Originally posted by: Velk
Originally posted by: bignateyk<BR>Ive used my MX700 now for several months, and I love it. I am a hardcore gamer, and anyone who claims that a good USB or wireless mouse causes them lag is either just full of it and looking for excuses as to why the suck, or has a REALLY crappy computer otherwise.<BR>
<BR><BR>My previous PC was an athlon 2000 running on one of the early KT400 boards ( NS7 maybe ? I forget ) and it gave me a lot of problems with my MS USB mouse, mostly in the form of suddenly moving more slowly than before or pausing for brief fractions of a second. It drove me nuts, but worked fine using the PS/2 adapter. The same mouse on a newer nforce2 board had no problems at all. <BR><BR>Whether it was a driver problem or a hardware issue I don't know, but I can say you not having experienced lag using a USB mouse does not mean that *noone* experienes lag using a USB mouse.
Interesting data-point, thanks. It may well be just that Via cheaped out, and their USB ports are CPU driven (PIO mode - as I suspect is the case), and better chipsets (Intel, NVidia) are DMA-driven. That would correlate strongly with what I've seen, under nearly 100% application CPU usage, the mous skips around like a mofo.
Another reason could be issues with the software layer. USB mice have to interact with a user-mode process, running as a "hidden" window, called HIDSERV.EXE. Under very heavy CPU load, there is a noticable latency before Windows' can schedule the execution of that process, and if the data captured by the hardware driver (at interrupt level), has then to be processed by this user-mode intermediary process, before being sent to the other application that requested the data (a game for example), it could easily lead to high enough latencies that the internal hardware or device-driver axis counters could overflow, leading to skipping of the mouse. In fact this can happen even with a PS/2 mouse too (it's a software thing not a hardware thing in this case). That was one reason why I uninstalled the Logitech mouse drivers, and went back to the MS built-in ones - because the Logitech ones run a background user-mode process to process the data (and support things like button/keyboard re-mapping), and that adds noticable latency to the input stream - totally unacceptable for gaming purposes.
XP actually makes this "software latency" issue much worse as well, because of two things - mice are always seen as "emulated" devices under DirectX, and XP's "enhanced pointer precision" feature also necessarily has to process input-stream packets, and delay them by a few frames, in order to "smooth out" their movement. (It basically works almost like vsync, but for mouse inputs.)
In fact, I think that's exactly what
this article is talking about, when talking about "negative acceleration". The internal (to the hardware, or the hardware-level device-driver) counters are overflowing, because of the increased sensitivity of the device, as compared to how fast it is read (DPI vs polling Hz). Any temporary spike in CPU load, for a PIO-driven USB port can also cause this. Also, I'm not aware of any mouse input devices that implement true "negative acceleration" - only Trackpoint 3 and 4 series devices do.