d93### [at] efdlthse (Michael Lundahl) wrote:

>In article <357370fc.8075051@news.povray.org>,
>	new### [at] DESPAMpovrayorg (povray.org admin team) writes:
>
>>I would often see Win95 crash
>>several times a day. In the six months I've been running NT, it has crashed
>>perhaps five times. Occasionally (maybe once every few days) it gets into a
>>funny state and I'll choose to reboot it, but hard crashes are very unusual.
>
>You need to reboot Windows machines!? And they crash?!?!

Unfortunately, yes .. with all Microsoft's efforts, they have yet to reach the
level of reliability of, say, the FreeBSD machine that runs this news server,
or kernels such as QNX or even [ugh] OS-9 (which is what I do a lot of work in
- uptimes of several years are not unknown).

Nevertheless, they are on their way ; the NT kernel design is very nice. In
general, when I get problems with NT, it's because one of the higher-level
components gets screwy (in particular, the GUI). Unlike Unix, where you can
just CTL-ALT-BKSP exit from and restart XFree86 if it goes ga-ga, with NT there
is no such option (though perhaps there should be).

Post a reply to this message

Andrew M <tot### [at] inamecom> wrote:

>How about the fact that it costs ~200 dollars, OEM.

I was going to mention that myself. This is the primary reason why I had to
choose not to make the new POVWIN rendering engine an NT-only service. Many of
our users simply can't afford the high cost of NT.

Post a reply to this message

>Also, the STOP's are a pain. Whenever an app crashes badly, NT generates a
>STOP; just what it sounds like, a STOP halts the operating system and forces
>you to restart.

An app (presuming it's a user-level app) cannot (to my knowledge) induce a stop
(i.e. bugcheck). Only a faulty kernel-mode component (i.e. ring 0) can do this.


This generally means a bug in a device driver. It is possible, of course, for
an app to induce a bugcheck (AKA BSOD, the 'Blue Screen Of Death') if it is
communicating with a device driver in such a way as to flush the bug out. Also,
some apps come with their own kernel-mode drivers (though this is uncommon).
However, the same principle applies there - the bug is in the device driver.

Post a reply to this message

Yes, but since NT 4.0 moved a lot of the GUI stuff into ring 0, there
are potentials for applications to do this.

There was the one where a 4-line Windows App could cause a BSOD 100% of
the time. Then the Internet Explorer/F00F bug combo was also a good
one...



povray.org admin team wrote:
> 
> >Also, the STOP's are a pain. Whenever an app crashes badly, NT generates a
> >STOP; just what it sounds like, a STOP halts the operating system and forces
> >you to restart.
> 
> An app (presuming it's a user-level app) cannot (to my knowledge) induce a stop
> (i.e. bugcheck). Only a faulty kernel-mode component (i.e. ring 0) can do this.
> 
> This generally means a bug in a device driver. It is possible, of course, for
> an app to induce a bugcheck (AKA BSOD, the 'Blue Screen Of Death') if it is
> communicating with a device driver in such a way as to flush the bug out. Also,
> some apps come with their own kernel-mode drivers (though this is uncommon).
> However, the same principle applies there - the bug is in the device driver.

Post a reply to this message

povray.org admin team wrote:
> 
> Andrew M <tot### [at] inamecom> wrote:
> 
> >How about the fact that it costs ~200 dollars, OEM.
> 
> I was going to mention that myself. This is the primary reason why I had to
> choose not to make the new POVWIN rendering engine an NT-only service. Many of
> our users simply can't afford the high cost of NT.

You mean so there is no windows 95/98 verson?  What about ports to other
OSes? $200 buys a hefty amount of processing power, like it could buy a
moniterless K6 200, maybe even a 233 or 266, for a renderfarm. Or it
could upgrade me to K6 300 plus 30 bucks to spare.

Andrew McDavid

Post a reply to this message

povray.org admin team wrote:
> 
> An app (presuming it's a user-level app) cannot (to my knowledge) induce a stop
> (i.e. bugcheck). Only a faulty kernel-mode component (i.e. ring 0) can do this.
> 
> This generally means a bug in a device driver. It is possible, of course, for
> an app to induce a bugcheck (AKA BSOD, the 'Blue Screen Of Death') if it is
> communicating with a device driver in such a way as to flush the bug out. Also,
> some apps come with their own kernel-mode drivers (though this is uncommon).
> However, the same principle applies there - the bug is in the device driver.

Just as another data point...

I've been running NT (3.51 and 4.0) for about four years, on several
different boxes, using it for intensive development work as well as many
other applications (including POV-Ray) (not counting editors and web
browsers, the program I run most often is probably EGCS). In all that
time, I've had exactly three system crashes (and one of those was under
an early beta of NT4, so I don't think it's fair to count it), plus two
occasions when the OS was malfunctioning badly enough that I rebooted to
fix it. My record for contnuous up time is about two months. (The usual
reason for rebooting is badly written installers that insist on a
restart.)

So don't take those "blue screen of death" tales too seriously. Yeah, it
happens, but an average of one system crash per year is good enough for
me.

-- 
Ross Smith ..................................... Wellington, New Zealand
<mailto:r-s### [at] ihugconz> ........ <http://crash.ihug.co.nz/~r-smith/>
   "Remember when we told you there was no future? Well, this is it."
                                                         -- Blank Reg

Post a reply to this message

Andrew McDavid <tot### [at] inamecom> wrote:

>povray.org admin team wrote:
>> 
>> Andrew M <tot### [at] inamecom> wrote:
>> 
>> >How about the fact that it costs ~200 dollars, OEM.
>> 
>> I was going to mention that myself. This is the primary reason why I had to
>> choose not to make the new POVWIN rendering engine an NT-only service. Many of
>> our users simply can't afford the high cost of NT.
>
>You mean so there is no windows 95/98 verson?  What about ports to other
>OSes? $200 buys a hefty amount of processing power, like it could buy a
>moniterless K6 200, maybe even a 233 or 266, for a renderfarm. Or it
>could upgrade me to K6 300 plus 30 bucks to spare.

Where was it stated that there is no 95/98 version ?

Post a reply to this message

povray.org admin team wrote:
> 
> Andrew McDavid <tot### [at] inamecom> wrote:
> 
> >povray.org admin team wrote:
> >>
> >> Andrew M <tot### [at] inamecom> wrote:
> >>
> >> >How about the fact that it costs ~200 dollars, OEM.
> >>
> >> I was going to mention that myself. This is the primary reason why I had to
> >> choose not to make the new POVWIN rendering engine an NT-only service. Many of
> >> our users simply can't afford the high cost of NT.
> >
> >You mean so there is no windows 95/98 verson?  What about ports to other
> >OSes? $200 buys a hefty amount of processing power, like it could buy a
> >moniterless K6 200, maybe even a 233 or 266, for a renderfarm. Or it
> >could upgrade me to K6 300 plus 30 bucks to spare.
> 
> Where was it stated that there is no 95/98 version ?
	You said you decided _not_ to make it an NT service.  I was just
wondering if that meant, if you had decided otherwise, there wouldn't
have been a 95/98 verson?

Andrew M

Post a reply to this message

>>Also, the STOP's are a pain. Whenever an app crashes badly, NT generates a
>>STOP; just what it sounds like, a STOP halts the operating system and
forces
>>you to restart.

>Not the OS my friend, the _processor_ ;)

If it halts the processor. then why does the option "Automatically reboot"
in System Properties under "When a STOP message occurs" work?

It must be... uh... the FLOPPY DRIVE that automatically reboots it! Yeah! ;)

-aardvarko

Michael Lundahl wrote in message <6l37to$a37$4@oz.aussie.org>...

Post a reply to this message

>An app (presuming it's a user-level app) cannot (to my knowledge) induce a
stop
>(i.e. bugcheck). Only a faulty kernel-mode component (i.e. ring 0) can do
this.
>
>
>This generally means a bug in a device driver. It is possible, of course,
for
>an app to induce a bugcheck (AKA BSOD, the 'Blue Screen Of Death') if it is
>communicating with a device driver in such a way as to flush the bug out.
Also,
>some apps come with their own kernel-mode drivers (though this is
uncommon).
>However, the same principle applies there - the bug is in the device
driver.

DeBabelizer does sometimes when it hits a faulty TGA file. Beats me. :)

Also Total Annihilation (an awesome RTS game) STOP's the system if you let
it install DX5 and then try to run it; it's pretty stupid to install a
Windows 95 version of DirectX. :|

-rdv

povray.org admin team wrote in message
<3577695e.43768355@news.povray.org>...
>>Also, the STOP's are a pain. Whenever an app crashes badly, NT generates a
>>STOP; just what it sounds like, a STOP halts the operating system and
forces
>>you to restart.
>

Post a reply to this message