Stupidity or Genius?
I'm not much of an over-clocker but I have knowledge of it and have dabbled before.
My friend, who works for a well-known hardware retailer specialising in over-clocked systems, has said that I cannot possibly get more processing power out of a processor than he can. In fact, he went so far as to mock me.
Gauntlets were thrown, various words were exchanged, and there is now a bet on, for £1000, for whoever gets the highest average clock speed.
Now this is where I need some advice. My plan is to lap the processor heat exchange panel to a mirror finish, then polish it. After that, I intend to solder a lapped and polished water block onto the CPU. The reservoir and piping will be running through an old refrigeration unit from a commercial fridge.
You're probably thinking, "Yeah, do it, it might be fun to see the results," however, the chip is an i7 965 Extreme.
So the question is: Is the soldering going to turn this very, very expensive chip into a very, very expensive paper weight? I am thinking about using silver dust and flux as the solder.
The only way you'll get a good solder joint is to heat both the heatsink and the IHS (assuming you're still using it somehow) to the point the solder will melt.
If you can pull that part off without damaging either component, then it may work, but I don't see the results being all too spectacular unless you pull the job off perfectly.
It may be wiser to just lap both sides as you've suggested and apply as thin (and even) a layer of a good thermal paste as possible.
I'm not certain that the CPU is going to heat up the solder to melting point. If I merely use something like Arctic Silver then the heat transfer is not going to be efficient enough.
Does anyone know of a good mixture for a lower melting point solder?
Really? Enthusiasts have been cooling systems with refrigerants and liquid nitrogen for years now. Do you think they've been soldering chips to their blocks?
Originally Posted by Winters
The entire purpose of a thermal paste is fill in the minute gaps between a heatsink and the chip itself with something that offers as high a thermal conductivity as possible. Have you checked the thermal conductivity specs of compounds such as AS5 or AS Ceramique vs silver-based solders?
Also, how do you intend to get the solder to actually melt between your IHS and heatsink correctly?
Well, the idea was to run the CPU for a couple of minutes without the heatsink, then use a solder gun to apply the solder but as I said, I am not certain that the CPU will take the heat.
Liquid Nitrogen won't work as one of the rules is to have the case sealed without any external components. The cooling solution also has to be permanent, so a container of liquid inside does not pass the rule.
I have looked into conducivity and I know that something such as silver or copper with complete surface area will conduct far more efficiently that a paste.
Another option would be to submerge the components in oil. What do you think?
July 16th, 2009, 08:09 AM
A tin, silver and lead based flux will destroy your CPU. F*** it.
July 16th, 2009, 10:22 AM
As an avid OCer, having built over 300 systems ranging from high end water setups to phase change, and watching what others have done, I can tell you that Arctic Silver is more than enough for what you want to do; and yes, the conductivity will pose some HUGE issues. Especially because you run the risk of melting the heat spreader(if it has one.). Main reason it wont work: You will be soldering around the heat spreader. Even if it's a dual core; you will be soldering around a part that doesn't see much heat. The center of the chip is where all the important stuff is going to be heat wise, and even if you use AS5, if it's not properly applied you can get worse temps than before.
As for overclocking results; It varies so much it's not even funny. You would need IDENTICAL setups, and in the end it comes down to the CPU build(C0 chips in general are getting pretty good results), your motherboard(Asus Rampage II is a serious OCing board), RAM(use crappy RAM get crappy results), PSU, OS(Windows 7 doesn't enjoy OCing all that much), Cooling(The cooler you can keep the chip the more voltage you can safely run(hence LN2 setups)), and lastly, your own judgment. Would you rather have a higher overall CPU speed but sacrifice the speed of your RAM? Or do you want a system that will perform well in every aspect and not be a benchmark queen? If you want some serious results though, water will be the way to go. Run a T-line and you wont need some huge resi inside the case which will save on space and you can get a pretty serious CPU cooling setup for around $200.
July 16th, 2009, 11:46 AM
Thanks for the advice, Cerberus.
The i965 is dead. Very, very dead. I removed the heat spreader to apply the solder and the heat killed it. I am going to try and get the solder off it and send it back as faulty.
With regards to your suggestions. If I provide you with the exact system specs and cooling solution. Would you be able to provide me with some rough clocking specs please?
July 16th, 2009, 01:07 PM
Might be able to get you some ranges and I know a few guys who have done a dozen or so LN2 runs on the EE i7s, so they know the ins and outs of that chip as well as one of the engineers who designed the chip. As for the RMA, they might not help you there. "Burn and return" as we call it is highly frowned upon as in the end; just raises prices and in this hobby you just gotta know the risks. I have about 12 mobos and chips I've killed over the years hanging in the garage at my parents.
Originally Posted by Winters