Overclocking Your Intel Processor--All You Need To Know

[THE JEW (RaVeN)]

INFORMATION:


Guides To Overclocking Conroe:
http://anandtech.com/cpuchipsets/intel/ ... spx?i=3184
http://anandtech.com/mb/showdoc.aspx?i=3208&p=1
http://techreport.com/articles.x/13815/1
http://anandtech.com/cpuchipsets/showdoc.aspx?i=3239

i7 Overclocking:
http://www.pcper.com/article.php?aid=63 ... ert&pid=13

LGA Pad Modding:
http://overclockers.com/articles1527/

Motherboard History:
http://www.tomshardware.com/picturestor ... adaxe.html

PCIE At X1 Bug And Fix:
http://www.techarp.com/showarticle.aspx?artno=522

Pentium M/Mobile Pentium Overclocking Database:
http://www.wwwwwoc.net/db1.htm

Xeon Database:
http://www.hardwaresecrets.com/article/277

[Tempus]

This should be included in both because there is more to overclocking then knowing your processor

Ram is important particularly if you are locked into FSB overclocking

in general, loosening your timings allows the ram to function at a higher FSB. thus if you have say a 4-4-4-12 ram bumping that to 5-5-5-15 will provide more overclocking head room

[THE JEW (RaVeN)]

True, but you also need to balance out your overall performance with your clock speed:

http://www.xtremesystems.org/forums/sho ... ostcount=2

Heres a Interesting one.

3960 mhz with cas 5-4-4 is slower the cas 3-3-3 and 3830 mhz.
weiiiiiiird.... I knew these old 975's are real sensitive to latency but this definatly had me grinning.

Here the lower overclock gave better performance and could be done with less voltage. A better solution in my book. However, he doesn't give his max 5-5-5 score so that could swing back to that solution. I guess all I'm trying to say is your "best" overclock is dependent on more than just max mhz.

Gimme your source and I'll be sure to include it.

[Tempus]

the source was actually inside of a review of some crucial ballistix ram that i was reading. The had a section on does ram matter for over clocking and thats where the chart came from

[Tempus]

Another important factor in overclocking particularly the Intel chips because the IHS is concave, is lapping. So I put together a how-to guide... because well it was more interesting then studying for midterms

So I decided to put this together since it has been a while since I have seen any how-to guides for lapping been posted. (I appologize in advance my gf has our good camera at the moment so we are stuck with 1.5 mpixel)

So first off the system specs so you know what we are talking about.

550W VX psu
pentium D 820 with a blue orb II
asus p5wd2-p
2 X 1 gig Gskill DDR2-800
8800GTS 640 meg with stock cooler




here is the orthos temp reading after just shy of an hour on both cores Forgive the fact that I am using the ASUS Probe II. Core temp didnt want to go with this processor and neither did Intel's TaT



so to begin I got some 1000 grit sand paper and taped it down to the workbench


I took staged pictures here is the cpu before lapping has begun




At this point I would like to say that unlike some people that I have read, I did not dampen the sand paper.

Also as this is a very basic attempt at summarizing the process, I would explain that the process of "lapping" at least in my case is applying slight pressure (the amount of pressure used depends on the grit of the sand paper, the lower the grit the less pressure is needed)

Now please note that by pressure I do NOT mean to push down hard. I simply mean a steady, firm grip on the edge of the processor making sure that the concave nature of the cpu IHS is slowly being worn down.

A single lap in the case of my counting is once pushing away from yourself and once coming back (I know that some people have done lapping continuously away from themselves and this is why I clarified)



after 50 laps


and the coin test after 50



As you can see, there is some reflection of the coin now which was not present before hand due to the finish on the IHS before hand.

I did not find this to be satisfactory as I could still make out the wording on the IHS which tells me that the IHS is not yet smooth.

I continued smoothing the processor for another 50 laps

CPU IHS at 100 laps



and the matching coin test



Still not satisfied with the outcome, you can still see some of the Intel writing on the chip so I decided to do 100 more laps instead of 50.

[Tempus]

CPU IHS at 200 laps



As you can see, the writing is completely gone. This is a good indication that the processor has begun to become even.
At this point I would have put a razor blade along the top of the cpu to check for perfect flatness but I found i was all out so I continued to the coin test



as you can see the coin now looks quite a bit better then the original coin test.

At this point I would have put a razor blade along the top of the cpu to check for perfect flatness but I found i was all out



At this point you have a few options

1) continue doing a few more laps. This would reduce the amount of IHS between the fan/waterblock and the cpu core itself. If you chose to do so, be warned that you have to closely monitor your progress so that you do not take off too much of the IHS. On top of providing heat dissipation, the IHS also acts as a sort of protective shield protecting the core from cracking when you apply the pressure of the fan/waterblock

2) Gently clean the IHS with either a lint free type cloth or a paper towel etc. with rubbing alcohol and reapply your choice of thermal paste (AS 5 is recommended)

3) Change the sand paper out for an even finer grade.

I opted for #3. I cleaned the IHS with rubbing alcohol and then swapped out the sandpaper for 1500 grit



I continued to lap the IHS with this very fine grit until I got it nice and shiney. This serves a few purposes, though mostly aesthetic, it helps to remove any small grooves you may have accidently put in the IHS. At this point these grooves would not have affected overall heat dissipation to a great extent so this step is not needed. However I felt that after putting in this much effort why not?

And finally the coin test



What a difference! The cpu IHS is almost mirror like.

The final step is to reapply the Arctic Silver 5 and put the chip back in the machine

This next part I throw in for the sake of completeness as, if you are taking a shot at lapping, you probably already know how to properly apply thermal paste.

For this step get either a sandwich bag or plastic wrap and put it around your finger. Apply only the smallest amount of AS5 to the IHS and spread out a very thin layer.




The last step is to check to see if anything actually made a difference.


as you can see, the times being relatively equal I dropped about 7C off the overall temps. This probably has something to do with the Blue Orb and something to do with the fact that I only have 1 7V delta fan on the plexy glass side blowing in so there is relatively no air flow in the case.

I hope that beginners find this helpful. I scoured the forums and the internet and this is the cumaltive result of many different posts.
I always found the guides with pictures more helpful which is what prompted this post.

[THE JEW (RaVeN)]

Excellent guide! The pics really made it. They really only wet stand stuff for 2 reasons that I know of:

1. prevent excessive heating when sanding to get a flat surface quicker
2. cuts down on dust

For your flatness test, there's always this from the AMD Overclocking Sticky:

Flatness--A Poor Man's Guide:
http://overclockers.com/tips1224/
http://raven.kalshadowtech.com/webstuff ... ngGrid.pdf