CPU Cooler install trouble(long and lots of questions).

Terry1212

Distinguished
Aug 3, 2009
143
0
18,690
Some of you might remember me posting a thread about running Prime95 and my CPU reaching 63C. I had been using the stock cooler, but this was too hot for me. The stock cooler was also very loud, which I didn't like. I decided to order the Zalman 9700 NT CPU cooler from Newegg. Initially I was going to order AS5 along with the cooler, but decided not too because I had read the included Zalman Super Thermal Grease(STG-1) was just as good.

I decided to install the cooler today. When I took off the stock HSF, I noticed the cured pre-applied thermal compound was thicker than what people have ALWAYS recommended to me: the layer should be as thin as possible. Moving on, I took a lint free cloth, dabbed it with isopropyl alchohol, and cleaned off the CPU. Out of all the computers I've built, this is the first that I have installed an aftermarket cooler.

I opened the STG-1 and noticed it's a lot thinner than the AS5. More like a nail polish consistency. I put a circle of the grease in the center of the CPU. Diameter was about the diameter of a quarter(obviously not as thick though). I took a piece of grocery bag, put it over my finger, and spread grease outward trying to make it as thin as possible.

I took the cooler and put the AM3 latch through it as shown in the picture. I took out the video card so I could put it on easier. I put it on, pushed down the latch until it locked on. Plugged in the cooler. I hooked up the computer and booted into Windows. After all the startup items were done loading, I opened HWMonitor and watched the temp slowly decrease. I let the computer idle for about 5 minutes and the temp had reached a minimum of 35C(hovering up slightly every so often). Previously the temp had reached a minimum of 37C. I'm down 2C on idle. Not breathtaking, but still lower than stock.

I wanted to get a load temp, so I opened Prime95 and ran the blend test. Something that I did not expect happened. Previously, when I ran the test with the stock HSF, the temp would gradually increase and eventually would peak at ~61-63C. This did not happen with this cooler. The temp increased rapidly and I watched it climb, 62, 63, 64, and eventually reached 70C before I would NOT let it go any further so I stopped the test. I decided to check and make sure the HSF was on right, so I powered down. I noticed it wasn't EXACTLY(I emphasize this word) centered, I turned it ever so slightly so it appeared to be as centered as possible. Booted up again. Idle was the same, but again when running Prime95, the temp SHOT up. 62, 63, 64, 65, 66, 67. I didn't want to chance it, so I stopped the test.

I was thinking the consistency of this stuff is so thin, and maybe I didn't put enough on the CPU. So I powered down. Took the HSF off, cleaned it and the CPU off as I did before. This time I put a slight bit more, but still rather thin, because, again, as I've heard from the majority, it SHOULD be very very thin.

Booted into Windows again. Minimum idle temp again at 35C. When it came time to run Prime95, again, the temp rapidly increased. This time not as fast but still fast. I decided to let it run for a few minutes. It got to 66 rather quickly, but the increased slowed. 67 came more slowly, and 68 a lot slower. So I assume the extra bit helped very slightly being the temp wasn't increasing as fast, but this was still way too hot, so I stopped the test.

I began getting worried, so I decided to put the stock HSF back in and just put the new cooler in next weekend. I cleaned off the stock HSF and CPU. I applied some of the STG-1 to the CPU. This time just a bit more than what I previously used on the Zalman. FAST FORWARD. Min idle temp 36C. When running Prime95, the increase was a lot more gradual, but I still peaked at 65C. Still too hot. Also, the stock HSF is no longer revving up to 6700rpm under load with CnQ OFF(I'll have a question on this). I began thinking maybe this "grease" needs to be applied thicker than traditional compound. I reapplied it again, this time using a bit more.

So I tried a bit more and now and have the current results. Min idle down to 35C. When I ran Prime95, the temps increased a lot slower this time. But still, the fan didn't rev up to 6700rpm max like it had originally. After 15 minutes, I was peaking 62-63C. This duplicates what I had previously experienced with everything stock.

Outside of Prime95 testing, things seemed slightly cooler than stock. Idle was down 1-2C. I reset my temps and played COD4 for a bit. Previously when playing COD4, my CPU would max out at 61C. This time it maxed out at 58C. So now my questions:

1. Question about the stock cooler: Why is AMD including a HSF with the CPU that allows the CPU to reach its upper limits in the first place? I know stock is stock, but come on. Upper limits?

2. Does the "thinner is better" apply to both compound and grease, or just compound? In my situation, this didn't apply to grease.

3. Previously my stock fan would rev up to 6700rpm, but now it doesn't. I have CnQ off. Why would just switching to grease cause this to happen?

4. Is Prime95 considered a good OVERALL load test? I mean, my gaming load temp has dropped, but Prime95 is still reaching upper limit temps. Should this program only be reserved for truly maxing out the CPU, but not necessarily be considered a good OVERALL load test? Am I giving Prime95 too much credit?

5. Should I use AS5 thermal compound the next time I install the Zalman?
 

dokk2

Distinguished
Jul 1, 2007
1,450
0
19,310



For what it is worth,, I personally would never use a Zalman cpu cooler,,because it is like a manure spreader ,in that it throws it's heat everywhere and thus might raise the overall temp of all the rest of things ,ie,,mobo,ram,hdd's.the ideal cpu cooler would push the heat to the rear where the exhaust fan will help it to exit the case,different strokes,etc..
Now then I have an AMD 7850be which I cool with an Arctic Cooler 64 Pro and I use AC5 spread as thin as possible,like a cigarette paper,my idle temps never exceed 30c or thereabouts I never check the tempa after heavy usage except to open the case and feel the exhaust from the cooler and the back of the case..:)
 

Devastator_uk

Distinguished
Jan 11, 2009
649
0
19,010
Well I have a 955BE @ stock with stock cooler and it has never ever reached 6700rpm, nor has it ever reached to 63C. I ran Prime95 the other day (room temp was 26C) for about 1 hour and the CPU got to 58C and fan got to just under 4000rpm.

Might be a case/airflow thing, I have a Sharkoon Rebel 9 Economy with two 120mm (ultra quiet) fans and one 80mm (also ultra quiet). To give you an idea, my MB temp was 37C.
 

selea

Distinguished
Aug 16, 2009
83
0
18,640


1. Because stock coolers are created just to to assure you can operate the PC at stock speeds. They are not done with the lowest temperature at the lowest rumor possible in mind or with overclocking in mind. Stock coolers are barely adeguate, they work sufficently only when not overclocking and/or if the airflow in the case is enough. If that's not the case problems arise. The adeguacy of the stock cooler also depends on the TDP of the processor. For a Core 2 Duo the stock cooler can suffice also for mild overclocking since the TDP is very low. If you have a Core i7 with a much higher TDP the stock cooler is barely adeguate also at stock speeds; you need good airflow or you will either have a jet inside the PC or too higher temps.

2. The "thinner is better" parable for thermal paste is not always true. It depends on the surfaces having contact. The less lapped the heatsink and the HIS are, the less the parable is right. If you have an heatsink that it's not lapped well or is rough (as the Thermalright 120 Ultra Extreme) you have to put more paste. The thermal compound must fill the gaps between the two metal surfaces, so it depends on how big the gaps are. In little words the bigger the gaps between the two metal surfaces the thicker the paste layer must be. If you put a too thin layer for the gaps the thermal compound will not fill them and air will flow instead, and air is a very bad thermal conductor. If the layer of the paste is too thick, however, the thermal diffusion will be lower, since the best thermal spread is between two metals. In a lapped heatsink plus lapped HIS you have to put a very thin layer. If the two surfaces are not lapped and one of the two is rough then you have to apply a thicker layer. There are two school of thoughts: the first believe that the best thermal diffusion is applied when you lap both heatsink and HIS and put a very thin layer of paste between. The second believes actually that a lapped heatsink reduces thermal diffusion and having "lines" in the base of the heatsink it's better. A thicker layer of paste is used in this case, and sometimes two layers (one in the heatsink and one in the HIS). One thing to notice: if you spread the paste to fully cover the heatsink or the HIS don't put the heatsink on top of the processor in a perfectly horizontal manner. This creates air bubbles. Instead put the heatsink at an angle and then lay it over the HIS (as you do when you apply a sticker to a surface) as not to create air bubbles.

3. Probably it doesn't go up as before because it doesn't need to. It is a bit difficult to explain but also if the temperatures are higher doesn't mean necessarily that the heatsink will let the fan go to the maximum. If it "thinks" that the fan speed is enough to keep the temps down it will not use more speed. The paste you applied makes the heatsink behave better so the fan speed goes lower.

4. Prime95 is used to test the Cores to the maximum, so it is used primarly to test the max temps to see if there are problems in overheating. Also in a stressed PC usage it's very difficult that all the cores will be in use at the same time at 100% as in Prime95, so the normal temperatures will never reach what you see when it's in use. It is a program to stress the system and it's used just for it. It's not a progam that simulates the normal PC use and the temps associated with it. It stress the system to evidence some problems of overheating in an ipotetic all-cores 100% load.

5. AS5 is a bit better than the compound given in the Zalman coolers, but the difference will be 1 or 2 degrees at most.
 

Terry1212

Distinguished
Aug 3, 2009
143
0
18,690
Thanks for the replies guys. I think AS5 would be easier to work with than STG-1. This STG-1 stuff is like nail polish. I have applied AS5 to computers I have previously built. It's much easier to work with. That alone makes me want to buy some before trying to install the Zalman again.