Pentium4 and older OpenServer5

0 views
Skip to first unread message

Don Yakubowski

unread,
Jan 8, 2003, 12:43:44 PM1/8/03
to
I have noted the reqt for P4 systems to run 5.0.6 with 648a supp. Anyone
know what can/would happen if a drive with 5.0.4 on it was put in a P4
system?
Could one even load a non-5.0.6 OS in a P4 system?

Don Yakubowski
Tri-Comp Systems Ltd.


Stuart J. Browne

unread,
Jan 8, 2003, 5:15:05 PM1/8/03
to
> I have noted the reqt for P4 systems to run 5.0.6 with 648a supp. Anyone
> know what can/would happen if a drive with 5.0.4 on it was put in a P4
> system?
> Could one even load a non-5.0.6 OS in a P4 system?

I beleive it would work, but you'd be in danger of damanging the P4. The P4
has thermal throttling and a number of other neat features, to try and keep
the processor working as efficiently as possible.

If you overworked the CPU, had a cooling fan die, or something else equally
unexpected, it's entirely possible you could burn your CPU out.

Bela, a quick question with regards to the termal throttling. Does the CPU
make a request of the OS as to whether it can be turned on/off, or does it
just do it with the OS-be-dammned?

bkx


Bill Vermillion

unread,
Jan 9, 2003, 12:27:16 AM1/9/03
to
In article <avi7pb$2di$1...@perki.connect.com.au>,

Stuart J. Browne <stu...@promed.com.au> wrote:
>> I have noted the reqt for P4 systems to run 5.0.6 with 648a supp. Anyone
>> know what can/would happen if a drive with 5.0.4 on it was put in a P4
>> system?
>> Could one even load a non-5.0.6 OS in a P4 system?

>I beleive it would work, but you'd be in danger of damanging the
>P4. The P4 has thermal throttling and a number of other neat
>features, to try and keep the processor working as efficiently
>as possible.

>If you overworked the CPU, had a cooling fan die, or something
>else equally unexpected, it's entirely possible you could burn
>your CPU out.

That's on the Athlon before the mfrs kludged a board fix in, which
is nasty for Unix systems. On those if the fan fails and the CPU
starts heating the power supply is turned off.

And you know how bad that cam be pretty bad.

>Bela, a quick question with regards to the termal throttling.
>Does the CPU make a request of the OS as to whether it can be
>turned on/off, or does it just do it with the OS-be-dammned?

Remember Jeff's posting where he showed the times just slowing
down further and further? That was a few months ago. The CPU
clocking is shown at boot up and he rebooted a few times and
watched the CPU clock down to less tha 1/4 of it's rated speed.
It throttles itself down just to protect against things like
this.


--
Bill Vermillion - bv @ wjv . com

Stephen M. Dunn

unread,
Jan 9, 2003, 8:26:13 PM1/9/03
to
In article <avi7pb$2di$1...@perki.connect.com.au> "Stuart J. Browne" <stu...@promed.com.au> writes:
$> I have noted the reqt for P4 systems to run 5.0.6 with 648a supp. Anyone
$> know what can/would happen if a drive with 5.0.4 on it was put in a P4
$> system?
$> Could one even load a non-5.0.6 OS in a P4 system?
$
$I beleive it would work, but you'd be in danger of damanging the P4. The P4
$has thermal throttling and a number of other neat features, to try and keep
$the processor working as efficiently as possible.
$
$If you overworked the CPU, had a cooling fan die, or something else equally
$unexpected, it's entirely possible you could burn your CPU out.

I'm not so sure that's true. It's been a while since I looked into
this, but I think the thermal throttling in the P4 functions even if
the OS knows nothing about it, and that it's pretty difficult to
fry a P4 this way. The CPU simply sllllooooooowwwwwwwwwwwwssssssssssssssss
down to reduce its heat dissipation. Again, from memory, the P3
takes another effective, but much more disruptive, approach: it
shuts itself off if it's about to set itself on fire.

There is, or at least used to be, a pretty nifty video of a
test of several CPUs' overheating behaviours on tomshardware.com.

So I think you can run pre-5.0.6 on a P4; however, don't be
surprised if the CPU responds by slowing itself way down over time,
to the point where you may find a higher-end P3 outperforms it.

Remember, it's been a while since I looked into this so take
my post with a grain of salt and do some research yourself. I'm
not responsible for any spontaneous CPU combustion that may result
from you running your choice of non-P4-aware OS on a P4 - you are!
--
Stephen M. Dunn <ste...@stevedunn.ca>
>>>----------------> http://www.stevedunn.ca/ <----------------<<<
------------------------------------------------------------------
Say hi to my cat -- http://www.stevedunn.ca/photos/toby/

Jim Sullivan

unread,
Jan 9, 2003, 12:18:17 PM1/9/03
to

"Stuart J. Browne" <stu...@promed.com.au> wrote in message
news:avi7pb$2di$1...@perki.connect.com.au...

> > I have noted the reqt for P4 systems to run 5.0.6 with 648a supp. Anyone
> > know what can/would happen if a drive with 5.0.4 on it was put in a P4
> > system?
> > Could one even load a non-5.0.6 OS in a P4 system?
>
> I beleive it would work, but you'd be in danger of damanging the P4. The
P4
> has thermal throttling and a number of other neat features, to try and
keep
> the processor working as efficiently as possible.
>
> If you overworked the CPU, had a cooling fan die, or something else
equally
> unexpected, it's entirely possible you could burn your CPU out.

If you want to see what happens when the fan dies, check out:
http://www6.tomshardware.com/cpu/20010917/index.html

WHen I first saw this, I suddenly realized that the computer in my office
could be a major fire hazard.


--
Jim Sullivan
seattl...@comcast.net
Don't plant your bad days, they turn into bad weeks,
and then bad months, and before you know it,
you've had a bad year!


Rainer Zocholl

unread,
Jan 10, 2003, 4:38:00 PM1/10/03
to
(Jim Sullivan) 09.01.03 in /comp/unix/sco/misc:


>"Stuart J. Browne" <stu...@promed.com.au> wrote in message
>news:avi7pb$2di$1...@perki.connect.com.au...
>>> I have noted the reqt for P4 systems to run 5.0.6 with 648a supp.
>>> Anyone know what can/would happen if a drive with 5.0.4 on it was
>>> put in a P4 system?
>>> Could one even load a non-5.0.6 OS in a P4 system?
>>
>> I beleive it would work, but you'd be in danger of damanging the P4.
>> The P4 has thermal throttling and a number of other neat features, to try
>> and keep the processor working as efficiently as possible.
>>
>> If you overworked the CPU, had a cooling fan die, or something
>> else equally unexpected, it's entirely possible you could burn
>> your CPU out.

>If you want to see what happens when the fan dies, check out:
>http://www6.tomshardware.com/cpu/20010917/index.html

>WHen I first saw this, I suddenly realized that the computer in my
>office could be a major fire hazard.

The problem only arose on AMD CPUs.

All Intel CPUs survived the "lost cooler test".
All AMD CPUs died, more or less effectfull ,-)


Rainer Zocholl

unread,
Jan 10, 2003, 4:44:00 PM1/10/03
to
(Bill Vermillion) 09.01.03 in /comp/unix/sco/misc:

>In article <avi7pb$2di$1...@perki.connect.com.au>,
>Stuart J. Browne <stu...@promed.com.au> wrote:

>>If you overworked the CPU, had a cooling fan die, or something
>>else equally unexpected, it's entirely possible you could burn
>>your CPU out.

>That's on the Athlon before the mfrs kludged a board fix in, which
>is nasty for Unix systems. On those if the fan fails and the CPU
>starts heating the power supply is turned off.

>And you know how bad that cam be pretty bad.

Ouch.

At least the data in RAM will not be lost and the
file systems stays consistent... ;-)

I assume the athlon will start to make errors when getting
hotter and hotter. Why is it bad to stop it the "hard way"
allocating wrong inodes etc.?
A fschk would never be abel to fix such -written- missallocations.
And a few moments later the CPU may run into a senseless loop
because address calculation will not work any more, and now
overheat faster.
Result:
unrepairable broken filesystem and a burned out CPU.

Better use an Intel in a server
or a Siemens termal management unit: a second (smal) CPU only
to check and control clock and temperature...

Bill Vermillion

unread,
Jan 10, 2003, 10:25:59 PM1/10/03
to
In article <8d$nq-A$H...@zocki.toppoint.de>,

Rainer Zocholl <UseNet-Pos...@zocki.toppoint.de> wrote:
>(Bill Vermillion) 09.01.03 in /comp/unix/sco/misc:

>>In article <avi7pb$2di$1...@perki.connect.com.au>,
>>Stuart J. Browne <stu...@promed.com.au> wrote:

>>>If you overworked the CPU, had a cooling fan die, or something
>>>else equally unexpected, it's entirely possible you could burn
>>>your CPU out.
>
>>That's on the Athlon before the mfrs kludged a board fix in, which
>>is nasty for Unix systems. On those if the fan fails and the CPU
>>starts heating the power supply is turned off.

>>And you know how bad that cam be pretty bad.

>Ouch.

>At least the data in RAM will not be lost and the
>file systems stays consistent... ;-)

>I assume the athlon will start to make errors when getting
>hotter and hotter. Why is it bad to stop it the "hard way"
>allocating wrong inodes etc.?

If the fan fails your Athlon is toast in about 1 second. Time for
new CPU. That's why the turn-the-power-supply-off fix was
implemented.

Well you typically don't want a multi-user system to just stop dead
in it's tracks. Of course if your turn it back on it should turn
right back off. So you call the tech support and find the fan has
failed. On a system with more than a handful of users the price
differenece between an Athlon that did that, and a P4 that just
throttles itself down actually makes the Athlon cost more - when
you include lost production time.

See Tomshardware for details and gory pictures.

>A fschk would never be abel to fix such -written- missallocations.
>And a few moments later the CPU may run into a senseless loop
>because address calculation will not work any more, and now
>overheat faster.
>Result:
>unrepairable broken filesystem and a burned out CPU.

The CPU will burn out in about 1 second with a failed fan.

>Better use an Intel in a server
>or a Siemens termal management unit: a second (smal) CPU only
>to check and control clock and temperature...

Yup.

Rainer Zocholl

unread,
Jan 11, 2003, 6:43:00 AM1/11/03
to
(Bill Vermillion) 11.01.03 in /comp/unix/sco/misc:

>In article <8d$nq-A$H...@zocki.toppoint.de>,
>Rainer Zocholl <UseNet-Pos...@zocki.toppoint.de> wrote:
>>(Bill Vermillion) 09.01.03 in /comp/unix/sco/misc:

>>>In article <avi7pb$2di$1...@perki.connect.com.au>,
>>>Stuart J. Browne <stu...@promed.com.au> wrote:

>>>>If you overworked the CPU, had a cooling fan die, or something
>>>>else equally unexpected, it's entirely possible you could burn
>>>>your CPU out.
>>
>>>That's on the Athlon before the mfrs kludged a board fix in, which
>>>is nasty for Unix systems. On those if the fan fails and the CPU
>>>starts heating the power supply is turned off.

>>>And you know how bad that cam be pretty bad.

>>I assume the athlon will start to make errors when getting


>>hotter and hotter. Why is it bad to stop it the "hard way"
>>allocating wrong inodes etc.?

>If the fan fails your Athlon is toast in about 1 second.

Yes. Time enough to generate bad data.
@2Ghz 2 billion CPU clocks.

>Time for new CPU.

Of cause.

>That's why the turn-the-power-supply-off fix was implemented.

That would only decrease the amount of CPUs AMDs can sell.. ;-)

>Well you typically don't want a multi-user system to just stop dead
>in it's tracks.

Yes.
In both cases the CPU is "dead" for the OS.

>Of course if your turn it back on it should turn
>right back off.
>So you call the tech support and find the fan has failed.

Hopefully...

>On a system with more than a handful of users the price
>differenece between an Athlon that did that, and a P4 that just
>throttles itself down actually makes the Athlon cost more - when
>you include lost production time.

Yes, no discussion about needed.
throtteling the cpu down to 600Hz is far better than letting
burn out or stop it the hard way.
As i said: The dead of a CPU is not a "digital binary" go/nogo.
It dies with first generating random numbers.
It's more like go/fail seldom/fail sometimes/fail often/dead.
1sec are 2 Billion CPU clocks and much more operations.

>See Tomshardware for details and gory pictures.

Yes, real nice. But those "1sec" ocures when the cooler
was removed on the running CPU, IIRC. Not just blocking the fan.
And: The AMDs "dirty fan hack" would not detect that at all!
Because AMDs needs a big cooler one MUST open the box
after each transportation!

>>Better use an Intel in a server
>>or a Siemens termal management unit: a second (smal) CPU only
>>to check and control clock and temperature...

BOT:
Does an unpatched 5.0.6 OSR "burnout" (break, fire) the P4-CPU
or does it trigger only the overheat throtteling so it "looks like"
a broken CPU (600Hz instead of 2000MHz is like "stopped")
In the first article that did not become clear.

If someone made the mistake and bought an AMD for a server
(because the company is using only AMDs for political reasons
or there were applications that really runs faster on an AMD)
will the unpatched 5.0.6 (or before) burnout the AMD and overheat it too?
(IIRC P4 do more with "predition", so software (delay) loops(ouch)
may be "optimized" too much...)
Will the rs506a solve the problem on AMD if it occured?

Dave Butler

unread,
Jan 13, 2003, 10:41:02 AM1/13/03
to
> So I think you can run pre-5.0.6 on a P4; however, don't be
> surprised if the CPU responds by slowing itself way down over time,
> to the point where you may find a higher-end P3 outperforms it.
>
> Remember, it's been a while since I looked into this so take
> my post with a grain of salt and do some research yourself. I'm
> not responsible for any spontaneous CPU combustion that may result
> from you running your choice of non-P4-aware OS on a P4 - you are!

I would appreciate it if someone from SCO could confirm or deny this.
There is a world of difference between 1) Frying a CPU and 2) Simply
running slower, but doing no damage.

If possible, I would like to run OSR 5.0.5 on P4 systems, but have
been scared away because of that tech note on damaging the CPU.

Thank you,

Dave Butler

Tom Parsons

unread,
Jan 13, 2003, 12:06:54 PM1/13/03
to sco...@xenitec.on.ca
Dave Butler enscribed:

Are the P4's really that wonderful?

Ever wonder why IBM's Intel server line uses P3's and the next step is
to Xeon's?
--
==========================================================================
Tom Parsons t...@tegan.com
==========================================================================

Don Williams

unread,
Jan 13, 2003, 6:17:49 PM1/13/03
to

"Don Yakubowski" <do...@tricomp.ca> wrote in message
news:v1oom3g...@corp.supernews.com...

THIS IS A CAUTION NOTE!

I know you can do it and I did, but don't try an Intel D845PEBT2 or similar
Intel board. OSR 5.0.6 and the Bios (or board) don't communicate to set up
the display so you will have nothing on your screen if you update a running
system to that board from another board.

I actually didn't do a new install, just replaced the mother board, had been
using an Intel SE440BX. With the new board there were no characters, just
blocks of color and black on the screen.

Bela will probably publish a manual patch to the OS which will force OSR to
expect a VGA card but you may have to do it in the blind, that is with
nothing on your screen. It does work and I ran that way for a couple of
days until I saw the warning notice quoted below.

Then I added Release Supplement 5.0.6A for OpenServer 5 and once more had no
video and had to do the patch again. That Release Supplemt contains the
following comment in it's text file:

"WARNING Do not run OpenServer on Pentium 4 systems without Release
Supplement 506A. Doing so might damage your hardware.

OpenServer drivers have been upgraded to support Pentium 4 technology.
Before using other drivers on your system, consult your vendor to ensure
that they conform to Pentium 4 specifications."

I have sent several queires to Intel about that board and Unix and they keep
wanting to talk about Windows. In view of the fact that the world is
populated with Open Server systems, and people will want to upgrade the
servers, I think there is a disaster waiting to happen, as it did to me.
Right now I have a forced-Console system which I hope continues to run.
Also, if you go that upgrade route (with that board family) do not install
the USB drivers which come with the supplement. Your system will hang and
you will have another patch job on your hands.

Don Williams

Bela Lubkin

unread,
Jan 14, 2003, 4:54:42 AM1/14/03
to sco...@xenitec.on.ca
Dave Butler wrote:

I have never heard of a machine _actually_ being damaged by OSR5
overheating a P4 CPU. As far as I know, the warnings are all
theoretical. However, we know by code analysis that if everything went
wrong in just the right way, and if we understand Intel's description of
the CPU's behavior, it _would_ be possible to damage a P4 CPU. Fixes
for these theoretical issues were developed and shipped in rs506a.

Operating OSR5 on a P4 without rs506a is a gamble. Are you likely to
win? Yes, but you might not.

What are you doing that you need OSR505 specifically?

>Bela<

Dave Butler

unread,
Jan 14, 2003, 1:31:35 PM1/14/03
to
Thanks for clarifying. We have no specific need for v. 5.0.5, except
that a large number of servers in the field are running it and we
prefer not to support (internally) more than one version. As you know,
it can be challenging to upgrade remote servers.

Also, with 5.0.7 just around the corner (right?), our preference is to
skip the upgrade to 5.0.6.

Thanks for your help,

Dave Butler

Bela Lubkin <be...@caldera.com> wrote in message

Erwan LE BIHAN

unread,
Jan 16, 2003, 1:01:35 PM1/16/03
to
"Bela Lubkin" <be...@caldera.com> a écrit dans le message de news:
2003011401...@mammoth.ca.caldera.com...

> Dave Butler wrote:
>
> > > So I think you can run pre-5.0.6 on a P4; however, don't be
> > > surprised if the CPU responds by slowing itself way down over time,
> > > to the point where you may find a higher-end P3 outperforms it.
> > >
> > > Remember, it's been a while since I looked into this so take
> > > my post with a grain of salt and do some research yourself. I'm
> > > not responsible for any spontaneous CPU combustion that may result
> > > from you running your choice of non-P4-aware OS on a P4 - you are!
> >
> > I would appreciate it if someone from SCO could confirm or deny this.
> > There is a world of difference between 1) Frying a CPU and 2) Simply
> > running slower, but doing no damage.
> >
> > If possible, I would like to run OSR 5.0.5 on P4 systems, but have
> > been scared away because of that tech note on damaging the CPU.
>
> I have never heard of a machine _actually_ being damaged by OSR5
> overheating a P4 CPU. As far as I know, the warnings are all
> theoretical. However, we know by code analysis that if everything went
> wrong in just the right way, and if we understand Intel's description of
> the CPU's behavior, it _would_ be possible to damage a P4 CPU. Fixes
> for these theoretical issues were developed and shipped in rs506a.

I have see one.
My "Beloved Chief Technician", who is unable to read or even more write in
english, has tried the P4 1.5GHz + Sco 5.0.4.
After 1 day, the computer was having difficulties to boot and ,when
achieving booting sco, was very slow.
The 2nd day, there was a system crash. He never achieve to power on the
computer after that. no post. Nothing at all.
I changed the Mobo (Intel 845) and CPU and give him a printed copy of the
oss readme.....
We are now using 5.0.6 with the oss. But He seems to have problems recently,
but I'am a poor technician, so asking me or one of my coworkers is not for
tomorrow !


Reply all
Reply to author
Forward
0 new messages