A7N8X - Temps when PCI card installed?

[Ed]

A7N8X non-dlx v2.0
Barton @ 3200+ 1.675v
2 HDD, 2 CD
Ti4200 4X 64MB
Antec full tower w/400w smartpower

When I disable the onboard sound and use a SB PCI 512 sound card the CPU
temp is higher at idle, even with the case open. (fwiw, the CPU usage
hasn't changed according to WinXP task mngr. with or without the card.)

Why would the CPU run hotter at idle with the SB card in the slot?, is
it a bad sound card? (seems ok), a weak PSU? (hasn't crashed in 5
months), any thoughts?

Temps: room/case/socket/die (MBM5 Avg. @ idle)
Onboard sound - 25.4/27/35/40
SB card - 25.4/28/38/46

btw, I thought maybe it was just a bug reading the temps when I added
the SB card but my thermal sensor case fans (sensors are on CPU
Heatsink) are running faster also.

Cheers,
Ed

 

[Rob Stow]

A7N8X non-dlx v2.0
Barton @ 3200+ 1.675v
2 HDD, 2 CD
Ti4200 4X 64MB
Antec full tower w/400w smartpower

When I disable the onboard sound and use a SB PCI 512 sound card the CPU
temp is higher at idle, even with the case open. (fwiw, the CPU usage
hasn't changed according to WinXP task mngr. with or without the card.)

Why would the CPU run hotter at idle with the SB card in the slot?, is
it a bad sound card? (seems ok), a weak PSU? (hasn't crashed in 5
months), any thoughts?

Temps: room/case/socket/die (MBM5 Avg. @ idle)
Onboard sound - 25.4/27/35/40
SB card - 25.4/28/38/46

btw, I thought maybe it was just a bug reading the temps when I added
the SB card but my thermal sensor case fans (sensors are on CPU
Heatsink) are running faster also.

Cheers,
Ed

Your sound card puts out heat inside the case, which
raises the general air temperature in the case, which
means that being used to cool your cpu is warmer, which
means your cpu runs hotter, which means your cpu fans
spin faster, which means they put out more heat of their
own, which means, ...

Eventually all the interacting thermal effects of all
of your components stabilize at new and higher temperatures.

The difference is particularly dramatic for the cpu because
the efficiency of the heatsink+fan combo is, if IIRC,
proportional to the square of the temperature difference
between the heatsink and the air being blown over it.
In other words, small changes in the ambient air temperature
in the case can result in large changes in the cpu temperature.

As a reverse example, I recently added a case fan to a
friend's computer that only dropped the ambient air
temp in his computer by 4'C, but the cpu temp dropped
by a whopping 17'C.

FWIW, I saw very similar temperature increase when I
disabled the video integrated into my motherboard and
starting using an AGP card. The ambient air temperature
in the case only went up by 2'C, but the cpu temperature
went from 42'C to 55'C.

 

[Ed]

Your sound card puts out heat inside the case, which
raises the general air temperature in the case, which
means that being used to cool your cpu is warmer, which
means your cpu runs hotter, which means your cpu fans
spin faster, which means they put out more heat of their
own, which means, ...

Eventually all the interacting thermal effects of all
of your components stabilize at new and higher temperatures.

The difference is particularly dramatic for the cpu because
the efficiency of the heatsink+fan combo is, if IIRC,
proportional to the square of the temperature difference
between the heatsink and the air being blown over it.
In other words, small changes in the ambient air temperature
in the case can result in large changes in the cpu temperature.

As a reverse example, I recently added a case fan to a
friend's computer that only dropped the ambient air
temp in his computer by 4'C, but the cpu temp dropped
by a whopping 17'C.

FWIW, I saw very similar temperature increase when I
disabled the video integrated into my motherboard and
starting using an AGP card. The ambient air temperature
in the case only went up by 2'C, but the cpu temperature
went from 42'C to 55'C.

I know what you are saying, that is obvious, this is an instant
difference, I can shut down, let everything cool down, with open case,
boot WinXP in about 12 secs and the temps are already higher then they
normally are without the card installed. I actually think the SB card is
sucking too much power or something, the card was running on a faster
bus then 33MHz for last 3-4 years, the overclocked PCI bus might have
screwed it up no?

Cheers,
Ed

 

[Paul]

I know what you are saying, that is obvious, this is an instant
difference, I can shut down, let everything cool down, with open case,
boot WinXP in about 12 secs and the temps are already higher then they
normally are without the card installed. I actually think the SB card is
sucking too much power or something, the card was running on a faster
bus then 33MHz for last 3-4 years, the overclocked PCI bus might have
screwed it up no?

Cheers,
Ed

Does WinXP have any way of measuring the rate at which devices
give interrupts to the system ? Maybe the SB is constantly interrupting
the CPU and causing the kernel load to go up ? On a Unix box, this
would be easy to spot, as some Unix OS have performance monitors
that keep stats like that. I don't know if WinXP has a similar
performance monitor stat for interrupts or not.

The temp increase isn't imaginary, and I doubt the sound card throws
off enough heat for that much temperature change. (A sound chip has no
reason to dissipate more than about 1W.) Perhaps if it disturbed the
air flow enough... but a PCI card is too far away to do anything like
that.

So, it has to be computing load, and it is just a matter of figuring
out whether the load is caused by an interrupt handler, or is due
to some stupid functions added to a SB mixer control panel or related
interface cruft. I had a video card once, where the control panels
and assorted goodies caused a measurable load on the CPU even when
the desktop was idle. Backing off to a previous release of
driver/control panel fixed it.

Paul

 

[Ed]

Does WinXP have any way of measuring the rate at which devices
give interrupts to the system ? Maybe the SB is constantly interrupting
the CPU and causing the kernel load to go up ? On a Unix box, this
would be easy to spot, as some Unix OS have performance monitors
that keep stats like that. I don't know if WinXP has a similar
performance monitor stat for interrupts or not.

Not that I know of.

The temp increase isn't imaginary, and I doubt the sound card throws
off enough heat for that much temperature change. (A sound chip has no
reason to dissipate more than about 1W.) Perhaps if it disturbed the
air flow enough... but a PCI card is too far away to do anything like
that.
Yep


So, it has to be computing load, and it is just a matter of figuring
out whether the load is caused by an interrupt handler, or is due
to some stupid functions added to a SB mixer control panel or related
interface cruft. I had a video card once, where the control panels
and assorted goodies caused a measurable load on the CPU even when
the desktop was idle. Backing off to a previous release of
driver/control panel fixed it.

Paul

Well I have an Audigy 2 coming... we'll see what happens. ?

Cheers,
Ed