Here's something I should have tested a long time ago but never did...PS/2 vs USB keyboards. In the past, I noticed plugging virtually any USB device into the computer besides the mouse, such as a PC gamepad, would alter mouse movement for the worse. Some gamepads only affected it a little, some to a bigger extent.
After unplugging my USB keyboard, I noticed a similar effect. Mouse movement got better. Most people know USB keyboards are polling based, while PS/2 keyboards are interrupt based, so it's an overall superior interface for keyboards anyway. You can't just buy a PS/2 adapter for your crappy USB keyboard and have it work though. Some keyboards send out both a PS/2 and USB signal, some send out only USB, so a PS/2 adapter does not work on all keyboards.
Here's the part where it gets real interesting. Most or all high end motherboards now only include a single transaction translator on USB hubs. This means that your mouse, keyboard, and any other device that you plug into your PC must ALL operate at either USB 1.x or USB 2 speed to not run into issues. For example, if you have a $10 keyboard that operates at USB 1.1, while your mouse connects with USB 2.0, how I understand it, you have two possible problems:
1) According to Cypress, data is not routed through the transaction translator (TT) at all unless you match up differing USB link speeds, and since all of these "high end" motherboards only include a single TT, you have now not only added an additional step onto the USB pipeline, both devices are also sharing a single TT which further degrades performance. I have no idea if that data is specific only to hardware that Cypress makes either. It could be that your motherboard in particular is routing all USB devices on the same hub through a single TT no matter what, even if they're all USB 2, or a bad case scenario for mice in other words.
http://www.cypress.com/?id=4&rID=54981
2) You can also have other issues where mixing USB 1.x and USB 2.0 just causes the PC to downgrade all ports to USB 1.x
So as you can see, there are multiple reasons for why USB keyboards seem to be a bad choice. Even if you don't run into one of the two problems above, just plugging the thing in is continuously polling your system while also typically being on the same hub as your mouse where both compete for resources.
For people who know nothing about keyboards, generally any keyboard guaranteed to work on PS/2 falls into one of two categories, $5 crap ones with only a PS/2 connector and no USB, and $100 mechnical keyboards that advertise things like infinite key rollover. You would probably want to make sure the keyboard includes the passive PS/2 adapter in the box, because like I said earlier, a keyboard can either send only a PS/2 signal, only a USB signal, or both.
I currently use CM Storm Quickfire with brown switches and 40A-R o-rings over PS/2 for the following reasons:
http://www.overclock.net/t/1525170/why-your-usb-keyboard-might-be-gimping-your-mouse-in-several-different-ways/100#post_23212418
After unplugging my USB keyboard, I noticed a similar effect. Mouse movement got better. Most people know USB keyboards are polling based, while PS/2 keyboards are interrupt based, so it's an overall superior interface for keyboards anyway. You can't just buy a PS/2 adapter for your crappy USB keyboard and have it work though. Some keyboards send out both a PS/2 and USB signal, some send out only USB, so a PS/2 adapter does not work on all keyboards.
Here's the part where it gets real interesting. Most or all high end motherboards now only include a single transaction translator on USB hubs. This means that your mouse, keyboard, and any other device that you plug into your PC must ALL operate at either USB 1.x or USB 2 speed to not run into issues. For example, if you have a $10 keyboard that operates at USB 1.1, while your mouse connects with USB 2.0, how I understand it, you have two possible problems:
1) According to Cypress, data is not routed through the transaction translator (TT) at all unless you match up differing USB link speeds, and since all of these "high end" motherboards only include a single TT, you have now not only added an additional step onto the USB pipeline, both devices are also sharing a single TT which further degrades performance. I have no idea if that data is specific only to hardware that Cypress makes either. It could be that your motherboard in particular is routing all USB devices on the same hub through a single TT no matter what, even if they're all USB 2, or a bad case scenario for mice in other words.
http://www.cypress.com/?id=4&rID=54981
2) You can also have other issues where mixing USB 1.x and USB 2.0 just causes the PC to downgrade all ports to USB 1.x
So as you can see, there are multiple reasons for why USB keyboards seem to be a bad choice. Even if you don't run into one of the two problems above, just plugging the thing in is continuously polling your system while also typically being on the same hub as your mouse where both compete for resources.
For people who know nothing about keyboards, generally any keyboard guaranteed to work on PS/2 falls into one of two categories, $5 crap ones with only a PS/2 connector and no USB, and $100 mechnical keyboards that advertise things like infinite key rollover. You would probably want to make sure the keyboard includes the passive PS/2 adapter in the box, because like I said earlier, a keyboard can either send only a PS/2 signal, only a USB signal, or both.
I currently use CM Storm Quickfire with brown switches and 40A-R o-rings over PS/2 for the following reasons:
http://www.overclock.net/t/1525170/why-your-usb-keyboard-might-be-gimping-your-mouse-in-several-different-ways/100#post_23212418