wow...it took me a long time

  • Thread starter Thread starter philo
  • Start date Start date
P

philo

I was working on a P-III-1ghz IBM Netvista today.
It's at an internet cafe...and I had loaded Vector Linux on it about 5
months ago.

For the past month or so...when I've come in...
the unit has been completely dead...but after a reboot
it seemed to work OK...at least for a while.

Anyway after I tried to reload both Vector Linux and a few other OS's
with no success (system blanked out during the install)
I took the thing hone to my shop.

After replacing virtually all components,
RAM, power supply, harddrive, net card, video card and cdrom...

sicne there was nothing else to try...
I replaced the cpu. The best I had was a celeron 533.

well that cured the problem...
I don't know (yet) if the cpu could be defective...
or if maybe it just was not quite seated properly...
or if perhaps...because of the slower cpu...the system bus was reset
from 133 - 100 mhz...anyway the thing works.
and for running Linux the C-533 is fine.


The only kicker is that upon bootup...the bios always shows the error:

"No Processor Bios Found"

but if I continue on...the machine runs fine...

when I look at the bios settings...for cpu ID...
it has just question marks for the cpu date stamp...
so I assume that the older cpu not having a date stamp
is the error the bios sees...
but there is no place in the bios to *not* check the date stamp...
and I don't want to put the P-III back in as it locks the setting to a
133 mhz bus speed...and with the C-533 the bus speed is 100mhz
so I can use 100 mhz ram...

I suppose I could just tell the cafe to leave the machine on all the
time...and if they have to reboot to just click past the bios error
 
I was working on a P-III-1ghz IBM Netvista today.
It's at an internet cafe...and I had loaded Vector Linux on it about 5
months ago.

For the past month or so...when I've come in...
the unit has been completely dead...but after a reboot
it seemed to work OK...at least for a while.

Anyway after I tried to reload both Vector Linux and a few other OS's
with no success (system blanked out during the install)
I took the thing hone to my shop.

After replacing virtually all components,
RAM, power supply, harddrive, net card, video card and cdrom...

sicne there was nothing else to try...
I replaced the cpu. The best I had was a celeron 533.

well that cured the problem...
I don't know (yet) if the cpu could be defective...
or if maybe it just was not quite seated properly...
or if perhaps...because of the slower cpu...the system bus was reset
from 133 - 100 mhz...anyway the thing works.
and for running Linux the C-533 is fine.


The only kicker is that upon bootup...the bios always shows the error:

"No Processor Bios Found"

but if I continue on...the machine runs fine...

when I look at the bios settings...for cpu ID...
it has just question marks for the cpu date stamp...
so I assume that the older cpu not having a date stamp
is the error the bios sees...
but there is no place in the bios to *not* check the date stamp...
and I don't want to put the P-III back in as it locks the setting to a
133 mhz bus speed...and with the C-533 the bus speed is 100mhz
so I can use 100 mhz ram...

I suppose I could just tell the cafe to leave the machine on all the
time...and if they have to reboot to just click past the bios error
Click to expand...

Try unplugging AC, then pulling the battery for 10 minutes.
Next, you power up the system and at least enter the bios
menu then exit and save it... even if you change nothing
though presumably you'd at least set the system clock.
 
philo <[email protected]> said:
I was working on a P-III-1ghz IBM Netvista today.
It's at an internet cafe...and I had loaded Vector Linux on it about 5
months ago.

For the past month or so...when I've come in...
the unit has been completely dead...but after a reboot
it seemed to work OK...at least for a while.

Anyway after I tried to reload both Vector Linux and a few other OS's
with no success (system blanked out during the install)
I took the thing hone to my shop.

After replacing virtually all components,
RAM, power supply, harddrive, net card, video card and cdrom...

sicne there was nothing else to try...
I replaced the cpu. The best I had was a celeron 533.

well that cured the problem...
I don't know (yet) if the cpu could be defective...
or if maybe it just was not quite seated properly...
or if perhaps...because of the slower cpu...the system bus was reset
from 133 - 100 mhz...anyway the thing works.
and for running Linux the C-533 is fine.


The only kicker is that upon bootup...the bios always shows the error:

"No Processor Bios Found"

but if I continue on...the machine runs fine...

when I look at the bios settings...for cpu ID...
it has just question marks for the cpu date stamp...
so I assume that the older cpu not having a date stamp
is the error the bios sees...
but there is no place in the bios to *not* check the date stamp...
and I don't want to put the P-III back in as it locks the setting to a
133 mhz bus speed...and with the C-533 the bus speed is 100mhz
so I can use 100 mhz ram...

I suppose I could just tell the cafe to leave the machine on all the
time...and if they have to reboot to just click past the bios error
Click to expand...

The BIOS could be missing a microcode patch for the processor.
Usually a BIOS will have a number of microcode patches stored
in one of the BIOS modules. Each one is intended for a different
family code of processor. The microcode patch fixes errors in
the processor design.

Flash upgrading the BIOS chip is one way to try to fix it.
The BIOS release notes would say something like "supports new
CPUs", to make such an attempt worthwhile.

In the case of Award BIOS, the CTMC program can be
used to load a microcode segment. Microcode in those days
was 2KB long. If you identify the processor, then get a
BIOS file from some other motherboard, you can extract
a microcode segment, and use CTMC to load it. Some Award
BIOS have a "volatile" area in the flash, which is a
cache for the most recently used microcode patch. CTMC
can load the volatile area with new microcode. The microcode
stays in place until you change the processor and muck things
up. Then, you would need to use CMTC again, if the newer
processor was also missing from the main BIOS. I've used
CTMC on a P2B-S, to add microcode for a Tualatin, which
would not otherwise be supported by the BIOS. It succeeds
at removing the warning message, and letting the machine boot.

ftp://ftp.heise.de/pub/ct/ctsi/ctmc10.zip

The benefit of using CTMC, is that you don't run the danger
of a flash upgrade failing. What CTMC is doing, is using a
BIOS call to only flash a 2KB area in the flash chips, so the
whole flash image is not endangered by the operation. That is
the main benefit of the method. And if the motherboard maker
doesn't offer a more up-to-date BIOS, then it may be the
only way to get microcode support. CTMC won't work on a
AMI BIOS, because the necessary BIOS call is missing.

What I find strange about your problem, is the P-III is supported
and the Celeron is not. Usually microcode for both should be
present in the BIOS, unless this "Netvista" box only had
microcode loaded for the exact processor shipped with the
box. Which would be shortsighted on the part of IBM.

Paul
 
What I find strange about your problem, is the P-III is supported
and the Celeron is not. Usually microcode for both should be
present in the BIOS, unless this "Netvista" box only had
microcode loaded for the exact processor shipped with the
box. Which would be shortsighted on the part of IBM.

Paul
Click to expand...

Yes...
IBM's later machines sure are not what they used to be...
I can see why they sold out of the PC business...
 
Try unplugging AC, then pulling the battery for 10 minutes.
Next, you power up the system and at least enter the bios
menu then exit and save it... even if you change nothing
though presumably you'd at least set the system clock.
Click to expand...


the sytem clock is ok...
just the cpu date stamp does not exist
 
the sytem clock is ok...
Click to expand...

Then did you not clear CMOS?
I"m suggesting, clear the cmos. That will reset the clock
so when you go into the bios and save it, you might as well
then set the clock right. Clock cannot be right if not
reset after a clear cmos.

just the cpu date stamp does not exist
Click to expand...

Yes so you wrote but that is not necessary to run any normal
sytem so it could be a random error with a non-obvious
solution.
 
Yes so you wrote but that is not necessary to run any normal
sytem so it could be a random error with a non-obvious
solution.
Click to expand...


the clock is fine!!!!

it just cannot pick up the cpu id date
as there is none on the older celeron
 
the clock is fine!!!!

it just cannot pick up the cpu id date
as there is none on the older celeron
Click to expand...


Read again:

I never suggested the clock was unfine.

I suggested, "maybe" the CMOS was not actually cleared.

IF you never had to set the clock to the correct time, then
you have not cleared the CMOS. That is why I mentioned
the clock in the first place, that AFTER YOU CLEAR THE CMOS,
then go into the bios menus, you will save the settings.

WHILE you're in the bios menu, you might as well set the
clock, because if you did clear the CMOS successfully, the
clock will not be correct anymore.

In other words, forget about what the system shows on the
screen and just clearn the cmos, enter bios, "save settings"
(Just DO IT, not "think about something or other"). This is
a very easy and standard practice but it seems we still have
no idea if you have done it yet.
 
Read again:

I never suggested the clock was unfine.

I suggested, "maybe" the CMOS was not actually cleared.

IF you never had to set the clock to the correct time, then
you have not cleared the CMOS. That is why I mentioned
the clock in the first place, that AFTER YOU CLEAR THE CMOS,
then go into the bios menus, you will save the settings.

WHILE you're in the bios menu, you might as well set the
clock, because if you did clear the CMOS successfully, the
clock will not be correct anymore.

In other words, forget about what the system shows on the
screen and just clearn the cmos, enter bios, "save settings"
(Just DO IT, not "think about something or other"). This is
a very easy and standard practice but it seems we still have
no idea if you have done it yet.
Click to expand...


Ok I should have mentioned then that I have cleared the cmos several times.

If I put in the p-III -1000 (and i tried it again)
I get *no* cmos errors...it's just that as the OS loads...the entore machine
just blanks out...

I have a number of celerons here, 400 mhz and a 533 which work fine in the
machine
as far as running the OS...
but when the machine is first turned on the bios stops with the error:

no processor bios found


this is not that big of a deal actually
as all one has to to is scroll down once and hit "continue"

but I think before I put the machine back in operation I'll look at a bios
"upgrade"... though it's really just to support a lesser cpu.

since I tried the p-III 1000 twice now and the same problem occurred...
it looks like the first cpu I've ever seen that was defective but had not
failed
completely
 
In the case of Award BIOS, the CTMC program can be
used to load a microcode segment. Microcode in those days
was 2KB long. If you identify the processor, then get a
BIOS file from some other motherboard, you can extract
a microcode segment, and use CTMC to load it. Some Award
BIOS have a "volatile" area in the flash, which is a
cache for the most recently used microcode patch. CTMC
can load the volatile area with new microcode. The microcode
stays in place until you change the processor and muck things
up. Then, you would need to use CMTC again, if the newer
processor was also missing from the main BIOS. I've used
CTMC on a P2B-S, to add microcode for a Tualatin, which
would not otherwise be supported by the BIOS. It succeeds
at removing the warning message, and letting the machine boot.

ftp://ftp.heise.de/pub/ct/ctsi/ctmc10.zip

The benefit of using CTMC, is that you don't run the danger
of a flash upgrade failing. What CTMC is doing, is using a
BIOS call to only flash a 2KB area in the flash chips, so the
whole flash image is not endangered by the operation. That is
the main benefit of the method. And if the motherboard maker
doesn't offer a more up-to-date BIOS, then it may be the
only way to get microcode support. CTMC won't work on a
AMI BIOS, because the necessary BIOS call is missing.
Click to expand...

thanks for the link...
however when i tried ctmc.exe
the message i got was " microcode update bios not supported"

so i think i'll just leave the machine as it is.

they will prob. leave the machine on most of the time anyway...
and to click past one bios message should not be a big deal...
 
Ok I should have mentioned then that I have cleared the cmos several times.

If I put in the p-III -1000 (and i tried it again)
I get *no* cmos errors...it's just that as the OS loads...the entore machine
just blanks out...

I have a number of celerons here, 400 mhz and a 533 which work fine in the
machine
as far as running the OS...
but when the machine is first turned on the bios stops with the error:

no processor bios found
Click to expand...


What was the past history of the system? You wrote about
installing Linux 5 months ago and it was in an internet
cafe, but what is the overall state of the system, has it
had a lot of "on" hours? Which Netvista case and power
supply? I have a couple here that were pretty stingily
endowed with about 75W PSU. Such a psu, if in yours, may be
just worn out at this point. Does the motherboard have the
full compliement of FSB settings? I wonder if downclocking
the CPU to 66MHz FSB would make any difference? That is,
the lower bus might make the board more stable or reduce
power consumption on what may be a dying PSU. If the PSU is
dying, it could be that you can manage to get a slower CPU
such as the celeron working but given another few weeks of
service, the system has degraded further and the Celeorn
won't work either.

There were also some IBM boards with faulty capacitors,
though I don't recall today which those were. You might
examine the board if you hadn't already.

this is not that big of a deal actually
as all one has to to is scroll down once and hit "continue"
Click to expand...

Is there a setting in the bios to disable "Halt on Error" or
something similar?

but I think before I put the machine back in operation I'll look at a bios
"upgrade"... though it's really just to support a lesser cpu.
Click to expand...

Also try a bios downgrade, an older one as it seems as
likely the old cpu microcodes were removed to make room for
newer ones.
since I tried the p-III 1000 twice now and the same problem occurred...
it looks like the first cpu I've ever seen that was defective but had not
failed
completely
Click to expand...


I'd suspect the bus speed or power more than a partial
failure. That is, unless you had a persistent overheating
situation such as a fan failure.
 
I'd suspect the bus speed or power more than a partial
failure. That is, unless you had a persistent overheating
situation such as a fan failure.
Click to expand...


As I think I may have mentioned previously I substituted
virtually all parts including the power supply before I decided to
substitute the cpu.

Other than the cpu...about all I did not replace was the mobo and the
case!!!!!

The machine is about a year old and the cafe has 4 other identical machines.

In the time I've been maintaining them...
one lost a HD and on another...the RAM died.

The machines run pretty close to 24/7 but since the cafe has only had
them a year...they seem to have had more than their share of problems...

Anyway...the bios in the machine has no option to adjust the cpu
clocking...it certainly is not one of the better machines I've worked on.


Once in a while I've come across a machine that's been totally dead...
and simply by removing the cpu and putting it back in...has fixed the
problem...
so I thought there was at least a small possibility that the cpu might
have just had a pin that was not making good contact...
but last night when I was trying a few different Celerons I had...
I put the original cpu back in the machine...and the original problem
returned.

Just a guess here...but perhaps one of the fine wires inside broke
off...I'd think the chances of a partial semiconductor failure would be
quite small.


Anyway I let the machine run overnite and it's fine.
With the original cpu...the machine would "blank-out" anywhere between
10 minutes and one hour (or so)
 
so I thought there was at least a small possibility that the cpu might
have just had a pin that was not making good contact...
but last night when I was trying a few different Celerons I had...
I put the original cpu back in the machine...and the original problem
returned.

Just a guess here...but perhaps one of the fine wires inside broke
off...I'd think the chances of a partial semiconductor failure would be
quite small.
Click to expand...

I suppose if it were a manufacturing error, some sort of
defect, it's theoretically possible but seems unlikely. The
wires inside aren't directly coupled to the pins on the
bottom though, they lead up to pads on the bottom onto which
the pins are bonded. You can break a pin completely off,
pull even the base of it off and the pad will still be
connected to the internal wire... something I verified by
taking a bit of conductive epoxy to reattach a pin once...
and it's a PITA to do that but it was more of a passing
interest than a necessity, I'd not put such a CPU into any
important use.
 
I suppose if it were a manufacturing error, some sort of
defect, it's theoretically possible but seems unlikely. The
wires inside aren't directly coupled to the pins on the
bottom though, they lead up to pads on the bottom onto which
the pins are bonded. You can break a pin completely off,
pull even the base of it off and the pad will still be
connected to the internal wire... something I verified by
taking a bit of conductive epoxy to reattach a pin once...
and it's a PITA to do that but it was more of a passing
interest than a necessity, I'd not put such a CPU into any
important use.
Click to expand...


I'm going to see if I have another socket 370 mobo in my shop
and try the questionable cpu in it.

I'm currently setting up another machine for the Cafe...
and will use the one from the Cafe elsewhere

thanks for the info
 
Back
Top