Re: Drivers, Upgrades and SW design (was Re: Dedicated Z-wave sites?)

[Marc_F_Hult <MFHult@xxxxxxxxxxxxxxxxxxxxx>]

On Tue, 2 Jan 2007 19:50:07 -0500, "Robert Green"
<ROBERT_GREEN1963@xxxxxxxxx> wrote in message
<o8idncuA36h9nwbYnZ2dnUVZ_vmqnZ2d@xxxxxxx>:

>I don't know how a discussion of driver reliability morphed into a belief
>that I want to run games on my HA PC, but that's not the case.  For
>review:
>I noted an article on NOOKS, that talked about 85% of crashes in PC's
>being the result of bad drivers.
>
>I mentioned games as an application that caused video drivers, in
>particular, to be very heavily stressed and that caused bugs to be found
>and new drivers to be released, sometimes as often as every week when a
>video board is new.  I had hoped that would have been a cue that this
>discussion was about driver issues, and not about CQ.

And if you don't run the games that 'stress' the video drivers that cause
the crashes (HA apps don't or shouldn't), where is the problem? Don't run
'video-compute-intensive' apps on production servers, scientific
instrumentation, or HA controllers and the 'problem' disappears. Stick to a
$50, fanless, plain-Jane video card and the problems you cite disappear.

>The same problems facing video drivers happen to be true for motherboard
>BIOS's when a new board is launched.  As the boards leave beta testing and
>reach the real world, loads of problems begin to surface, many of which
>have to be fixed by rewriting the BIOS code.

Reality check. You wrote that new MB's have bios updates as often as
weekly.  I have at this moment 7 PC's running. The only one that has *ever*
had a bios update is an 11-year-old Thinkpad sold with W95/98 that needed a
new BIOS to run W2K (which wasn't successful and IBM lost a class action
suit over).

(Yes, I have had other PC's for which I did update BIOSes on  -- for
example, the first dual 1ghz Pentium III Tyan MB was a nightmare I won't
forget. I bought it before the 1ghz chips were shipping. But I also bought
a 1975 Pinto new ...  nuff said ! ;-)

As I have written before, my experience in running CyberHouse HA SW
24x7x365x7 years is that IT DID NOT crash, or freeze, or hang, or have any
of the problems that you seem to want it to have. I can't change my
experience. I have also run scientific instrumentation that I built myself
to run on PC when the MS OS _was_ notoriously unstable and succeeded even
that problematic environment by *active* problem *avoidance* .

>When a manufacturer changes drivers or firmware frequently they often
>break something that was previously unbroken.

Does the manufacturer sneak into your basement in the dead of night and
change out your video bios? Or do you cause the problem by applying every
patch and update that comes along, needed or not? And what you write is
universally true of _all_ computer code, firmware or not. It shouldn't
paralyze one.

>That means there's a possibility that other programs depending on previous
>versions of that driver no longer work as expected.  I used the example of
>video drivers used by gamers because those drivers get worked so hard by
>so many people in such a short time span that errors that might languish
>undetected for years in other types of drivers are quickly exposed.

ROTFL. And "there's a possibility" that a problem that does not happen is
not a problem ;-)

>From what the professor who developed NOOKS wrote, it's the driver bugs
>that only appear under certain circumstance and combinations of factors
>that are the most insidious.  Truly bad drivers are shot down right away
>and fixed. Intermittent bugs, especially related to specific hardware
>interactions, are very hard to find.

And there is no need to find the bugs if one doesn't use the particular,
troublesome hardware involved because the bugs don't actually exist in
one's hardware instantiation except as a hypothetical.

I found my spread sheet for the configuration of the instrument backplane
80386-16 that I referenced in my previous post. It all came back in a flood
of nostalgia ;-) :  multiple DAQ and PIO cards, four serial, two parallel,
bus-mouse, IEEE-488, HP-IL, mono and SEGA video cards and more -- it took
me forever to get all the interrupts and I/O addresses and drivers
configured, working and stable. But I did. Do you think I loaded a new
driver or replaced the BIOS chip jist because a newer version was available
that fixed a problem with hardware that I didn't have?

... Marc
Marc_F_Hult
www.ECOntrol.org