Holger Granholm,

06-Dec-99 18:29:00, Holger Granholm wrote to Andy Roberts
 HG> 12-04-99, Andy Roberts said to Holger Granholm:
          Subject: TVFS & USB

 HG>> One thing I noted on the CD was an USB program that wasn't usable
 HG>> for my new (still uninstalled) motherboard, an ASUS P5A-B with
 HG>> ALI BIOS.

 AR>> Humm...

 HG> Yes, look at the txt file of usbuhcd.zip.

--- Excerpt usbuhcd.txt ---
This is a modified version of IBM's OS/2 driver (Jan-06-99) for UHCI style 
USB host controllers. It should support mainboards equipped with an INTEL, 
VIA or ETEQ(?) chipset, as well as VIA USB addon cards.

USB chipsets from OPTI, SIS, NEC, CMD, ALI/ACER or AMD are not supported,
as their OHCI architecture is totally different.
---
1. Q: Will there be support for OHCI style USB chipsets soon?
   A: Most likely not. Use a cheap "VIA 83C572" USB addon card instead.
--- End Quote ---

For the extra cost of an addon card, you might as well have bought an Intel
and avoided a slew of incompatibility problems.

Don't get me wrong, I'm not in favor of supporting an over priced near
monopoly OEM.  But many years ago I was beta testing software on a
AMD386-DX40 that was very popular and had a timing problem in the cache.
After all the other testers and the author were satisfied, I still had many
problems.  We spent another month working out my problems before public
release.  When it was all done, that version became renowned as the most
solid version ever, even better than later versions.  I seriously doubt most
authors are willing to go to the extra effort and time to achieve near 100%
compatibility, but they almost always find a way to work with Intel.  Seeing
how that incident cost me several hundred lost hours, that I wasn't getting
paid for, after that I always insisted on Intel.  Sure that costs me more $
than AMD or most any other OEM, but my time is worth something and I'm
getting too old to put up with aggravation I can do without.  Granted that
was a long time ago, but I've followed the many reports of quirks with newer
non-Intel HW that may have had some little advantage in a specific area, but
almost always had an undesirable side effect, such as requiring a special
MoBo that could handle more power consumption, and a special power supply to
deliver that extra power.  Or gaining in 1 function, but losing performance
in another function.  AFAIK it was and still is always a trade-off or a
compromise to use non-Intel HW.

While almost always non-Intel HW will be considerably cheaper.  And in many
cases non-Intel HW will perform quite well.  IMO all too often, when you least
expect it, some little incompatibility shows up that doesn't exist with Intel.

In your case I don't know whether to blame USB or ALI or the driver author.
But considering the long list of other non-supported chipsets, it looks like
USB is far from standard.

About ASUS P2BL ATX:
 AR>> ......  My only complaint with that MoBo is that there is something
 AR>> strange about it's ATX power management, which the dealer and I tried
 AR>> to force to act like an AT.

 HG> I'm not taking any chances with ATX mobos.

I had several reasons for selecting the ASUS P2BL and ATX.  ASUS is near
top-of-the-line and claims support for OS/2.  And it has the Adaptec SCSI
controller built-in.  If I had to buy that Adaptec controller separately,
then it would have cost me as much as the MoBo.  So I figured I was getting
the MoBo free for the expense of the SCSI controller.  I had read that the
power management could be forced to act like an AT.  I like the positive
pressure air flow of the ATX design.  I like the massive heat-sink design of
the PII 450i.  And I needed another case, since this was for a new system.

I will admit that a year ago when I bought that ATX, I didn't know as much as
I do now about the undesirable side effects.  And while the matched case has
only 1 power switch, it does not seem work well with the AW37Pro.  The AW37Pro
works fine in the AT machine.

 HG> I've got a huge (2ft 1") AT tower for my work machine.

I've got 1 of those here and I love it.  I converted it to positive pressure.

 HG> It will now go through its third upgrade with said mobo and an K6-III
 HG> 400 - 450 MHz.

 HG> It came to a standstill because I didn't buy the memory at the
 HG> same time as I couldn't see that I had the time to do the upgrade
 HG> at that time.  Now we all know where the prices went. I can afford
 HG> to buy it now but if I wait a bit prices may go down again (or
 HG> up).

Seasonally November through January is not a good time for bargains.

About ATX:
 AR>> It apparently leaves the buss powered all the time, even with the
 AR>> power switch off, since the AW37Pro always complains after a cold
 AR>> boot but never complains after a warm re-boot.

 HG> Yes it apparently does. THAT is one thing I don't like with those
 HG> ATX boards. When you shut it off I'd expect it to really be dead.
 HG> Same thing when starting up, I don't fancy the use of two switches
 HG> before it starts

This case/PS only has 1 power switch.  And yes it is a real ATX PS with the
proper plugs, but no additional switch on it.  If I had known in advance, I
would have gladly put up with 2 switches rather than fight the other problems.

     Thanks and Good Luck,        Andy Roberts
                                  andy@shentel.net
--- Terminate 5.00/Pro*at 
347