OEM AMD fan or Super ORB?

[MUSTANG347]

I recently put a super ORB on my 900mhz T-bird and it seems almost the same as my OEM fan. I put the thermal grease on after having to carefully scrape the original gum off.

Temps range from high 30's to high 40's.

Why is it about the same performance? Does that fan/heatsink suck or do I have too much grease on there or what?

 

[Mikewarrior2]

a) socket-a temp readings are highly compressed

b) what was the stock heatsink? was it a taisol unit?

C) The s-orb isn't all that.



Mike

 

[ChipNOW]

Have you overclocked? Those temps aren't huge so I'd doubt you'd see much of a difference a/w

 

[Mikewarrior2]

ChipNOW.

Even if he were higher overclocked, wtih the compressed nature of socket-thermistor readings he'd likely never see a huge temp drop or gain switching heatsinks or greases.


Mike

 

[Mindlink]

Regardless, the heat dissipation of the sink by simple physics is related to the difference in temperature of the case air and the fins of the sink. If you put the case in the freezer, the all heatsinks would work better because the CPU will still run at ~40C and the external temp would be ~5C instead of 20C. The s-orb is not cryo-cooling, so don't expect a great drop in temps unless the CPU is reaching 45-50C. That higher gradient is needed to allow the heat to transfer. In addition, Mikewarrior2 makes a point about inaccuracies. Without a thermistor in the CPU die itself, everything is based on approximate input, and there is not guarentee the die is at or even near what the bios says it's at.
Hope this clears it up for you.

Regards.

 

[Hanky]

Mike,

one question came just to my mind...



<< socket-a temp readings are highly compressed >>



...and you also stated temp changes are compressed. So...let's say we have the K7T Pro2 and a T-Bird 800 at idle temperature 38°C. then under full load it reaches 44°C. The measured temp change is 7°C. The idle temp of 38°C is not THAT bad, let's say it's 15°C low with an average HSF. Then on the other side the board has a very hard time reaching higher temps like 55°C (because of compressed temp CHANGES), which is roughly the CPU temp (according to the board) where the system starts to get unstable (as you once wrote), meaning that the real temp is likely in the 70's or even 80's (as this T-Bird is specified for 90°C) so the temp is at least 20°C low.
So is it true the absolute temperature compression is even higher when the real CPU temp is higher?

 

[Mikewarrior2]

the IDLE versus Full Load doesn't apply as much anymore..

Moreso the compressed readings are &quot;compressed&quot; changes when overclocking, changing greases, and heatsinks... The reason the original idle theory of mine doesn't play out is because when the cpu idles, there is virtually no heat on the back. I'd guess that idle temps are probably more accurate versus core temp than are full-load temps.



Mike

 

[Hanky]

Mike,

what is your &quot;original idle theory&quot;?



<< I'd guess that idle temps are probably more accurate versus core temp than are full-load temps. >>



Yep, that's just what I suspected.

 

[Hanky]

...so that's more or less a new effect...compressed changes without changing the &quot;cooling solution&quot;...it's a little bit confusing

 

[Mikewarrior2]

I originally thought that the vast majority fo people getting a 5C change between IDLE and Full Load was one sign of compression.

Its weird, but a lot of people seem to be getting 15-or higher C changes. Don't know why, and I'm confused too...

The compression when changing overclock levels, voltage, heatsinks and gresae still definately applies though.


Mike

 

[Hanky]

What do you think is a realistic temp change between idle and full load for a T-Bird?

 

[Mikewarrior2]

I'm not sure... I hear that t-birds/durons do not have the same HLT command &quot;efficiency&quot; as p3s do, and my p3 drops about 10C, or goes to Half-W output at full idle.

I'd guess that a &quot;realistic&quot; duron/t-bird would drop anywhere from 8 to 12C, depending on heatsink.


Mike

 

[rmzalbar]

Well, my socket A tbird 1.1Ghz non-overclocked with AMD-supplied heatsink (taisol) on a KT7 motherboard in a not-very-well vented case goes from 34°C at idle to a max of 53°C at full load. &quot;system temp&quot; ranges from 28°C to 30°C.

This was after I replaced the original thermal gum with radio shack paste. This dropped my max temp down 2 or 3 degrees, and my idle temp down by about 3 degrees.

Rule out the video card heatng things up, since I test my temps using either Prime95 or Zip Key with brute-force mode. both these programs take my CPU up to the highest temps I've ever seen within about 5 minutes.

If I run in pure DOS mode or let it sit at my bios screen, the CPU stabilizes around 45°C. So yes, ACPI does significantly cool the CPU when you aren't using it. Web browsing, playing MP3s, etc. keep the CPU down around 35°C or so.