Anyone running their USB port polling frequency at greater than 125hz?

[PaulG]

So I've been playing around with NLite today (Freeware for making optimized/ slipstreamed/driver installed XP install disks.)  One of the options under patches is changing the USB port polling frequency from the standard 125hz to 250 or 500hz.

Did a search and found Razor is making mice that are coming with drivers for 1000hz and Happs has this device:

http://www.happcontrols.com/joysticks/950800xx.htm

that's also polling at 1000hz.

Has anyone every tried upping their USB polling frequency to 500hz?  Seems like there could be benefits.  Also seems like there's some risk?  Don't wanna install the patch (Cause if it doesn't work, I'm not sure you can go back) and I don't wanna be making my NLite discs at 125hz if 500 would be better for trackballs and gaming in general.  Anybody ever experiment with this?

[tetsujin]

What are the "risks" you envision?

I guess the most obvious one is that you'd saturate the USB bus with low-speed communication with your input device - but for a game cab with few USB devices this isn't likely to be a problem, and the problem could be mitigated with a USB 2.0 hub with a transaction translator.

It also is possible, in theory at least, that a device may not perform as it's supposed to if you start polling it faster than it was tested - but it's hard to imagine you could do any damage to the device this way.

A bit of background info for those less familiar with this: USB input devices (keyboards, mice, joysticks, etc.) feed a description of themselves to the computer, which includes a request for how frequently the device would like to be polled for data.  The computer is at liberty to ignore this request and poll the device at more or less whatever speed it chooses.  (typical polling time for a USB input device is 10 ms or 100 Hz, fastest rate is 1ms or 1000 Hz.  IIRC the computer has no obligation to poll the device faster than 100Hz, no matter what the device requests.)  So the polling rate is, in the end, determined by the computer.