PCI audio crackles P4P800 help needed!!

  • Thread starter Thread starter Terry
  • Start date Start date
T

Terry

I wonder if anyone can throw any light on a problem I'm having-
With my new P4P800 board, I am getting a crackling sound with my
PCI sound card when I use it at 88.2K or 96k sample-rates.
I have no other PCI cards installed and I've disabled EVERYTHING
I can on the P4P800 board, but this seems to make no difference.
(This sound card works fine on my old pc which is an AMD 1Ghz
with Gigabyte GA7ZXR m/board.)

Looking close up at a sine wave I recorded, there are errors in
the values of some individual samples throughout the recorded
wave.
I suppose it's to do with the PCI bus, something interupting the
data to and from the sound card, but what?
I'm using a P4 2.8Ghz FSB800, 2x512mB Kingston PC3500 ram, 2x
SATA 120GB Seagate, ATI radeon 7000. WIN98SE

I would appreciate any help on this.

Regards,

Terry
 
I wonder if anyone can throw any light on a problem I'm having-
With my new P4P800 board, I am getting a crackling sound with my
PCI sound card when I use it at 88.2K or 96k sample-rates.
I have no other PCI cards installed and I've disabled EVERYTHING
I can on the P4P800 board, but this seems to make no difference.
(This sound card works fine on my old pc which is an AMD 1Ghz
with Gigabyte GA7ZXR m/board.)

Looking close up at a sine wave I recorded, there are errors in
the values of some individual samples throughout the recorded
wave.
I suppose it's to do with the PCI bus, something interupting the
data to and from the sound card, but what?
I'm using a P4 2.8Ghz FSB800, 2x512mB Kingston PC3500 ram, 2x
SATA 120GB Seagate, ATI radeon 7000. WIN98SE

I would appreciate any help on this.

Regards,

Terry
Click to expand...

Try moving the card to a different slot. That usually works.

Dan
 
Heihachi said:
Try moving the card to a different slot. That usually works.

Dan
Click to expand...

I'd suggest the slot trick as well, but when I look at the IRQ table
in the manual for this board, if you believe the table, there isn't
an optimal slot to be had. I have trouble believing the contents of
this table, because the USB controller is inside the Southbridge,
and I'm not certain whether the interrupt signals as a result, are
physically sharing with actual PCI cards or not. If you choose to
believe that the USB ports aren't really in that table, then things
don't look so bad.

In addition to playing "whack a mole" with the slots, I would
recommend checking "ICH Delayed Transaction" and making sure it is
[Enabled]. Delayed transaction is a mechanism, where if an access is
attempted to a slow peripheral device (IDE disk interface seems to be
a culprit), the access is delayed and the PCI bus becomes freed up
for other (potentially higher priority) access cycles.

Without enabling Delayed Transaction, what happens is, an access into
slower parts of the Southbridge can last up to 50 PCI clock cycles.
During this time, the PCI bus is just hung up, waiting for a response.
When Delayed Transaction is enabled, the transaction is remembered,
and the transaction on the bus aborted. Some number of PCI accesses
by other devices (say a sound card for example) can take place, until
the transaction to the Southbridge is tried again. If the resource
in the Southbridge is now ready with its data, the "slow" transaction
completes. Otherwise, the transaction is disconnected again, only
to be tried again later. At some point, the slow data comes back.

By doing this, it is possible for fast and slow devices to get along
with one another, without a slow device starving the fast device
when it needs to do DMA. I believe sound cards have FIFOs, so data
is delivered in bursts, but I've read some posts on USENET from people
who claim that the FIFO isn't handled very intelligently. The FIFO
is allowed to get very full or very empty, before it is serviced,
rather than setting the threshold around half full. Depending on the
depth of any FIFO, and at what point it requests service,
determine how much abuse from a slow device it can take before being
starved or overflowing. That is where the snap, crackle, pop comes from.

Another setting of secondary value is the "PCI Latency Timer". This
timer prevents bus hogging. Many motherboards default to a setting of
32, and some people who are trying to get artificially high transfer
rates sometimes try to increase this to 64 or 128. A high value for
this timer leads to bus hogging, where a device with large chunks of
data to burst, prevents other cards from gaining access to the PCI bus.
Since you've disabled so much stuff in the machine, I wouldn't expect
changing this to help in your current situation. In my tests with an
older board, I found acceptable values to be between 16 and 32. Values
less than this, like 12 or 8, lead to the whole Windows desktop slowing
right down (almost stuttering). So, for me, 16 was as low as I could
comfortably set this.

HTH,
Paul
 
Hi Paul,

Thanks for your input on this, I appreciate your help.

I have tried all sorts of settings, including the "ICH Delayed
Transaction" and "PCI Latency Timer" which you mentioned, but
with no apparent effect. I've tried latency settings of 32
(minimum on this board) right through to max.
As I mentioned earlier, this card works fine with my old PC, a
far less powerful setup.

I've been wondering if the ACPI has anything to do with this? A
while back I remember reading some DAW users saying ACPI is bad
for audio use.

If you have any other tips or info, I would appreciate hearing
from you.

Regards,

Terry.
 
I wonder if anyone can throw any light on a problem I'm having-
With my new P4P800 board, I am getting a crackling sound with my
PCI sound card when I use it at 88.2K or 96k sample-rates.
I have no other PCI cards installed and I've disabled EVERYTHING
I can on the P4P800 board, but this seems to make no difference.
(This sound card works fine on my old pc which is an AMD 1Ghz
with Gigabyte GA7ZXR m/board.)

Looking close up at a sine wave I recorded, there are errors in
the values of some individual samples throughout the recorded
wave.
I suppose it's to do with the PCI bus, something interupting the
data to and from the sound card, but what?
I'm using a P4 2.8Ghz FSB800, 2x512mB Kingston PC3500 ram, 2x
SATA 120GB Seagate, ATI radeon 7000. WIN98SE

I would appreciate any help on this.

Regards,

Terry
Click to expand...

What sound card?



--
Free Windows/PC help,
http://www.geocities.com/sheppola/trouble.html
email shepATpartyheld.de
Free songs download,
http://www.soundclick.com/bands/8/nomessiahsmusic.htm
 
Paul said:
Heihachi said:
Try moving the card to a different slot. That usually works.

Dan
Click to expand...

I'd suggest the slot trick as well, but when I look at the IRQ table
in the manual for this board, if you believe the table, there isn't
an optimal slot to be had. I have trouble believing the contents of
this table, because the USB controller is inside the Southbridge,
and I'm not certain whether the interrupt signals as a result, are
physically sharing with actual PCI cards or not. If you choose to
believe that the USB ports aren't really in that table, then things
don't look so bad.

In addition to playing "whack a mole" with the slots, I would
recommend checking "ICH Delayed Transaction" and making sure it is
[Enabled]. Delayed transaction is a mechanism, where if an access is
attempted to a slow peripheral device (IDE disk interface seems to be
a culprit), the access is delayed and the PCI bus becomes freed up
for other (potentially higher priority) access cycles.

Without enabling Delayed Transaction, what happens is, an access into
slower parts of the Southbridge can last up to 50 PCI clock cycles.
During this time, the PCI bus is just hung up, waiting for a response.
When Delayed Transaction is enabled, the transaction is remembered,
and the transaction on the bus aborted. Some number of PCI accesses
by other devices (say a sound card for example) can take place, until
the transaction to the Southbridge is tried again. If the resource
in the Southbridge is now ready with its data, the "slow" transaction
completes. Otherwise, the transaction is disconnected again, only
to be tried again later. At some point, the slow data comes back.

By doing this, it is possible for fast and slow devices to get along
with one another, without a slow device starving the fast device
when it needs to do DMA. I believe sound cards have FIFOs, so data
is delivered in bursts, but I've read some posts on USENET from people
who claim that the FIFO isn't handled very intelligently. The FIFO
is allowed to get very full or very empty, before it is serviced,
rather than setting the threshold around half full. Depending on the
depth of any FIFO, and at what point it requests service,
determine how much abuse from a slow device it can take before being
starved or overflowing. That is where the snap, crackle, pop comes from.

Another setting of secondary value is the "PCI Latency Timer". This
timer prevents bus hogging. Many motherboards default to a setting of
32, and some people who are trying to get artificially high transfer
rates sometimes try to increase this to 64 or 128. A high value for
this timer leads to bus hogging, where a device with large chunks of
data to burst, prevents other cards from gaining access to the PCI bus.
Since you've disabled so much stuff in the machine, I wouldn't expect
changing this to help in your current situation. In my tests with an
older board, I found acceptable values to be between 16 and 32. Values
less than this, like 12 or 8, lead to the whole Windows desktop slowing
right down (almost stuttering). So, for me, 16 was as low as I could
comfortably set this.
Click to expand...

That was a very informative reply. I've always wondered about those two
BIOS settings. So should "ICH Delayed Transaction" ever be set to Disabled?
Is there any downside of setting it to Enabled?

Thanks,
Dave
 
My first advice would be to not use Windows 98 - use 2000 or XP instead.

Modern motherboards support APIC interrupt delivery, which allows more than
16 physical interrupt lines (24 or more). It's possible you have something
sharing an IRQ with the sound card and its driver is delaying the interrupt
servicing for the sound card. With APIC it's much less likely that devices
will end up sharing.

However, Windows 9x/Me do not support APIC at all, only PIC mode which jams
everything into 16 interrupt lines.
 
Terry said:
The sound card is a "MARIAN MARC4m" 4 analog. in/out +2 midi
in/out. 24bit at 44.1k, 48k, 88.2k, or 96k sample-rates.

(www.marian.de)



Regards,

Terry
Click to expand...

Another setting you could play with is "Plug And Play OS". Setting
this to [No] will allow the BIOS to define the resources before
boot. Setting it to [Yes] leaves most of the setup to Windows.
Some people find the OS is poor at this, and the BIOS does a better
job of setting IRQs and I/O spaces. Some people in this group even
recommend a No versus Yes setting for this, based on which OS it is
you are using.

HTH,
Paul
 
(e-mail address removed) (Paul) wrote in
Another setting you could play with is "Plug And Play OS".
Setting this to [No] will allow the BIOS to define the
resources before boot. Setting it to [Yes] leaves most of
the setup to Windows. Some people find the OS is poor at
this, and the BIOS does a better job of setting IRQs and
I/O spaces. Some people in this group even recommend a No
versus Yes setting for this, based on which OS it is you
are using.

HTH,
Paul
Click to expand...

Thanks Paul, but I have tried this setting both ways, but no
luck.
In my attempts to solve this problem, I did manage to get the
sound card on a high priority IRQ, and apparently not sharing
with any other stuff (except for the ACPI IRQ Holder for PCI IRQ
Steering), but the problem persists!
I wonder if it could be something interupting the PCI that is
not visible in the normal way, maybe some process to do with the
ACPI? I am a little bit in the dark on this sort of thing.

Regards,

Terry
 
(e-mail address removed) (Paul) wrote in
Another setting you could play with is "Plug And Play OS".
Setting this to [No] will allow the BIOS to define the
resources before boot. Setting it to [Yes] leaves most of
the setup to Windows. Some people find the OS is poor at
this, and the BIOS does a better job of setting IRQs and
I/O spaces. Some people in this group even recommend a No
versus Yes setting for this, based on which OS it is you
are using.

HTH,
Paul
Click to expand...

Thanks Paul, but I have tried this setting both ways, but no
luck.
In my attempts to solve this problem, I did manage to get the
sound card on a high priority IRQ, and apparently not sharing
with any other stuff (except for the ACPI IRQ Holder for PCI IRQ
Steering), but the problem persists!
I wonder if it could be something interupting the PCI that is
not visible in the normal way, maybe some process to do with the
ACPI? I am a little bit in the dark on this sort of thing.

Regards,

Terry
Click to expand...

Try the card in each slot but boot to Safemode as per,
http://www.geocities.com/sheppola/drivers.html
each time you switch off and re-insert the card.
Also if you have the option
Device Manager/System Devices/PCI bus/IRQ steering make sure it's
enabled but only tick the Get IRQ Table from BIOS option and if no
good try the others.
Requires a re-boot but make a reg back up 1st of your default settings
as per below,

If a problem is less than 5 days old in win98 then restart the system
repeatedly pressing the F8 key and choose Command prompt only Dos or
restart in Msdos mode and type scanreg/restore and try an older
reg(cab file)backup,however you may have to re-install some programs
and user changes if they were installed after this date.
You can type just scanreg and say,"Yes" if you want to make a fresh
registry back up.
You can then also type
scanreg/fix and then scanreg/opt to fix and compact the registry.You
can type win or exit from Msdos mode to get back to windows.
You can use the free WRP program off here to back up the registry as
well as it saves more files.
http://www.geocities.com/sheppola/freeprog.html
HTH :O)



--
Free Windows/PC help,
http://www.geocities.com/sheppola/trouble.html
email shepATpartyheld.de
Free songs download,
http://www.soundclick.com/bands/8/nomessiahsmusic.htm
 
Try the card in each slot but boot to Safemode as per,
http://www.geocities.com/sheppola/drivers.html
each time you switch off and re-insert the card.
Also if you have the option
Device Manager/System Devices/PCI bus/IRQ steering make
sure it's enabled but only tick the Get IRQ Table from BIOS
option and if no good try the others.
Requires a re-boot but make a reg back up 1st of your
default settings as per below,

If a problem is less than 5 days old in win98 then restart
the system repeatedly pressing the F8 key and choose
Command prompt only Dos or restart in Msdos mode and type
scanreg/restore and try an older reg(cab
file)backup,however you may have to re-install some
programs and user changes if they were installed after this
date. You can type just scanreg and say,"Yes" if you want
to make a fresh registry back up.
You can then also type
scanreg/fix and then scanreg/opt to fix and compact the
registry.You can type win or exit from Msdos mode to get
back to windows. You can use the free WRP program off here
to back up the registry as well as it saves more files.
http://www.geocities.com/sheppola/freeprog.html
HTH :O)
Click to expand...

Hi Shep,

Thanks for the tutorial! very useful tips.

Regards,

Terry
 
Back
Top