Pentium D 920 - What temperature to believe?

[jp4444]

The short:
I've paired a D 920 with the Intel 945GPM board (mATX) in my first HTPC project. The chassis is an nMediaPC 200, which is little cramped, but not so bad - there's lots of ventilation slots on the top, sides, back exhaust, etc. but I am concerned it's a little (a lot?) too hot for my liking.

The long:
The problem in trying to evaluate what to do/improve/replace - I don't know what temp monitor is actually accurate--
In Intel's Utlity "Hardware Monitor" (Windows XP MCE 05) it reads as this (at idle):
"Processor Zone Temp" 50C
"System Zone 1" 42C
"System Zone 2" 42C
"Processor Fan" 2550 RPM
Which is all "in the green" - But the BIOS sings a slightly different story (at idle/sitting in BIOS):
"Processor Temp" 63C
"Internal Temp" 47C
"Remote Temp" 45C
"Processor Fan" 2640 RPM

Everything is about 2-4C cooler when the top/cover is removed - so I'm not TOO worried about overall case ventilation. (I let things idle for about 10 minutes in each state before writing down readings)

To confuse matters more, this chassis comes with a front LCD panel with 2 temps sensors, but they are both in Farenheit and when converted come out 10-20 degrees cooler than both Windows and BIOS temp readings above...
I've tried the attached the "CPU" sensor in two places, 1st shoved as far into the fins on HS as possible, and right now its on the ledge/side of the processor same temp range in both places - 109F-115F (42C-46C) UGGGGGHHHH

What I do know is I've looked over Intel's tech specs for the D9xx, and am not real comfortable with the CPU being up near 65c - if that's the case (see BIOS reading).

So can I trust any of the above methods as accurate? I'm about to give this all up, close things up and wait for an alarm or auto-shutdown, cuz I'm tired of agonizing over it. How can I decide on new HSF without being sure what the actual temps are?

1 other note to this diatribe - this is my first build where I've selected my own PSU (and my first mATX) - does it look beefy enough??? It's pretty high quality IMO - I didn't really consider the wattage very long when I was selecting since I was only planning on 1 HD....

Full specs
nMEDIAPC HTPC 200SA
Intel Pentium D 920 Presler
Intel D945G-PM-LKR (MicroATX)
SILVERSTONE SST-ST365 ATX 360W Power Supply
pqi TURBO 1GB DDR2 SDRAM (PC2 5400)
POWERCOLOR 1600XTBRAVO256 Radeon X1600XT 256MB
Seagate ST3300622AS 300GB 7200 RPM SATA 3.0Gb/s
LITE-ON 16X DVD±R IDE

 

[PaulWuzHere]

Is a 360 watt enough for a system like that? That x1600xt has gotta take alot of juice

 

[jp4444]

yes, that's one of my questions. I haven't really done anything to tax the video card yet - all these temps/problems are at idle. Both BIOS and Intel monitor are reporting sufficient voltage in all areas though....

 

[jp4444]

:hotbounce

I've been messing with the CPU/HS off and on this evening. I was afraid on my initial install that perhaps I tweaked the thermal pad. Anyhow- I removed stock Intel thermal pad (looked a little uneven, but not too bad) and replaced HS with some non-branded compound I had... Booted to windows and immediately got CPU temp warnings. :hotbounce

I'm gonna get some AS 5 tomorrow and try that with stock Intel HSF. That shouldn't be a problem should it (or do I need to use a lame pad)?

If all that fails I'll break the bank open again on a better HSF.

 

[CMH]

uneven = not good.

There's some ways you can make it even, wuch as sanding it on your own, if you know what you're doing. However, I don't really recommend that for the stock aluminium heatsinks, they're way too soft, much harder to sand than copper ones where sanding is recommended to extreme users (need that 100% flatness?).

 

[Samstoned]

I friend of mines new build with a celeron d has the same problem
I don't like thermo pads,but from what I read in some cases without this there is no warrenty
that said
I showed him how to plan the heatsink dwn flat and polish it
his system has no issue with temp now
and is 5 to 15 C cooler than before stock heatsinks are not made flat enough
thats why the pad
and I know this is wierd but I don't use standard compounds
I use copper based antiseize but I grind it a littl finer not one of my proceesors have over heated

PS:

oh ya that pad has to do a burn in for about 4 to 6hrs

 

[jp4444]

uneven = not good.

There's some ways you can make it even, wuch as sanding it on your own, if you know what you're doing. However, I don't really recommend that for the stock aluminium heatsinks, they're way too soft, much harder to sand than copper ones where sanding is recommended to extreme users (need that 100% flatness?).

Sorry- I meant when I took off the heatsink that the original thermal pad looked like it was uneven/displaced a little heavier on one side than the other. Also a very thick ring of material, delineating the outer edges of the HS mating surface, left on the processor (for some reason the HS surface is round and the processor square?). to my eyes, the surfaces are flat

This second crap compound I just tried seems so fluid that it has been nearly completly flushed out of mating area just by the force of clipping in the heatsink. I've read it's good to have it on thin, but the system is NOT stable at all.

For the record I'm not a big fan at all of the clip method on this stock HS - with four seperate areas to push on at once it seems near impossible to get it secured in an even motion/manner.

This is my 4th or 5th build, never had HS mounting trouble b4

 

[jp4444]

samstoned,
did your buddy go with compound or pad on newly surfaced HS w/D proc? I'm thinking of using pad after seeing what just happened to compound - like it all clamps down too tight & squeezes it all out. and yes, I also saw in the Intel docs where things aren't warrantied/reccomended w/o pad.

On the burn-in- yea, I think I got a little antsy when I saw all the conflicting temp readings. Probably should have let it settle in.

Sounds like you have develeoped quite a unique method for mounting cpu/hs. I'm not looking to become a chemical engineer - just get this thing operating at safe temperatures

 

[Samstoned]

ya I get anal with somethings
I was a machine repair maintaince sup for 10 yrs before accident
anyway
he use silver compound on his build after the pad messed up the first time around
after the milling and polish
the only purpose is to remove air voids in the surfaces
a litle goes a long way if both surfaces are really flat
ideally you should see athin line around the chipwhere it is squeezed out
about 5 yrs ago I read an artical ,some guy polished his chipsets too
that is a little to far for me

 

[jp4444]

I'm going to pick up some AS5 today and try reseating it all again.

Anyone have an opinion on if I should go with a higher wattage PS for this system?

 

[jp4444]

I've reseated the stock Intel heatsink with AS 5 and things are back to normal now (no alarms). I'll be sure to give it a lot of time to break-in this time before reassessing the cooling. Still not really sure what sensor readings to trust...