Strange problem

[pek]

This isn't life or death, I'm most likely going to rma this MB anyway,
but I'd like everyones _constructive_ thoughts as to the cause of my
problem.
My set-up is about 8 months old as follows:

ECS N4SLI-A9 (I know, but it was on sale)
2 gig corsair vs1gb400c3
amd x2 4200+
Saphire x800 gto2
antec smartpower 2.0 500 w
creative xfi extreme music

The problem is that it will occasionally freeze, at least video will
freeze, won't respond to mouse/keyboard and file transfers will stop. And,
here's the strange part, I played music to see if it was a usb problem, the
music will hang in a repeating loop. i.e., it plays the same 5 sec of
sound, 2 sec quiet, 5 sec sound over and over. Everest log says that temps
for cpu never get above 38c, gpu stays in 40's, voltages, as far as on-board
sensors go, are within specs, cpu load never gets higher than 70% on either
core. Drivers and bios are all the latest. The key to the symptom is
probably the repeating music, but even though built pc's back to the 286
days and used pc's back to vic20 days, I can't think of any fault that will
cause these symptoms. Normally, if a pc freezes it freezes solid, cpu
doesn't compute, clock stops, etc. It may be a unique property of the
Nforce 4 chipset that it buffers the pci data and that's why the repeating
sound, but that seems to be really reaching for a solution. Any
suggestions?

pek

 

[paulmd]

This isn't life or death, I'm most likely going to rma this MB anyway,
but I'd like everyones _constructive_ thoughts as to the cause of my
problem.
My set-up is about 8 months old as follows:

Open up the case and check for bad capicitors. if any are bulging or
leaking..... RMA that thing whilst it's under warrenty.

 

[pek]

Caps look ok, Biostar still features this board on their home page, so
I'm pretty sure production probs have been worked out. Of course, they
might just have one humungus unsold inventory. I discovered something else
in my troubleshooting; The freeze happens without fail when I am
transfering a lot of large files (movies or mp3's). Which points me at the
chipset for sure. I have an aftermarket Thermatake fan on it, the old one
was way too noisy.

pek

 

[meow2222]

Caps look ok, Biostar still features this board on their home page, so
I'm pretty sure production probs have been worked out. Of course, they
might just have one humungus unsold inventory. I discovered something else
in my troubleshooting; The freeze happens without fail when I am
transfering a lot of large files (movies or mp3's). Which points me at the
chipset for sure. I have an aftermarket Thermatake fan on it, the old one
was way too noisy.

The sound looping means it hasnt frozen at all. Many audio players do
that when the cpu isnt able to keep up with their demands. The audio
playing stream defaultly gets higher priority than everything else,
including taskman for some reason, so when the cpu is maxed out you get
no response from anything except the player, which just loops because
it isnt getting data refreshed.

Definitely looks like a software problem to me, something maxing out
the cpu.


NT

 

[pek]

Well, the Everest log, up to the freeze point, anyway, shows less that
70% on one core and about 5% on the other, so unless something snuck in
there between one log update and the freeze, I don't _think_ that's it.
I've never seen the dual core maxed out on both cores unless I'm running 2
compute intensive things. The file transfer just hits one heavy and the
other normally stays less than 30%. I'm not ruling it out, but I think
that's lower on the probability list. I wish Biostar had a northbridge temp
sensor, they monitor the shipset voltage for some resaon, but not the temp.
Really threw me at first, Everest reads it as the 2.5vdc mem voltage sensor
but says that it's 1.6 vdc, the chipset voltage.

pek

 

[meow2222]

Well, the Everest log, up to the freeze point, anyway, shows less that
70% on one core and about 5% on the other, so unless something snuck in
there between one log update and the freeze, I don't _think_ that's it.
I've never seen the dual core maxed out on both cores unless I'm running 2
compute intensive things. The file transfer just hits one heavy and the
other normally stays less than 30%. I'm not ruling it out, but I think
that's lower on the probability list. I wish Biostar had a northbridge temp
sensor, they monitor the shipset voltage for some resaon, but not the temp.
Really threw me at first, Everest reads it as the 2.5vdc mem voltage sensor
but says that it's 1.6 vdc, the chipset voltage.

pek

what does your hand say about northbridge temp? I cant say I have the
greatest faith in these hw test suites, they have their uses but
routinely give wrong results.

When an endless loop occurs, logging wont log, the cpu will be too busy
with the player app and the looping app. If taskman is already running
and on performance page, you should still see retraces of cpu use, just
very slow. So this may tell you either way re cpu.


NT

 

[kony]

pek wrote:

what does your hand say about northbridge temp? I cant say I have the
greatest faith in these hw test suites, they have their uses but
routinely give wrong results.

When an endless loop occurs, logging wont log, the cpu will be too busy
with the player app

Again, no.

It is not at all a matter of CPU busy with player or audio.
The audio app could instead be buggy, but that won't make
the CPU any more busy per se, even that endless loop should
not be at high priority.

...and the looping app. If taskman is already running
and on performance page, you should still see retraces of cpu use, just
very slow. So this may tell you either way re cpu.

It isn't.

 

[kony]

Caps look ok, Biostar still features this board on their home page, so
I'm pretty sure production probs have been worked out.

Is it Biostar or PCChips?

Having it on their homepage is no evidence of anything
regading production problems or not.

Of course, they
might just have one humungus unsold inventory. I discovered something else
in my troubleshooting; The freeze happens without fail when I am
transfering a lot of large files (movies or mp3's). Which points me at the
chipset for sure. I have an aftermarket Thermatake fan on it, the old one
was way too noisy.

Maybe chipset but be sure to check the usual things like
temps, voltages, of other parts too, and cables (incl. drive
cables).

If all else fails take off the heatsink/fan from the chipset
and check the interface.

If you had made bios changes, clear CMOS and try it at
default values. Look for a newer board bios. Try another
OS or at least another OS installation to see if same
activities still freeze it.

 

[kony]

The sound looping means it hasnt frozen at all. Many audio players do
that when the cpu isnt able to keep up with their demands.

That never happens when the CPU isn't able to keep up.
When CPU can't keep up, it will stutter, every single time.

The audio
playing stream defaultly gets higher priority than everything else,

Normally, no, only rare malcoded software.

including taskman for some reason, so when the cpu is maxed out you get
no response from anything except the player, which just loops because
it isnt getting data refreshed.

Again, no.

More often the problem is a buggy driver, though if the chip
itself were having problems all bets are off. It certainly
is not a CPU or priority problem.

Definitely looks like a software problem to me, something maxing out
the cpu.

No. We can rule this out even without knowing what is wrong
because of the symptoms.

 

[meow2222]

That never happens when the CPU isn't able to keep up.
When CPU can't keep up, it will stutter, every single time.



Normally, no, only rare malcoded software.



Again, no.

More often the problem is a buggy driver, though if the chip
itself were having problems all bets are off. It certainly
is not a CPU or priority problem.

No. We can rule this out even without knowing what is wrong
because of the symptoms.



Again, no.

It is not at all a matter of CPU busy with player or audio.
The audio app could instead be buggy, but that won't make
the CPU any more busy per se, even that endless loop should
not be at high priority.

The short version is I've had it happen on 2 machines, a P1 and a
celeron. All tasks are at normal priority, including taskman, except
the player audio which is at real-time priority. This is default
settings. Once the audio player runs out of enough CPU to do all the
tasks it needs to do it just loops. CPU sits at 100%, presumably due to
some unstable process somewhere. Getting the PC out of the looping
audio doesnt happen because the looping audio has higher priority than
taskman, and theres no CPU cycles to spare. Theres just enough CPU
available to refresh the CPU use graph slowly. The player in both cases
was winamp.


NT

 

[kony]

The short version is I've had it happen on 2 machines, a P1 and a
celeron.

A p1 may indeed be marginal if not insufficient (depending
on speed and whether MMX & player using it). Any celeron
besides the L2-cacheless 300MHz version is fast enough for
MP3, let alone lesser compressed formats.

All tasks are at normal priority, including taskman, except
the player audio which is at real-time priority.

Are you 100% certain the audio player was at realtime
priority? If it was, this is a very severe defect in the
software, there is no audio player that should EVER be at
realtime priority. I'm talking lawsuit-serious, it should
never be released into the market as it will guarantee
problems.

So take any random audio player, actually all of them except
this one you think might be at realtime priority, and it
should not do as you described.

This is default
settings. Once the audio player runs out of enough CPU to do all the
tasks it needs to do it just loops.

Either:

A) The system had enough CPU power for the audio regardless
of anything else or;

B) It didn't.

If it did, realtime priority didn't cause it from lack of
CPU, it caused it because the background OS functions
weren't working as they should. In short, no application
running should have realtime priority. Abandon such
software and demand money back if it wasn't free.

CPU sits at 100%, presumably due to
some unstable process somewhere.

Getting the PC out of the looping
audio doesnt happen because the looping audio has higher priority than
taskman, and theres no CPU cycles to spare.

Your problem has nothing to do with CPU or audio, the
operating system itself was malfunctioning and would have
done so just the same had it been some other thing being
done besides audio.

just enough CPU
available to refresh the CPU use graph slowly. The player in both cases
was winamp.

Which version?

Again, a Pentium 200MMX, Celeron 300A or better can play MP3
or other audio formats without what you describe. Most
certainly anything that came thereafter can.

It is just not even slightly likely that OP has anything
resembling what you have described.

 

[meow2222]

On 28 Aug 2006 04:25:02 -0700, (e-mail address removed) wrote:

A p1 may indeed be marginal if not insufficient (depending
on speed and whether MMX & player using it). Any celeron
besides the L2-cacheless 300MHz version is fast enough for
MP3, let alone lesser compressed formats.

yes, the running out of CPU resources was not simply due to it playing
an mp3.

Are you 100% certain the audio player was at realtime
priority? If it was, this is a very severe defect in the
software, there is no audio player that should EVER be at
realtime priority. I'm talking lawsuit-serious, it should
never be released into the market as it will guarantee
problems.

eh??

Lemme fire one up now. OK this is a different player on different OS on
a different machine, and am not getting any answers - will try one of
the machines thats prone to this later.

So take any random audio player, actually all of them except
this one you think might be at realtime priority, and it
should not do as you described.



Either:

A) The system had enough CPU power for the audio regardless
of anything else or;

B) It didn't.

If it did, realtime priority didn't cause it from lack of
CPU, it caused it because the background OS functions
weren't working as they should. In short, no application
running should have realtime priority. Abandon such
software and demand money back if it wasn't free.



Your problem has nothing to do with CPU or audio, the
operating system itself was malfunctioning and would have
done so just the same had it been some other thing being
done besides audio.



Which version?

2.79 iirc, will check later

Again, a Pentium 200MMX, Celeron 300A or better can play MP3
or other audio formats without what you describe. Most
certainly anything that came thereafter can.

yes, as can slower machines, if not too busy elsewhere.

It is just not even slightly likely that OP has anything
resembling what you have described.

more later


NT

 

[kony]

yes, the running out of CPU resources was not simply due to it playing
an mp3.

While that may be true, it's not really a sign of a problem
per se... take the highest performance dual or 2 x dual
system and give it enough load and that would happen, or
just wrong priority alone will cause it.

eh??

No software is supposed to run at realtime priority.
A typical and correct response when there is not enough CPU
time left to decode the audio, is the audio playback ceases
for that time interval (until they CPU can again allocate
more time), and whether the audio just pauses or skips ahead
on it's timeline depends on the audio software.

2.79 iirc, will check later

I don't have 2.79, but I had an old 2.91 installer and
installed it. It's playing at normal priority.

Next I'll que up a video file for encoding, set that to
normal, then higher priority to see what winamp does.

At normal priority, video encoding in focus, nothing wrong
with winamp playback (this on a single core system).
With the video encoding set to highest priority, it does
loop as you described.

So it would seem with winamp, wrong priority settings do
cause it. Winamp shouldn't be at realtime priority though,
you might try a newer version if 2.71 does that and
likewise, if it isn't winamp to blame but another app
instead set at an improperly high priority, the importance
of that other app should be questioned (ie- ideally no
software should be running at higher than normal priority
unless it is truely a high priority task such as capturing
*realtime* events like streaming or live/broadcast
audio/video. Even then, priority should be set to high(est)
at most, not realtime.

 

[pek]

At end.....

what does your hand say about northbridge temp? I cant say I have the
greatest faith in these hw test suites, they have their uses but
routinely give wrong results.

When an endless loop occurs, logging wont log, the cpu will be too busy
with the player app and the looping app. If taskman is already running
and on performance page, you should still see retraces of cpu use, just
very slow. So this may tell you either way re cpu.


NT

The task man freezes, too. Everything _apparently_ stops (including
logging) except the audio loop. I took my trusty dmm and the voltages are
all ok, no dips when file transfers take place.

pek

 

[Noozer]

The task man freezes, too. Everything _apparently_ stops (including
logging) except the audio loop. I took my trusty dmm and the voltages are
all ok, no dips when file transfers take place.

The loop is the buffer on the soundcard. It is not being driven by software.

 

[pek]

Is it Biostar or PCChips?

It is Biostar, labled Biostar, anyway. Don't know if Chips oem them for
Biostar. Everest says it's a Biostar, and they're pretty good at
identifying the original maker. Forex, I've got an ECS labled Abit
manufactured mb, Everest called it an Abit board, and Abit has the identical
board on their product list.

Having it on their homepage is no evidence of anything
regading production problems or not.

No, but it's been on sale since 2004, so I _asssume_ that the rev board I
bought new in Dec 2005 from Newegg is latest production model and _should_
have most bugs ironed out.

Maybe chipset but be sure to check the usual things like
temps, voltages, of other parts too, and cables (incl. drive
cables).

Checked with my old dmm, voltages look ok. Newest bios is from Sept 2005,
that's the one I had to load when I first installed the board to use the A64
version I had. Gave the chipset heatsink the finger test, didn't stick, so
it's not _too_ hot. Didn't pull the heatsink, don't want to pull the mb
until I pack it up to rma ( have to cut the plastic retainers and replace
them with new ones, too tight to remove/replace from above). Haven't made
bios changes for at least 6 months, when I installed my SATA drive. No
power sags, surges or other mishaps that I know of, something might have
happened when I was away from it, but that will usually bring the pc down if
it's bad enough to affect it that way. Did clear cmos and check setting, I
left things stock, this isn't an overclockers board, just an entry level to
s939. May try re-loading xp, just a pita to re-load all software,
re-loading is a last resort. Not to mention the damn WGA bs.

 

[kony]

The task man freezes, too. Everything _apparently_ stops (including
logging) except the audio loop. I took my trusty dmm and the voltages are
all ok, no dips when file transfers take place.

You mentioned file transfers, but not where?

Were they over the lan or was source or destination drives
connected to a PCI controller card, OR a discrete controller
chip on the motherboard?

This is a Creative Labs sound card, rigth? Try a different
driver, try tweaking the PCI latency in bios menu or use
"Latency Tool" (Google search it) to raise latency to at
least 96-128, or preferribly remove the sound card entirely
from the system and see if the onboard sound exhibits the
same problem.

 

[kony]

No, but it's been on sale since 2004, so I _asssume_ that the rev board I
bought new in Dec 2005 from Newegg is latest production model and _should_
have most bugs ironed out.

It's not at all uncommon for a revision to change from Dec
2005 to present, though less likely they'd update any
picture just to show a newer silkscreen.

If there is a newer bios available, try it.

Checked with my old dmm, voltages look ok. Newest bios is from Sept 2005,
that's the one I had to load when I first installed the board to use the A64
version I had.

Ok, did you clear CMOS/reset/load defaults?

You might also visit a motherboard oriented web forum in
search of fellow users of that board. Perhaps
http://forums.pcper.com/

Gave the chipset heatsink the finger test, didn't stick, so
it's not _too_ hot.

That isn't revealing, a poor thermal contact will result in
chip getting hotter, heatsink not as hot. Even so, if the
fan isn't spinning incredibly slow, odds are it isn't
overheating unless they'd simply failed to apply thermal
compound at all. Even so, I'd probably pull off the
headsink anyway, probably lap it's base (they are often
rather crude, rough on the bottom), and put a very small
blob of good thermal compound on the center of the chip,
letting pressure spread it out.

Didn't pull the heatsink, don't want to pull the mb
until I pack it up to rma ( have to cut the plastic retainers and replace
them with new ones, too tight to remove/replace from above).

If you can't find the fault, you might as well pull the
board out now, have it lying on a non-conductive surface
(not anti-static packaging) with all non-essential parts
pulled out already to test as a barebones config with only
integrated features (then disabling those for another test),
1 memory module, CPU, heatsink/fan, video.

Haven't made
bios changes for at least 6 months, when I installed my SATA drive. No
power sags, surges or other mishaps that I know of, something might have
happened when I was away from it, but that will usually bring the pc down if
it's bad enough to affect it that way. Did clear cmos and check setting, I
left things stock, this isn't an overclockers board, just an entry level to
s939. May try re-loading xp, just a pita to re-load all software,
re-loading is a last resort. Not to mention the damn WGA bs.

You don't necessarily have to install every last thing, just
use another drive or partition on that one to do a testbed
installation with just the OS, applicable drivers per the
functions (like drive controller and audio, video, chipset
but maybe MS IDE driver instead of the chipset driver).
Only if that works properly would you then look at building
up that installation to a production state.

 

[pek]

This one has replies to your other ms as well....

You mentioned file transfers, but not where?

Were they over the lan or was source or destination drives
connected to a PCI controller card, OR a discrete controller
chip on the motherboard?

Nothing on lan, just internal drive to internal drive and external usb
driver to internal drive and vise versa. All internal are mb ports, no
extra cards except video and xfi, tried the usb drive on other ports as well
as a powered hub also tried another drive in another enclosure. Each time
the file transder got going the pc would freeze. The nforce 4 provides all
the conectivity for the mb as I recall, no extra chip for the sata.

This is a Creative Labs sound card, rigth? Try a different
driver, try tweaking the PCI latency in bios menu or use
"Latency Tool" (Google search it) to raise latency to at
least 96-128, or preferribly remove the sound card entirely
from the system and see if the onboard sound exhibits the
same problem.

I'll try the tool and jack up the latentcy, can't hurt. I'll also look for
an older driver, I always try get the latest driver for anything before I
install. But I've read from other groups that there are some serious
problems with older drivers and static. I may just take out the board in
prep for an rma, since I haven't heard form biostar support, yet, and go to
newegg and start the process there. Looked at other boards, no problem with
biostar mb there, and the biostar usenet group is kinda dead. I can see
that if the bus locks up the xfi would play out it's buffer, but I don't
know why it would repeat. All in all it's the hardest problem I've had on a
pc, except for the time on an asus board (a7v333) the pc wouldn't boot if
the boot drive was the pri master, but would if I moved it to any other
position. I eventually replaced the drive and fixed the problem, it was
just weird that it wouldn't boot with the drive in the pri master. Come to
hink of it, I've got 5 WD 40 gig drives that are problematic.

I'll try some more things today and then send the board back, but I'll need
to have a new one in hand since this is my 'work' pc. I'll keep the
replacement biostar as a backup, or possibly a folding box.

pek

 

[meow2222]

While that may be true, it's not really a sign of a problem
per se... take the highest performance dual or 2 x dual
system and give it enough load and that would happen, or
just wrong priority alone will cause it.

No software is supposed to run at realtime priority.
A typical and correct response when there is not enough CPU
time left to decode the audio, is the audio playback ceases
for that time interval (until they CPU can again allocate
more time), and whether the audio just pauses or skips ahead
on it's timeline depends on the audio software.





I don't have 2.79, but I had an old 2.91 installer and
installed it. It's playing at normal priority.

Next I'll que up a video file for encoding, set that to
normal, then higher priority to see what winamp does.

At normal priority, video encoding in focus, nothing wrong
with winamp playback (this on a single core system).
With the video encoding set to highest priority, it does
loop as you described.

So it would seem with winamp, wrong priority settings do
cause it. Winamp shouldn't be at realtime priority though,
you might try a newer version if 2.71 does that and
likewise, if it isn't winamp to blame but another app
instead set at an improperly high priority, the importance
of that other app should be questioned (ie- ideally no
software should be running at higher than normal priority
unless it is truely a high priority task such as capturing
*realtime* events like streaming or live/broadcast
audio/video. Even then, priority should be set to high(est)
at most, not realtime.

OK, I got to one of the machines that is known for doing this when CPU
gets maxed out. Its the oldest one, a 75MHz pentium. I looked at winamp
2.79's priority and it was set to normal, and played fine. Max it out -
which is very easy - and it stutters. Now I know its gone into looping
at times, producing exactly the symptoms the OP mentioned, and I could
swear it was set to a higher priority at the time... but that priority
bit is only from memory.

My best guess is that the win9x somehow got it set to wrong priority at
times. I do know for sure it has exhibited this lock and loop behaviour
several times, but it wouldnt do it now when I checked. Not sure what
else to say, am not getting as clear a picture as I'd expected to see


NT