MC462 Stupid Question

[datwater]

Hi. This doesn't really pertain to overclocking - I was running an Athlon 1200C at 1200, in an MSI K7T Turbo motherboard, with an AMD Approved Coolermaster hs/fan. Now one day, after months of running with no problems, I get a temp warning that I'm over 70 degrees C. Yipes - fan is still running - so I shutdown the system. I overnight express a Swiftech 462A. I install it exactly as the documentation describes, and power up - get to desktop, then the system shuts down. I can't get it to boot back up. Bring it (CPU) to a friend's place, he has a K7T Turbo as well - won't boot up in his machine either. Red lights across the board on the D-LED. So I purchase a 1333C - only one available at my local shop. Run it with the coolermaster - it runs fine - but temp sensor says it's at about 73 deg C. Is my temp sensor smoked? I put the Swiftech on it - probably a dumb mistake - and it won't boot at all. Power it right down, back to the coolermaster, and it's fine. Now, the only thing about the swiftech documentation that struck me as odd, is that it tells you to install the fan such that it blows downward into the heatsink. This struck me as odd, but this is how I installed it. Is this common? I would think it would be forcing the hot air back onto the processor rather than drawing it away. I guess I have 2 questions - Is my new processor really running at 70-74 deg C all the time, or is my temp sensor hosed? and secondly, Is the fan really supposed to blow toward the heatsink in a Swiftech 462 setup, and if so, what else could I check in that setup?

Thanks,

Dale

 

[peteb]

Are you installing the 462 correctly? Is it seated properly? There are ways to easily get it wrong, shouldered washers, spacers on the springs etc. Did you follow the instructions properly?

What thermal compound are you using etc.

-* This Space For Rent *-
email for application details

 

[Oni]

I have a swiftech mc370 and it took me like 8 times to get it mounted correctly. I still think its not seated correctly because I only get 43-45 degrees celsius teps with my Tbird 1333 and I put a more powerful fan on it than the stock one (delta 38 cfm). In toms review his cooled a Tbird 1 ghz to 35 degrees celsius. I guess I'm still not mounted right but I followed the web site exactly, I even used arctic silver II.
My suggestion is try again and make sure nothing is blocking your HSF from touching anything

 

[Guest]

Oni

Actually I think what you have for temps is close to where it should be. Check out this review of several coolers. The guy is using a 1.33 OC to 1.4. They don't have the swiftech but many others.
http://www.hardocp.com/reviews/cooling/roundup0501/index5.html

The 1.333 puts out a lot more heat than a 1gig. I don't even think its possible to get a 1.33 down to 35C with air cooling. If you are under 45C at 100% load I would say you are doing good.

"Sorry Sir we can't replace your Toshiba cup holder....Toshiba does not make cup holders."

 

[Guest]

sounds like you probably did what I did with mine. The washers for between the mobo and the screw casings is not needed on all mobo's. If put in when not needed the core gets very little contact and thus burns up.

 

[peteb]

wrong - my 1.33 runs at 1.6G with a 462 and does not get over 40-41 degrees, unless the room is really hot. I crunch seti 24x7

I've had it to 29 degrees on idle (stock 1.33).

Pete.

-* This Space For Rent *-
email for application details

 

[Guest]

29 idle not load.

"I don't even think its possible to get a 1.33 down to 35C with air cooling. If you are under 45C at 100% load I would say you are doing good."

So your temps are exactly what I said. (40C at load) Maybe I wasn't clear but I meant 35C under load.

One other thing to consider is how you measure your temps. Most motherboard temp readings are not accurate and can be off as much as 14C.



"Sorry Sir we can't replace your Toshiba cup holder....Toshiba does not make cup holders."

 

[peteb]

Sorry - you were saying it is not possible to get 35 degrees on a 1.33 under load, and mine is currently 11 degrees over system temp, which is 2 degrees over ambient. Currently that puts my loaded 1.33 cpu at 32 degrees.

My 40 degree loaded temp is for a 1.6 t-bird@1.85V - which is not the same thing.

-* This Space For Rent *-
email for application details

 

[Guest]

Well I guess its possible although you never know with the temperature measurment things. I doubt many people actualy have a thermoresistor attached to their core. Some of those things are major wrong. I just have yet to see any heatsink temp tests for the 1.33 that get that low.
Such as
http://www.hardocp.com/reviews/cooling/roundup0501/index5.html

They don't have the 462 in review but the lowest temp they see is 48C with the 1.33 at 1.4

This site did test your cooler though.
http://www.cluboverclocker.com/reviews/heatsinks/swiftech/swiftech_mc462-a/p3.htm
Only they get 33C for a 800 at 1gig. So needless to say I have a hard time believing you are getting your 1.33 to run at 32C. Your HW monitor may say that but I doubt if the core really is that low.

But you never know. Kudos to you if you really get your CPU that low. I am popping a Thermoengine on mine this weekend. Anything is better than the stock cooler.


"Sorry Sir we can't replace your Toshiba cup holder....Toshiba does not make cup holders."

 

[peteb]

Temperature measurements are always subjective. I base mine on calibrated mbm readings.

Some time back I realised that Asus Probe was totally off the scale. Tried MBM which seemed to tie up with BIOS readings, then I got a compunurse type digital thermometer and calibrated against that.

I found Asus probe to report 10 degreec C hotter than bios and MBM. When I checked these temps against a physical reading taken next to the core, I found mbm and bios to read 3 degrees C hotter than the thermistor reading.

So by todays measurements, at a room temp of 19C and 1.33 stock settings under load >1hr my cpu temp was 32C. Using Asus Probe it would have been closer to 45C I expect.

Your point is valid, since there is no international standard of measurement, all mobos have different methods etc, and I've even seen Asus Probe for some report the same as MBM, others like me not. Just telling you what I got and how I got it. You could argue that having a thermistor touching the side of the core is not the true temperature, and I won't argue - but this is how I have seen the majority of 'harder' core overclockers measure their systems when it is possible.

As a final note, I do not leave the thermistor in place all the time, just for spot checks, so it could probably use recalibrating to ensure accuracy. I'll see if I can do that over the weekend and see if my calibration is still true. If it wasnt at that temp though I don't see that I could boot over 1.6G on my AMD. 5 degrees (when I turn the AC up full) makes the difference between sub 1.6G and over 1.6G.

-* This Space For Rent *-
email for application details

 

[Guest]

19C room temp. Yikes you must live in a fridge. I am sure that helps a great deal. Most of the published tests are usually 23-25C room temp.

"Sorry Sir we can't replace your Toshiba cup holder....Toshiba does not make cup holders."

 

[peteb]

typically that is why I think deltas should be quoted, not actual temps.

Actual temperature affects the cpu clocking ability, but in reality all we can do is measure offsets from ambient and see which is easiest to change.

As you point out, a lower room temp affects overall temperature, but the delta of room-system-cpu should stay relatively constant. If my room temp rises to 24 degrees (usual temp) then my cpu will be 36 degrees loaded or so. At 1.6G and 1.85V it will be nearer 41. 17 degrees from room temp to cpu is okay I think for that sort of setup.

-* This Space For Rent *-
email for application details