they provide no data.
Fantastically in-depth, but to the point video. Thanks! I use a mac and PC on the 1 monitor and unfortuntely most monitors only have 1 port to connect to its internal hub so i'll be confidently be using a USB 3.0 switching hub!
I have a powered USB 3.0 10 slot replicator. I was wondering if I should plug my Mouse/KB dongles directly into the PC or on the hub itself with other stuff. Good video.
I've run a USB 2.0 hub for gaming in the past in specific setups and can for sure say there is a difference in input latency. I was running a SNES emulator while talking over voice and could measure the difference in input delay. I was streaming at the time as well so I had footage somewhere and talked about the issue with and without the USB hub. I ended up purchasing a powered USB 3.0 hub because of that problem.
Interesting results! At 200fps (5ms per frame), I would have expected much less variance in the min/max results. Can you publish the raw results of each trial/test? I would like to see the distribution of the results of each test. As a follow up, I'd be very curious to see a comparison of 125/250/500Hz mice vs 1000Hz, in both min/avg/max as well as a latency distribution.
PS2 Mice ARE better for input lag since they are interrupt driven. This means that as soon as you move, and the mouse has data to send, the mouse notifies the CPU that the mouse should move right now. With USB polling, when you move the mouse, it sits there holding that data until the PC asks for it 1ms later. But when we're talking about 33ms input lag vs 32, it's not worth the trouble really. If we're looking at 125/250/500Hz refresh rates, it's the same, but the delay between polls is now 2ms/4ms/8ms.
Interesting. Then do you know why isn't there a PS/2 gaming mouse market as far as I can tell? "Pro gamers" buy hardware to get 500+ FPS on CSGo, they would buy a mouse that gives them 1ms better input lag I think.
-BY design, delay will be added. It might be just millisecond or two, but its still an existing delay. USB is not designed with daisychaining in mind, and its polling based system, which will literally always add delay, from your case, to that fancy USB hub on your monitor, keyboard, or standalone, and the more you chain, the longer the delay gets. However, an extension cable *shouldn't* add more delay, as there is no inbetween point where input and output has to be handled. Now, you can decide if few milliseconds is worth worrying about or not, thats a whole other topic.
What? USB WAS designed for daisy chaining, this is why single controller can adress and poll up to 127 devices/hubs. Obviously it wasn't designed to daisy chain 50hub's and then mouse at the end, but no system was.
i do think about this, ive troubleshooting the hell out of it and think i got it under control, well done for shining a spotlight on this usb can be messy, cpu, power settings, network, windows 10, nvidia, theres, chipset, hard ware etc etc a long list
Amazingly clear answer and explanation of a commonly asked question. Thank you, sir. Your diagrams were exceptionally clear... I just finished the video. Amazingly concise and the results are somewhat surprising as you say. Webcams now connected to monitor to eliminate cable run. Thanks again.
I am going to conect two monitors to the same pc at different roms for playing. I ordered this: - 10 meters USB 2.0 active extensión usb cable (not 3.0 for costs). - USB 3.0 USB hub that will be connected to the extension usb cable. - 10 meters HDMI cable. I'll conect a mouse, keyboard and Xbox controller. Just for gaming and multimedia. This video encouraged me to do this because I was worried about input lag. I hope that the USB hub conected to the USB 2.0 10m extension cable doesn't affect input lag. I think that it was the only missing part at your tests. I'll try to make a comparation with you same metod but with a smarthponecamera at 960 fps (Galaxy S9).
I wish you could make some test with usb extanders/repeaters/usb-lan adapters with 15m or more for guys want to try to outsource pc from living room. You've done amazing job. Subscribed!
This is strange. I got a different result with my test using my MIDI keyboard and Audacity. I recorded myself tapping the drum pad while FL Studio played a sharp sound, and I looked at the waveform on the screen. I could measure the space between the peak of my tap to the beginning of the triggered sound. I found that connecting the keyboard through an unpowered USB 2.0 hub consistently added to the lag. I can't remember if it was 5 or 10 ms.
Bro I haven't watched the video yet but I just wanted to say that I literally just had this thought while on the toilet and immediately after that, I go on youtube and I find you with this video. God it's like you read people's minds or something. God bless you man you are doing life saving work.
Thank you so much!!! Thought about this so much as I need to switch computers but only have one set of mic, mouse and keyboard for my streams and recordings. This helps a lot mate!
This definitely used to be more of an issue with USB devices stressing old CPUs more than their PCI or PS/2 equivalents. I used to avoid USB network devices and such like the plague, but now with the limited PCI-E lanes and multi-core on modern processors I do the opposite. lol how times have changed.
I mean there is a difference, it's within 3 or 4 percent, but it does seem that there is a difference between the hypothetical fastest and slowest instances. It would be interesting to see if this type of test could be automated to allow for a much larger sample size. Awesome channel by the way, nice to see an actual breakdown of this kind of thing rather than just seeing peoples opinions and hearsay
Hello! I was wondering if with the introduction of new technologies and possibilities you could do an updated video! Is there a lag input between using a usb c hub with an hdmi connection to the monitor when compared to simply using the hdmi port in the pc?
Short answer no, long answer, for high bandwidth USB 3 ports no but usb 2.0 maybe like 1 or 2 ms which for some may be worth while moving shuffling your USBs around, I know I moved my mouse directly into my laptop after this and I have a lot plugged into my hub, sometimes even hdmi, ethernet and drives.
Nice, it would be interesting to see the results on the moving mouse test. I would advice to, instead of "pulling the mouse really fast", to use some weight attached to the rope, for repeatibility sake...
very good video and this gave me the confidence to connect my mouse to usb 3.0 hub. Only improvement proposal would be testing longer cables. I have 3m long cable and I think that many are using also 5m long cables.
The reason (I think) the stress and additional hubs cause no delay is that the OS or the mouse itself probably gets to tell the USB hubs to prioritize the mouse and not to delay its messages (delaying messages to combine them could reduce stress on the cpu for some devices).
mouse and keyboard take priority over every other action/imput that goes through the chipset. it is really only the length of the cable or the performance of the device that will create lag
The reason why you couldn't/shouldn't see any difference between back 2.0 and 3.0 ports is because those are usually operated by same physical controller. (USB3.0 controller can work ports as 2.0 no problem). Even on my old Z87 its just single 3.0 controller for all back USB ports. And the reason why there was no difference with added controller load is polling. Devices are polled at given frequency and no device can "steal" all communication.
Should use a simple application that changes colour on receiving mouse input. Games can introduce latency on input, also as you saw framerate made it hard to collect data for you last test.
No input lag, but occasionally my mouse freezes up completely if using a powered hub, my controller locks up, and my blue snowball mic doesn't even receive power to turn itself on.
Thank you. Exactly what i needed and perfectly testet and explained. Now i can buy the cheap 2.0 USB switch instead of expensive 3.0 and dont worry if i introduce response lag.
Oh god thank you. I was going crazy trying to find a decent AM4 board with enough USB ports for my needs but it was either Gigabyte (which i dont like) or a very expensive crosshair. Now i just get a good USB Hub and a cheap board that fits my needs :D
If he woud go really into detail, how about testing the effect of different anti virus solutions (Windows Defender, Avast, Kaspersky, AVG, Antivir, BitDefender, Sophos, Panda in their free versions to keep the cost down) and some of their settings (like "game mode" and modules like behaviour scan)
Yeah i think every device gets its own transmission window within the 1ms USB frame, so far for interrupt endpoints like HID input. Then isochronous endpoints get a piece of the frame, like soundcards, webcams and capture cards, and finally bulk devices like USB disk volumes. The hubs can't buffer any data either if i'm not mistaken, the timeframes are host-defined. It's pointless to test a USB 3.0 harddisk together with a USB 2.0 mouse though, because they're not even on the same bus, even if they're on the same physical port. You can just reconfigure the harddisk to USB2.0 for a potentially more interesting test, by using a USB 2.0 cable instead of the one that came with the disk, which however will still show nothing changing, i think. The common webcams like C920/C922 are USB2.0 and use a lot of the bandwidth, and that's still not a problem. One might expect differences in behaviour between cheap mice, operating on a LowSpeed bus, and gaming mice, which operate on FullSpeed, because the LowSpeed devices consume more time out of a frame, but it makes them share and simply staggers the frames during which they talk, so say keyboard talks during one frame and mouse during another, this is why there's 8ms between messages. And yet, i don't expect this to actually be an issue still, because the bus must foresee enough time for interupt endpoints, though isochronous endpoints eat into the available time per frame overall.
Good to hear. Plz do a vid with inputlagtest gsync+ vsync on and capped framerate ingame because gsync.on+vsync off+capped framerate can cause partial tearing. So best experience is both on.
The partial tearing is caused by "micro FPS spikes" - the frame delivery is not on point. To fix this you need to do one of the following things: 1. enable the high performance power plan inside windows, that might fix the issue 2. lower the FPS cap _(i.e. 140, instead of 142 on a 144Hz display)_ 3. use RTSS instead of the FPS limiter built into the game, it increases the input delay by 1 frame but RTSS has more stable frame delivery 4. also enable v-sync _(i am not a fan of this since I paid 200$ extra for the g-sync support of the monitor ;-) )_ When you have a FreeSync or G-Sync monitor and cap your frame rate below the displays refresh rate, then v-sync on or off does not make a difference (however that is on my system where this did not cause tearing near the bottom of the screen) ua-cam.com/video/mVNRNOcLUuA/v-deo.html
Battle(non)sense thats not rly true I have capped all my games at 144fps (165hz). And playing with gsync and vsync on because there is a difference. Especially battlefield 1 does partial tearing at the bottom of the screen. There are threads that say this is the best way you configure gsync with both on and cap fps. They also did input lag tests which show no difference if vsync is on. But it fixes that partial.tearing. so I want a second lagtester to know if their results are true. Plz just try it in some games. Gsync was meant to work with vsync (nvcp). You can notice a difference in smoothness with it on in some games. If you do not cap fps and are in gsync range there is no partial tearing. It just occurs with capped fps with vsync off.so if you cap your fps vsync helps. Just wanted to ensure that there is no difference in input lag. (Blurbuster forums input lag thread gsync)
Battle(non)sense so according to their results its better to use vsync on(nvcp on, ingame off) and ingame limiter than vsync off and rts (which doesnt rly help with partial tearing in bf1). So I just wanted a second good input lag tester if he (you) has the same results :-)
TL;DW -- USB Hubs Are Input Lag Monsters? No.
MVP
Nigga needs a scholarship for his comprehension skills
U da man
The thing is PCI USB cards are good cause it reduces IRQ and input lag.
You daaa man!
The gaming scene needs more people like you. Great video.
Back this guy on Patreon so he can get a better high speed camera!
Googled this because I'm out of usb ports, did not expect to find such a well tested answer. new sub
Hi. I'm using a gaming laptop and also running out of USB ports. Have you tried using a USB hub? How was your result?
Thanks in advance.
It works very well for me i i use usb hubs
Me too, I was worried maybe a USB 2.0 hub with my keyboard and mouse connected would be worse than connecting both to a dedicated USB 3.0 port, even though both are USB 2.0 devices of course. But it seems like USB hubs don't add any big amount of lag and it's barely even measurable.
@@laLuminescence ilmao same problem did you try i?
Same
Conclusion: 6:24 it does not affect the input lag
Brilliant test, exactly the scenarios I was looking for! And the results are so surprisingly positive! Thanks for tackling this question the right way and absolving me of my latency concerns. Subbed!
Never even tought about this till you made this video
He's lying, don't listen, Danny!
U really went deep... Very deep.
Give this man a nobel price for this research.
You are really answering the questions humanity needs answers to
Thanks for taking the time to conduct and publish these test. I'm sure we've all seen hundreds of posts with anecdotes or theoretically arguments over whether hubs affects input lag. But it takes controlled testing to get a definitive answer.
This makes me feel better as my keyboard, mouse and sound-card are connected through a KVM switch (acts like a hub but lets me switch between two systems). I guess I'll have to go back to blaming internet lag and bad hit-boxes when someone beats me in CS:GO.
exactly. My chipset caused problems with usb hub delay, for example.
6:40 - That's a Dirty Bomb. If someone is interested.
F
I was not expecting this video to be so informative. I guess LTT has been lowering my expectations for tech videos. Good job!
aged like fine wine lol
@@Maebbie fr
I came here hoping to find an excuse for why I'm getting my butt kicked in CoD...now I must look elsewhere
In theory, it's not surprising at all that the latency does not increase even during a bulk transfer, because that's the whole idea behind the USB polling scheme -- since the traffic is completely shaped by the host controller, it can bound the latency for interrupt-class endpoints and the bulk transfer cannot interfere. It's very nice to see that being confirmed in practice by testing, however!
If anything, it might be interesting to test whether the latency is affected by having more devices on the same USB root hub, since that means more devices that need to be polled, which might affect the overall polling rate.
Very good to know that an extension cable is completely fine to use without any additional lag.
Well it does add additional lag to the raw signal, since speed of electricity through copper = speed of light. It's just that the timing of the USB protocol itself forces the USB controller to wait for the next packet anyway, to keep everything at a very predictable and steady update rate, much higher than what the extra copper length could introduce.
Well, by definition, electricity can't travel at the speed of light due to it being dependent to some degree on baryonic interactions. If it were solely dependent on bosons (like the photon) it would be able to move at the speed of light.
I'm not a nerd; _you're_ a nerd!
@@redchris05 you got the you're part right tho idiots dont check their grammar before typing like complete a holes.
But 2m at speed of lights is an increase of a few nanoseconds? You might get in to the millisecond category when your cable starts to be 100km...
@@aplatypus7651 this aint about gramar, check you're privilege
Great video. Curious why click latency isn't effected by hubs but sometimes double clicks can be registered because of hubs. I've experienced this several times and have read other cases of it happening online.
thank you for this awesome test. i always hesitated to use my monitors usb hub for my input devices because of being worried that it might cause "problems", but now i have to rethink my cable management thanks to you!
Thank you for yet another great test! I haven't been able to use one of my USB hub inputs because I feared I would get additional input delay, so I had to plug a larger connector in which blocked one of the inputs. Now I can swap that and use all of the inputs :)
From the moment I saw the title, I was prepared to answer "no"....the reason:
USB is a polling communication, with the controller as master (why you can't just connect 2 PCs with a simple USB cable). While it's communicating with a device, that is the only time the device can give back data (unlike ethernet where packets can come and go independently).
So all actual communication is directly between USB controller and device. All that USB hubs do is identify to which port they have to make ELECTRICAL connection. If that would be delayed, USB controller would close the communication window and "disconnect" the device. The only delay a hub could do is the "repeating" of the electrical signal, but even that should be done shorter than the USB clock signal (I assume; even if longer, we are talking nano/micro-seconds, not milliseconds).
If I connect a 1 meter USB 3.0 extender to a 15 cm USB Hub 3.0, will I get any delay?
@@Tufet Delay, no. The only worry with extenders is signal degradation and wire resistance, but 1m should be negligible. I personally use 3m USB3 extender for my simracing setup, which goes to hub with KB, mouse and wheel. The wheel has it's own power, so it's a small load on the hub. But I would avoid charging a phone, even with USB3.
@@mateiberatco500 i will use for a headset with rgb and condesator microphone fifine k669, should it be ok? My USB hub has a type c power supply port as well
@@Tufet I don''t know how much power they need, but if you have usb3 up to the hub, it should be fine. USB2 hub/extender might be problematic, but I guess you should see that immediately when listening at high levels. My biggest worry is HDDs, as that could screw their content on writing.
Great video. I was curious about this for a while and it's really interesting that there's no disadvantage!
man you have alot of patience for all these tests...
Thanks for doing this video. Just thought of this question this morning when I was considering getting a USB switch to quickly swap my mouse and keyboard between my gaming computer and work laptop. Didn't want my work life to potentially slow my gamer reflexes!
This video answered and solved several of my problems:
1. Lack of USB ports
2. Cables too short
3. Bad cable management
Thank you!
nice to see youre still trying to improve the content with new tests like the mouse movement, good job
This is a cool video. Wondering if there’s an update. I noticed if results had been on a line graph showing normalized data (perhaps a simple % increase/decrease) then there are a number of places where you said “no change” that was actually a 5% change. Perhaps you could argue that it’s negligible, but it’s not zero.
Finally found someone covering a related topic! I was wondering if I introduce any sensor error or input latency while using an USB A adapter for USB c gaming mouse or keyboard. I pick them to have a 5 ms advantage, but if it's diminished by the adapter and not having a USB c port in my pc, it doesn't make sense of course. Thanks!
Great video! (I was planning the very details of my new PC setup when this question came up.)
I expected to only find the usual "guesswork / non technically founded" discussions, but your video really blew my mind.
Physically _measuring_ and then _veryfiying_ the results is a very scientific approach, I really appreciate this :).
I was curious about this topic and couldn't find extensive information on it. Thank you
Excellent! Will plug my mouse and keyboard into my monitors USB hub then :D Should make cable management easier with a sit/stand desk!
I genuinely think you are the best tech UA-camr on the planet. You have answered all the tough gaming questions with objective test results on so many topics I have needed answers from. We all really appreciate your hard work.
I never actually thought about it. I had a bit thought, about that distance to motherboard a bit longer when connecting mouse/keyboard to front panel instead of rear, but it is like 0.5 or less meters, what is like maybe 0.01ms delay.
So, thanks for such informative video!
This makes me wonder about raw PS/2 mouse vs USB->PS/2 converted mouse vs raw USB mouse. I've heard that PS/2 has its benefits because it sends interrupts to the CPU but I've never looked into if it actually works.
ua-cam.com/video/AWkvzycD5PE/v-deo.html look this video, short answer , ps/2 its not better.
You earned yourself a sub! This nice, good to understand and accurate testing is what i personally prefer to some "5 min production" videos. Keep the great work up!
Nice video, i was impressed with the way you came up to measure USB latency with high camera and rigging a LED to mouse.
I love how every website out there swears that usb hubs gives you a shit ton of input lag and you should never use one for gaming, but they provide no data.
I'm glad to see someone doing this sort of testing, to get rid of these dumb myths for good.
7:08 wow that deagle is larger than the m16
Fantastically in-depth, but to the point video. Thanks!
I use a mac and PC on the 1 monitor and unfortuntely most monitors only have 1 port to connect to its internal hub so i'll be confidently be using a USB 3.0 switching hub!
Man you are the frikin BEST!!!
I'm shocked and super happy with the outcome.
I have a powered USB 3.0 10 slot replicator. I was wondering if I should plug my Mouse/KB dongles directly into the PC or on the hub itself with other stuff. Good video.
Man, you Optimum Tech, and GN are honestly the best. This video is exactly what I was looking for. Well done!
I've run a USB 2.0 hub for gaming in the past in specific setups and can for sure say there is a difference in input latency.
I was running a SNES emulator while talking over voice and could measure the difference in input delay. I was streaming at the time as well so I had footage somewhere and talked about the issue with and without the USB hub. I ended up purchasing a powered USB 3.0 hub because of that problem.
in my case the problem was the chipset in my laptop+hub's in general
Interesting results! At 200fps (5ms per frame), I would have expected much less variance in the min/max results. Can you publish the raw results of each trial/test? I would like to see the distribution of the results of each test. As a follow up, I'd be very curious to see a comparison of 125/250/500Hz mice vs 1000Hz, in both min/avg/max as well as a latency distribution.
and maybe a PS/2 mouse? I've been told they are better in term of input lag, but that seems odd.
XseuguhX They work with interrupts instead of polling so they are in theory 0ms latency but it would be a nice thing to actually have a test :)
At 1000Hz you can expect at least 2ms variance even in the perfect world. He's only getting max 3. That's pretty close to baseline.
PS2 Mice ARE better for input lag since they are interrupt driven. This means that as soon as you move, and the mouse has data to send, the mouse notifies the CPU that the mouse should move right now. With USB polling, when you move the mouse, it sits there holding that data until the PC asks for it 1ms later.
But when we're talking about 33ms input lag vs 32, it's not worth the trouble really.
If we're looking at 125/250/500Hz refresh rates, it's the same, but the delay between polls is now 2ms/4ms/8ms.
Interesting. Then do you know why isn't there a PS/2 gaming mouse market as far as I can tell? "Pro gamers" buy hardware to get 500+ FPS on CSGo, they would buy a mouse that gives them 1ms better input lag I think.
Wouldn't it be easier to just look at the Desktop Cursor (or Game Cursor) to see when it starts moving?
-BY design, delay will be added. It might be just millisecond or two, but its still an existing delay. USB is not designed with daisychaining in mind, and its polling based system, which will literally always add delay, from your case, to that fancy USB hub on your monitor, keyboard, or standalone, and the more you chain, the longer the delay gets. However, an extension cable *shouldn't* add more delay, as there is no inbetween point where input and output has to be handled.
Now, you can decide if few milliseconds is worth worrying about or not, thats a whole other topic.
What? USB WAS designed for daisy chaining, this is why single controller can adress and poll up to 127 devices/hubs. Obviously it wasn't designed to daisy chain 50hub's and then mouse at the end, but no system was.
i do think about this, ive troubleshooting the hell out of it and think i got it under control, well done for shining a spotlight on this
usb can be messy, cpu, power settings, network, windows 10, nvidia, theres, chipset, hard ware etc etc a long list
This is surprisingly interesting. Keep up the good work and cheers from Finland.
Your test setup and experiments are beautiful! Thanks for sharing :)
Amazingly clear answer and explanation of a commonly asked question. Thank you, sir. Your diagrams were exceptionally clear... I just finished the video. Amazingly concise and the results are somewhat surprising as you say. Webcams now connected to monitor to eliminate cable run. Thanks again.
I am going to conect two monitors to the same pc at different roms for playing. I ordered this:
- 10 meters USB 2.0 active extensión usb cable (not 3.0 for costs).
- USB 3.0 USB hub that will be connected to the extension usb cable.
- 10 meters HDMI cable.
I'll conect a mouse, keyboard and Xbox controller. Just for gaming and multimedia. This video encouraged me to do this because I was worried about input lag. I hope that the USB hub conected to the USB 2.0 10m extension cable doesn't affect input lag. I think that it was the only missing part at your tests.
I'll try to make a comparation with you same metod but with a smarthponecamera at 960 fps (Galaxy S9).
I wish you could make some test with usb extanders/repeaters/usb-lan adapters with 15m or more for guys want to try to outsource pc from living room. You've done amazing job. Subscribed!
This is strange. I got a different result with my test using my MIDI keyboard and Audacity. I recorded myself tapping the drum pad while FL Studio played a sharp sound, and I looked at the waveform on the screen. I could measure the space between the peak of my tap to the beginning of the triggered sound. I found that connecting the keyboard through an unpowered USB 2.0 hub consistently added to the lag. I can't remember if it was 5 or 10 ms.
Bro I haven't watched the video yet but I just wanted to say that I literally just had this thought while on the toilet and immediately after that, I go on youtube and I find you with this video. God it's like you read people's minds or something. God bless you man you are doing life saving work.
Thank you so much!!! Thought about this so much as I need to switch computers but only have one set of mic, mouse and keyboard for my streams and recordings. This helps a lot mate!
Thanks I was wondering about hubs affecting latency. really happy to see someone tested it out
You deserve so much more subs and views. So much work done in those videos! :D
So you completely dropped the 'niche' phrase in your outro. Love the confidence. Keep it up :)
This definitely used to be more of an issue with USB devices stressing old CPUs more than their PCI or PS/2 equivalents. I used to avoid USB network devices and such like the plague, but now with the limited PCI-E lanes and multi-core on modern processors I do the opposite.
lol how times have changed.
I mean there is a difference, it's within 3 or 4 percent, but it does seem that there is a difference between the hypothetical fastest and slowest instances. It would be interesting to see if this type of test could be automated to allow for a much larger sample size.
Awesome channel by the way, nice to see an actual breakdown of this kind of thing rather than just seeing peoples opinions and hearsay
Okay but does that still apply with a wireless USB dongle mouse? Or would having less power to it affect it?
Great work bro answered all my questions about using a USB hub. Thanks.
Well constructed and clear analysis, with a cool accent too! :) Thank you for doing this. Subbed.
Hello! I was wondering if with the introduction of new technologies and possibilities you could do an updated video!
Is there a lag input between using a usb c hub with an hdmi connection to the monitor when compared to simply using the hdmi port in the pc?
that's a really surprising result and trying to not be biased against Hubs for responsiveness is going to be hard!
I couldn't find hard test results for this ANYWHERE! Thank you so much.
Great Work, that was exactly what i wanted to know because i start running out of USB Ports for all of my Wireless Logitech Devices!
In theory, just having more copper makes lag worse by some nanoseconds, but it's not even measurable outside margin of error preceptable
This video came in clutch. On the hunt for a usb hub for my laptop right now.
Came here three years later, extremely useful video, thanks!
Short answer no, long answer, for high bandwidth USB 3 ports no but usb 2.0 maybe like 1 or 2 ms which for some may be worth while moving shuffling your USBs around, I know I moved my mouse directly into my laptop after this and I have a lot plugged into my hub, sometimes even hdmi, ethernet and drives.
you could have chained a few hubs together and then get the delay per hub by dividing by the amount of hubs
would be crazy if difference was imperceptable
Finally some good testing set up. Really appreciate it.
Nice, it would be interesting to see the results on the moving mouse test. I would advice to, instead of "pulling the mouse really fast", to use some weight attached to the rope, for repeatibility sake...
very good video and this gave me the confidence to connect my mouse to usb 3.0 hub. Only improvement proposal would be testing longer cables. I have 3m long cable and I think that many are using also 5m long cables.
Really useful, concise, well presented, thanks a lot for making this.
The reason (I think) the stress and additional hubs cause no delay is that the OS or the mouse itself probably gets to tell the USB hubs to prioritize the mouse and not to delay its messages (delaying messages to combine them could reduce stress on the cpu for some devices).
mouse and keyboard take priority over every other action/imput that goes through the chipset. it is really only the length of the cable or the performance of the device that will create lag
The reason why you couldn't/shouldn't see any difference between back 2.0 and 3.0 ports is because those are usually operated by same physical controller. (USB3.0 controller can work ports as 2.0 no problem). Even on my old Z87 its just single 3.0 controller for all back USB ports.
And the reason why there was no difference with added controller load is polling. Devices are polled at given frequency and no device can "steal" all communication.
Miss you, but glad you're out for you and yours.
For sure, some unbranded OEM ones lag for me. Good test!
This is such a great video. Thanks for putting the time and effort to make this.
Should use a simple application that changes colour on receiving mouse input. Games can introduce latency on input, also as you saw framerate made it hard to collect data for you last test.
No input lag, but occasionally my mouse freezes up completely if using a powered hub, my controller locks up, and my blue snowball mic doesn't even receive power to turn itself on.
Thank you. Exactly what i needed and perfectly testet and explained. Now i can buy the cheap 2.0 USB switch instead of expensive 3.0 and dont worry if i introduce response lag.
42 people are still living under usb1.0 environment today.
Oh god thank you. I was going crazy trying to find a decent AM4 board with enough USB ports for my needs but it was either Gigabyte (which i dont like) or a very expensive crosshair. Now i just get a good USB Hub and a cheap board that fits my needs :D
Great video. Quality is amazing ! Thank you for your efforts
Great video! Also what game were you playing towards the end?
You should make a video about windows 10 services and 3rd party programs effect on input lag.
Malwarebytes is definitely one of them
If he woud go really into detail, how about testing the effect of different anti virus solutions (Windows Defender, Avast, Kaspersky, AVG, Antivir, BitDefender, Sophos, Panda in their free versions to keep the cost down) and some of their settings (like "game mode" and modules like behaviour scan)
Thanks a lot! I was searching for this, your video is perfect for me.
I'm not a PvP or a "pro" gamer at all... but this was SUPER helpful and awesome! Thanks for... Just... Thanks. Great work.
Thnak you for your time investiment. This video is great!
This is exactly i was looking for. Thanks a lot
I had assumed there'd be some notable difference. Surprising results.
Amazing, Thorough and well explained as always. (Thumbs up!)
Great video! A question: Can I assume all tests you made are comparable to playing with a USB Controller on the Nintendo Switch plugged in a USB hub?
I’m tryna figure out if the adapter will make smash unplayable 👀
One of the best tech videos on YT
Yeah i think every device gets its own transmission window within the 1ms USB frame, so far for interrupt endpoints like HID input. Then isochronous endpoints get a piece of the frame, like soundcards, webcams and capture cards, and finally bulk devices like USB disk volumes. The hubs can't buffer any data either if i'm not mistaken, the timeframes are host-defined.
It's pointless to test a USB 3.0 harddisk together with a USB 2.0 mouse though, because they're not even on the same bus, even if they're on the same physical port. You can just reconfigure the harddisk to USB2.0 for a potentially more interesting test, by using a USB 2.0 cable instead of the one that came with the disk, which however will still show nothing changing, i think. The common webcams like C920/C922 are USB2.0 and use a lot of the bandwidth, and that's still not a problem.
One might expect differences in behaviour between cheap mice, operating on a LowSpeed bus, and gaming mice, which operate on FullSpeed, because the LowSpeed devices consume more time out of a frame, but it makes them share and simply staggers the frames during which they talk, so say keyboard talks during one frame and mouse during another, this is why there's 8ms between messages. And yet, i don't expect this to actually be an issue still, because the bus must foresee enough time for interupt endpoints, though isochronous endpoints eat into the available time per frame overall.
Very informative video, and quite surprising results. Cheers!
Good to hear. Plz do a vid with inputlagtest gsync+ vsync on and capped framerate ingame because gsync.on+vsync off+capped framerate can cause partial tearing. So best experience is both on.
The partial tearing is caused by "micro FPS spikes" - the frame delivery is not on point.
To fix this you need to do one of the following things:
1. enable the high performance power plan inside windows, that might fix the issue
2. lower the FPS cap _(i.e. 140, instead of 142 on a 144Hz display)_
3. use RTSS instead of the FPS limiter built into the game, it increases the input delay by 1 frame but RTSS has more stable frame delivery
4. also enable v-sync _(i am not a fan of this since I paid 200$ extra for the g-sync support of the monitor ;-) )_
When you have a FreeSync or G-Sync monitor and cap your frame rate below the displays refresh rate, then v-sync on or off does not make a difference (however that is on my system where this did not cause tearing near the bottom of the screen) ua-cam.com/video/mVNRNOcLUuA/v-deo.html
Battle(non)sense thats not rly true I have capped all my games at 144fps (165hz). And playing with gsync and vsync on because there is a difference. Especially battlefield 1 does partial tearing at the bottom of the screen. There are threads that say this is the best way you configure gsync with both on and cap fps. They also did input lag tests which show no difference if vsync is on. But it fixes that partial.tearing. so I want a second lagtester to know if their results are true. Plz just try it in some games. Gsync was meant to work with vsync (nvcp). You can notice a difference in smoothness with it on in some games. If you do not cap fps and are in gsync range there is no partial tearing. It just occurs with capped fps with vsync off.so if you cap your fps vsync helps. Just wanted to ensure that there is no difference in input lag. (Blurbuster forums input lag thread gsync)
Battle(non)sense www.blurbusters.com/gsync/gsync101-input-lag/
Battle(non)sense so according to their results its better to use vsync on(nvcp on, ingame off) and ingame limiter than vsync off and rts (which doesnt rly help with partial tearing in bf1). So I just wanted a second good input lag tester if he (you) has the same results :-)