Thoughts on first overclock

[known12345]

Overclocked my q6600 over the weekend to 3.0mhz at 1.2875v (BIOS number; lowest value I can put for my cpu voltage on my ABIT IP35-E) and ran successfully on prime95 for 8hours. Temps for each core were about 45-50C using an Arctic Cooling Freezer Pro 7 in a Antec P182 case with 3 fans on low.

Then overclocked my q6600 to 3.4mhz at 1.4275v (BIOS number; overestimated my core voltage as wanted to complete prime95 for 8 hours with no errors then decrease temp down from there). Temps each core were about 50-60C. PWM temp was pretty high unfortunately (can't remember exactly but I remember I was worried about it).

Finally, overclocked my q6600 to 3.6mhz at 1.52something v (BIOS number, pretty high I know but again wanted to ensure enough power was reaching the core though heat became a huge issue this time). Temps for each core were about 60-70C (very high I know) and PWM temp was extremely high, at 100-103C. At this point I tried whatever I could to decrease the temperature. I did not understand what PWM cooling was exactly but I remember someone telling me that the ABIT IP35 Pro has better PWM cooling then then ABIT IP35-E so I figured it had to do something with the motherboard (not the cpu which I figured I had somewhat covered). So I removed both the side panels of my P182, turned all 3 fans on high and put my 8800gt fan on 100%. As a result, the GPU temps dropped, and the core temps slightly decreased a few degrees C. Unfortunately the PWM temperature remained over 100C and did not drop more then a degree or two.

At that point I figured I should discontinue my efforts (I set the BIOS core voltage lower but it just resulted in an error in prime95 within a few minutes) but i realized my roommate had a small fan so I directed the fan (on high) toward my motherboard and my PWM temps started dropping to low 90sC. I then tried to angle the fan so that the air would hit the top section of my motherboard (the side of the CPU cooler) and blow the hot air toward the top fan of my P182 case. PWM temps then dropped to mid 80sC and leveled off at about 85C. This was around 10pm in San Diego so it wasn't exactly hot or cold. I then ran prime95 and kept it on for 10 hours. About 4 hours into the test the temps hit the low 90sC but when I woke up at around 7am the temps were at 85sC and when I checked back at 9ish am the temps were high 80sC with no errors.

From my first experienced I realized that in order to overclock higher (which I might consider) I have to get my PWM temperatures down and possibly my CPU temperatures yet I dont exactly understand how to cool down my PWM temperatures. I found this post by BillParrish that kind of clarified PWM:

It is a sensor that monitors the temp/health of the CPU Voltage Regulation Components around the cpu that are responsible for taking the 12V from the aux 12V connector and giving the CPU the voltage and power it needs. This circuity uses PWM type of switching to generate the voltages needed.

This has become important as a lot of big tower coolers and water cooling have eliminated the "blow by" airflow the downward facing Intel stock fan provides to these components surrounding the CPU socket. Quad cores also place large current demands on this circuitry and both of these reasons are why you now see heatpipes and heatsinks in/on the components aroud the cpu socket on high end boards.

From this statement I understood that the stock intel fan for the q6600 was better at cooling PWM temps then some of the bit tower coolers because the stock intel fan directed its cooling directly at the cpu which consequently not only cooled the cpu but the surrounding components. The Arctic Cooling Freezer Pro 7 that I use on the other hand has its fan directed toward the big heat sink attached to the cpu and blows air toward the top fan of my case. Now directing my roommates fan at my motherboard (specifically the upper portion of my board) was able to cool my PWM temperature and so I was wondering if someone would know if blowing a 120mm fan or two towards the CPU would possibly decrease the PWM temperatures in much the same way or is there a more efficient/better method to decrease PWM temperatures?

If its possible, I would REALLY like to avoid purchasing a new motherboard or a better cpu cooler. And a 120mm fan seems like a really cost effective method for doing so. I am currently living on a campus so power usage (right now) isnt really a problem as I dont pay my electricity bill. I only want to try to get my cpu to run at 3.6 for 8 hours on prime95 with both case panels closed and not using my roommates fan. Essentially I want to "prove" that I can run my cpu at 3.6 with both side panels closed. I mean I was already able to run at 3.6 but I felt that opening my case fans and putting a fan toward my motherboard was "cheating" the overclocking process. There really is no reason for me to overclock as once im done im going to restore everything back to factory settings (no need for the extra speed) but the feeling of "getting yours money's worth" is so strong for me in the process of overclocking that I just want to see how far I can go (once I hit 3.6 I think ill be good and content, 50% increase is a good number in my view).

 

[BillParrish]

If you look closely at the lower fins of the AF7 you will see a couple of the bottom ones are turned down to try to put some air on those critical components. That is good but frequently there are also PWM voltage regulator components off to the side as well, so placing a small fan or two blowing air in the area between the CPU and the I/O plate and the CPU and the top of the board (assuming "normal" mounting in a tower type case and if there are components there, look for the black squares and caps in a row) is always a great idea as your testing showed. With a quad and its high(er) current demands it is an even better idea for stability and long board life. Anything will do, I buy cheap HD bay coolers and rip the fans out because I can get 2 or 3 fans for $4.99 instead of paying $3.99 each and shipping on each.

Does not have to be fancy. a 120 hung anywhere you can get it to blow on that area would be great.

There is also a thing called a "spot fan" a fan on a flexible goose neck kinda thing but I cannot remember who makes it.

Practice what I preach:
http://mysite.verizon.net/ressdxka/sitebuildercontent/sitebuilderpictures/computer003.jpg

There is no "cheating" you do what you need to do to keep temps under control for high perfromance AND long life and reliability, thats just good engineering and it does not have to look pretty if it works.

 

[known12345]

Thank you for the response BillParrish and ive definitely tried to find the PWM voltage regulator components but I am not too sure if I have. Unless I am reading your helpful post incorrectly, the PWM voltage regulator components should be a row of black squares or something of the sort. I think I have found them but they are far away from the cpu, in fact my RAM separates the "PWM voltage regulator components" from the CPU. According to your post it probably should be close to the cpu. Here is a layout (from ABIT) of my bored:

http://www2.abit.com.tw/page/en/mot...ODEL_NAME=IP35-E&fMTYPE=LGA775&pPRODINFO=BIOS

If you look to the lower right hand side, I would assume those were the PWM voltage regulator components because there is a row of black squares and more importantly, when I took out my bottom P182 case fan out and had it face the area im talking about (with the fan directed at the RAM), my PWM temperatures dropped though only a few degrees (1-4C). From from the 5-15C dropped yesterday when I applied my roommates fan (not a comp fan but like a fan to cool you off when its hot; iono if I made clear earlier). Because it dropped only slightly in temperature, I was thinking that I was simply cooling the motherboard itself or the RAM and not the PWM voltage regulator components, is this the case?

Can someone please take a look at the picture in the link and please help me clarify where the PWM voltage regulator components are? I would assume, from the info ive learned from BillParrish's posts, that they would be close to the cpu yet by looking at the picture and at my bored, I do not see a "row of black squares." However I do see black squares so possibly it is the area. I can't test the area with a cooler right now as it is 4am in the morning here and I need a piece of string (which I dont have) to hold the fan more toward the cpu and my AF7 seems to be in the way of everything. Though tomorrow I will attempt to see if cooling around the CPU as opposed to the area near the RAM would drop the PWM temps significantly and clarify where the PWM voltage regulator components are.