Legacy free MBs cost performance...

[Cradle22]

I don't know where it was, but some hardware site ran some benchmarks a while ago, and discovered, that an activated USB-port (even if there is no device attached) costs about 1-3 % performance (when compared to the performace when USB is completely disabled).

Since the only device with USB-plug which I generally use is the MS Itellimouse Explorer 3.0, and this device can also be plugged into PS/2, I have USB completely turned off... And this also gains an IRQ (which is always a nice thing to have)... along with disabled COM1 and COM2 I now have three more IRQs than most other people... Soundcard, graphics card and network (on-board) have all an IRQ for themselve... no sharing

And don't forget that connected USB-devices cost even more performance, since the chipset of the MB does not all device interaction without the help of the cpu...

Just my view on the feature-bursting MBs common right now...

[ppera2]

This belongs rather in PC Hardware section.
And subject is very badly formulated.
USB is much used today - printers, scanners, Digital Photo, etc.
Try to use such devices without USB port...

I know that printing via parallel port causes extremely big slowdown, even 90% with my Canon. With USB it's not case.

Serial port is outdated too.

All in all you should name this just 'USB costs performance'.
And we don't know what is situation with USB 2.

[Doom9]

your subject per se is wrong. http://www.anandtech.com/mb/showdoc.html?i=1636 shows that in no way is the AT7 beyond its competition.
Of course it's obvious that if you disable onboard devices the performance will increase.. that's one of the main reasons intel and co are driving forces behind the legacy free scheme.. get rid of as many devices that are no longer used as possible. On any board, if you start shutting down unused onboard components you'll gain in performance, usually only in small and unnoticeable percentages but the gain is definitely there and has nothing to do with legacy free or not. My old board with floppy, serial and parallel ports disabled (which has been my default config ever since I got it.. I've been preparing for the legacy free future for a long time making sure that whenever I needed a new external device it had an USB connector) runs faster than when I turn on these devices. As well as the AT7 runs faster if I shut down the RAID (which I did as atm I have no use for it), the firewire controller (no DV cam) and the USB2 controller (no USB2 device for the moment). But once you have any of those (and the iPod is damned tempting if it had better windows software) all you have to do is turn on an onboard component instead of having to buy extra cards (one for DV, another for USB2.. in the end it gets cheaper having it all onboard). Oh yeah.. show me how you can simply plug in a scanner (Acer 640) and printer (HP 940) and start printing/scanning within less than 20 seconds of pressing the power button without USB
:devil:

[theReal]

I always had USB disabled as long as I had no USB devices and I never understood why I would need a USB mouse or USB keyboard.
But now I have a webcam and a PDA-cradle both connected to USB and I'm very glad these devices are USB. Unfortunately, my new printer doesn't work perfectly with USB connection - on the USB port it wouldn't use 100% CPU like the archaic parallel port does (still have a parallel port scanner and it's a pain in the ass!)

[Cradle22]

OK, understood. But one thing for the printing: The 90 - 100 % usage if printing via parallel port issue has only came up when they startet selling more and more GDI-only printers (so that the device depends a lot on the driver to do the calculations). If I use my crappy HP DJ 600 I have 90 % CPU usage, but if I use the good old HP 895 Cxi, the usage is somewhere around 10 %...

But I don't know why (if it really is so, can't verify it) printing via USP takes up lesser power, since the calculations should be the same...

[theReal]

My printer isn't a GDI printer, it's a Brother HL-1440 with 6 MB RAM, but the transfer over parallel port still uses a lot of kernel CPU-time. Also, the parallel scanner uses 100% cpu throughout all of the scan process, even Winamp stops playing when I scan (at least in Win98 - I haven't checked this in Win2k)

[Uli]

Originally posted by Cradle22
... And this also gains an IRQ (which is always a nice thing to have)...

Wrong: PS/2 needs an IRQ too, so there is no more...
You turn USB off: + 1 IRQ, You turn PS/2 on - 1 IRQ (12)

greetz, Uli

[ppera2]

Problem with parallel port is that it uses slow 8-bit I/O operations, without buffering, and it occupies CPU totally.

[Lionel]

The problem with some USB devices not working properly is well known, and mainly appear in power hungry printers and scanners with old mobos. Try attaching the problematic device to a powered usb hub and see if this solves the problem.

[theReal]

Lionel,
as for my printer not working correctly with USB in win2k, it seems the problem is known to Brother, they recommend not to use the USB cable and they put the blame on Microsoft's USB implementation in Win2k. My board is not too old (Epox 8K7A+/AMD761) and other USB devices work without flaws (webcam and PDA-cradle)

btw. a laser printer should never take its power from the USB cable, I guess... ;)

[Lionel]

If it has 1 year or more, its old :D

The only implementation I like of usb in windows is the one from xp, the others seem like a quick and dirty fix.

[theReal]

If it has 1 year or more, its oldI know - I only meant the board isn't one of the first generation USB mainboards :)

[twistee]

I have a lexmark z33 and whenever i put it into a usb hub it always stops after printing ~1/9 of what i am printing (it says that the printer isnt connected...or something). But whenever i put it into the actual computer, it prints perfectly...it happens on different computers with one being win2k and the other winxp home. I now never use usb hubs...as they seem to always make things not work 'perfect'