Take the 2-minute tour ×
Super User is a question and answer site for computer enthusiasts and power users. It's 100% free, no registration required.

I'm looking for an authoritative answer, backed by data. A friend asked me the other day whether he would benefit from plugging his mouse into a USB 3.0 port instead of an available 2.0 port. I flippantly replied that it would make no difference at all. Certainly I'm not the only one to think that. See answers on:

Intuitively, I don't think data throughput should be an issue. Mice worked fine over a serial port, and those transfer at a maximum of 112.5 Kbps. USB 1.0 runs at 1.5Mbps (slow) or 12Mbps (fast). USB 2.0 can handle 480Mbps, and 3.0 can reach 5 Gbps.

But what about response speed? Are there published studies of response time between USB 2.0 and 3.0 ports?

My question is, was I right? Would it matter if I have a really fancy mouse? Can my friend safely blame his loss in League of Legends on his slow USB port?

share|improve this question
    
mice worked fine over a serial port, and the 'S' in USB is "serial". –  tedder42 Jul 23 at 1:24
    
Good question, I have no idea what the input lag is like between USB 1, 2, and 3. And would a USB 3, operating a USB 2 device have a worse or better input lag? –  Jonathon Wisnoski Jul 27 at 20:51

7 Answers 7

up vote 107 down vote accepted
  1. A mouse is a slow device (the old PS/2 standard was RS232C-based), so USB1 is more than adequate.
  2. USB3 ports have extra connectors for the high-speed transfers, but also have standard USB2 connectors for backwards compatibility.
  3. Unless your mouse has these connectors (and I cannot imagine that any mouse has) it will connect via USB2 in a USB3 connector.

So you are entirely correct: plugging a mouse into a USB3 connector confers no benefit whatever. If a mouse responds slowly, it is because something else is hogging the CPU to the detriment of the mouse driver.

share|improve this answer
3  
It is a good point that the mouse isn't even going to use USB 3 even if connected to such a port. I would just like to add to that by saying that it's likely to not even use USB 2; the USB controller in a mouse isn't likely to support anything beyond USB 1.1. –  Dolda2000 Jul 21 at 8:06
    
@Dolda2000 - I entirely agree, and the point was made in some of the earlier comments. I glossed over it in order to keep my answer as simple as possible, and the question was about USB2 vs USB3. –  AFH Jul 21 at 12:27
    
@Dolda2000: A device using USB1.1 will force the port to use USB1.1, which is Not A Good Thing (it has various quirks in addition to being sloooow); most mice I could get my hands on report as USB 2.0 devices. –  Piskvor Jul 28 at 13:46

Just because data rate of a mouse is very low and can be handled by USB 1.x does not mean it can't benefit from being plugging into a faster controller.

There are a number of factors in play here:

  • USB 3.0 controllers support backward compatibility by performing enumeration in a way that is compatible with USB 1.x and 2.0 devices, and exposing a logical EHCI controller as part of the register map.

  • The mouse will identify itself as a USB full-speed device regardless of the port and controller it is connected to. This will cause the mouse to appear logically connected to the EHCI controller (sometimes even OHCI/UHCI).

  • Having a logical EHCI controller does not mean that the device is attached to a USB 2.0 circuit. The connection between system and USB 3.0 host controller will be higher speed, probably multi-lane PCIe. It's also much less likely to go through a PCIe-PCI bridge, which causes buffering and slight additional latency.

  • Besides the host-side connection, the transistors driving the USB pins will also have a higher switching rate, and instead of passive components for noise filtering, the bus will use digital switches in the filter network, to allow it to be removed and not slow down USB SuperSpeed devices. (This was probably a bigger deal for the jump between USB 1.x and 2.0)
    The passive filter components present in the mouse should dominate. Still, rise and fall times could be a few nanoseconds faster.

  • The registers of the EHCI controller also are implemented on the same die as the USB 3.0 UHCI logic, so the clock rate and digital performance will be higher.

  • Of these effects, the PCIe differences have the greatest impact on performance, especially compared to a USB 2.0 host controller connected via a PCIe-PCI bridge.

In summary, a USB 3.0 host controller with a full-speed or high-speed USB 2.0 device is not the same circuit as the same device attached to a USB 2.0 host controller, and the timing will not be identical.

However, any latency improvement will be at most a couple microseconds. I suppose it's possible that with some probability this would cause the mouse input to be relayed to the server in time to be processed one frame earlier; the probability is extremely low (well under 1%), and anyone claiming it affects their game performance is confused.

In particular, the effects of network congestion are several orders of magnitude larger. Anyone trying to give themselves every technical advantage in the competition should focus their efforts there. QoS markings probably do determine the outcome of games at the highest levels of play. Choice of service provider almost certainly does.

share|improve this answer
    
This is wishful thinking, a mouse uses a fixed polling rate. The speed an individual bit takes to get from a to b does not change that rate whatsoever. –  JamesRyan Jul 20 at 22:34
3  
@JamesRyan: There's no wishful thinking here. Rate and latency are different. The delay until the next transmission timeslot is a USB buffering delay, it gets added to transmission delay itself along with buffering delays for the rest of the path to the CPU, to the NIC, and ultimately to the server. And my answer already clearly states that the delay improvements from using a USB 3.0 controller are very small –  Ben Voigt Jul 20 at 23:06
    
Imagine there are 2 queues of people over a long distance. At the front of each they are letting out 1 person every 20s and at the back a new person is arriving every 20s. In the first queue the person walks to the back of the queue, in the second queue they run. If it takes less than 20s to walk to the queue do you see that the person who runs doesn't get let out any quicker, they just wait in the queue for longer? –  JamesRyan Jul 21 at 10:00
1  
@otus: While it is true that on-chip communication links can run quite a bit faster than board-level links such as PCIe and PCI, in practice the USB 2 controller inside the chipset is still connected via ordinary old PCI (not PCIe), at ordinary PCI speeds, with a PCI-PCIe bridge. Despite the USB 3 controller being at a greater physical distance, it's connected via PCIe, usually with no bridge, and has lower latency. –  Ben Voigt Jul 21 at 14:43
1  
@Vinayek: In case you missed the discussion in the comments above, there are many other factors affecting the latency. You can use Windows Device Manager "View by Connection" to see if the USB controller is on a PCI or PCIe port. But that won't tell you whether timing is affected or by how much. –  Ben Voigt Aug 4 at 2:49

We need to take a look at the mouse's polling rate and from that we can have a better idea of how much data is being transmitted. If a mouse has a 100hz polling rate, it is sending data to the computer 100 times a second.

A standard mouse will send a 3 byte packet containing info on X/Y position information as well as button information. Considering that 3 bytes are transferred each cycle of the polling rate, you could have 300bps being transferred.

By default, the USB polling rate is 125hz, so by our logic, data being transmitted is 375 bytes per second.

Based on this, I don't think USB 3 is going to be any more beneficial than USB 2 or even 1.

share|improve this answer
4  
And that USB 3 might be better used on a device that needs the speed. –  cliff2310 Jul 15 at 22:32
3  
It should be noted you can change the USB polling rate. High-end mice run in the 500-1000hz range (2 or 1 millisecond response rates, verse 8 millisecond at 125hz). –  philipthegreat Jul 16 at 13:30
16  
@philipthegreat a 1kHz polling rate is still four times faster than the refresh period of even the best screens I know, sixteen times faster than FPS of most games on a good PC, and 100 times the frame rate at which the human eye is able to detect discontinuity in motion. And you're still transferring at only 0.5% the USB 1.0 low-speed bandwidth. –  Jan Dvorak Jul 16 at 16:37
1  
@JanDvorak +100 to that comment. –  Jason C Jul 17 at 17:32

Technically there are some cases where it would make a difference. Generally speaking even when this difference exists, it won't be noticeable to even a professional player.

USB 3.0 and USB 2.0 controllers on modern motherboards are implemented differently, primarily due to the data throughput USB 3.0 requires. In some chipsets you'd find the USB 2.0 appear as essentially PCI device on a supercommunications hub that also handles all the other slow devices, like hard drives, audio, ethernet, parallel and serial ports, etc. This existed in the southbridge chip, which was then connected to the northbridge through a media layer similar to a PCIexpress bus. The northbridge handled the high speed stuff, such as memory, graphics, PCIexpress etc.

This meant that a USB transaction involved several other transactions - PCI, then the media interface, then the interface to the CPU, before it was handled.

USB 3.0 was introduced at the same time that the northbridge/southbridge gave way to the PCH(Platform Controller Hub). Most of the northbridge duties were absorbed into the CPU itself - memory, PCIexpress, etc, while the remaining northbridge duties and southbridge duties went to the PCH. The PCH is essentially running off a PCI express slot.

The paths, however, are still very different. Even though USB 3.0 and USB 2.0 are integrated into the PCH, the USB 2.0 is still implemented as though it's a slow PCI controller device. There's no reason for Intel to redesign the silicon of a proven part, so it's integrated into the PCH the same way it was integrated into the southbridge, with all the attendant bottlenecks and additional latency that it had before.

However, the USB 3.0 is much closer to the CPU. While this is meant primarily to account for the increased throughput, it also affects latency - there are fewer transactions involved in getting a USB 3.0 transaction to memory, or to the CPU, and the interrupts may be triggered more quickly.

However the difference between USB 2.0 latency and USB 3.0 latency would be measured in nanoseconds. Certainly not noticeable by even the best twitch gamers today. It exists, but it's practically useless.

share|improve this answer
2  
Probably not even measurable by objective timings as latency differences are several orders of magnitude lower than e.g. device polling intervals, kernel scheduler time slices, UI event queue processing times, etc. The OP's friend is simply defensively looking for something to blame his unfortunate loss on. –  Jason C Jul 17 at 19:35

Mouse is a so slow device in terms of data transfer that USB speed limit is not a limiting factor.

Communication latency also should not be a factor, a properly built (any generation) USB device should respond far faster than human reflex speed, so you should not experience any lag due the technology of the port.

Of course there are many other factor if you experience mouse lag: system may be busy, driver may be not working as intended, software can be stuck waiting for other events (most of times, it is a network-related event problem), mouse itself can be broke, and finally the port hardware may be faulty - so trying another mouse/port is a good start.

share|improve this answer

While I do not believe that on most conventional mouse models you would get any advantage whatsoever, you are likely trying to determine that even if there was any difference, it would have to be noticable by the user?

In that case, it is simple enough to test the case in a mouse-performance demanding game or application, by plugging it into both ports. I have tried this myself, and found no difference that I could notice, which was sufficient for me (test results beat theory for personal use scenarios, as in some cases even a psychological difference could help you perform better - in which case it would still be better for you).

share|improve this answer

It's worth noting that if the mouse is USB2, you should prefer USB2.

Using a USB2 device on a USB3 bus will bring the bus speed down for all devices that share that bus (so your USB3 hard drive would run slowly as soon as you plug the mouse into the free USB3 slot).

share|improve this answer
14  
That's wrong, actually each USB 3.0 port is connected to both an USB 3.0 and 2.0 bus, which are pretty much independent. Looking for some reference, I found this answer. –  Jens Erat Jul 16 at 8:45
2  
That principle only applied to using low-speed (1.5Mbps) devices on full-speed (12 MBps) USB 1.1. In 1.1, low-speed packets are transmitted at 1.5 Mbps through the entire bus hierarchy and starve the bus for time to transmit faster packets. From 2.0 on, low-speed and full-speed packets are transmitted at high-speed (480 Mbps) once they hit an intervening high-speed hub. For 3.0, devices negotiate a connection first over the 2.0 interface as a baseline and then enable the 3.0 interface if it exists. –  Kevin Thibedeau Jul 16 at 13:24

protected by random Jul 18 at 3:34

Thank you for your interest in this question. Because it has attracted low-quality answers, posting an answer now requires 10 reputation on this site.

Would you like to answer one of these unanswered questions instead?

Not the answer you're looking for? Browse other questions tagged or ask your own question.