Can't Believe What a Difference Lapping Made!

[Elfear]

I decided to try my hand at lapping a couple days ago after I saw horrible temps on my QX6700 with a decent watercooling setup. I was pretty nervous about lapping the most expensive chip I've ever owned but there are plenty of good guides online that simplified things a lot.

I also switched from a Swiftech MCW6002-64 waterblock to one of the new D-Tek Fuzions, so the temps aren't directly comparable but should be within a few degrees. Here are the initial results at stock speeds and volts:

With unlapped QX6700 and MCW6002 block

Idle 47-51C (temps varied a lot between cores)

Load (only one instance of Orthos for ~30sec) temps shot up to 90C very quickly


With lapped QX6700 and D-Tek Fuzion

Idle 30C

Load (two instance of Orthos running for 30min) 46C


I attribute 95% of the difference to the lapping. When I pulled off the MCW6002 there was only a very thin line of contact right along the edge of the chip. The IHS was horrible and was no where near flat. After lapping the contact was awesome. Almost the entire IHS made contact with the waterblock.

I don't recommend this for everyone (those who will be running at stock speeds and don't want to void the warranty), I just wanted to throw in my two bits about the effectiveness of lapping.

 

[Avalon]

That's nuts. It seems like when I finally get my Core 2 chip I'm going to pull the HSF back up after plopping it down on my chip to see if it was making good contact...otherwise...I may be tempted to try this.

 

[bobsmith1492]

Uh, I think your initial setup had some problems like not having good contact with the heatsink... 90C sounds ridiculous. I say you should attribute 95% of the increase to reseating your CPU and 5% to lapping. Lapping was not the only variable there.

 

[Elfear]

Originally posted by: bobsmith1492
Uh, I think your initial setup had some problems like not having good contact with the heatsink... 90C sounds ridiculous. I say you should attribute 95% of the increase to reseating your CPU and 5% to lapping. Lapping was not the only variable there.

I reseated the waterblock a couple times. I'm no guru by any means but I have properly seated a heatsink or two. The reason I credit lapping with the huge drops in temps is because of the pattern of the Ceramique when I pulled the MCW6002 off. The only contact the waterblock had made was a razor thin line around the outside of most of the chip. I should have taken a picture because it was almost ludicrous how concave the IHS was. Even if I had tightened one side down a little too much I would have seen more of a contact patch than just the outside of the chip.

 

[Duvie]

What speeds have you hit on that? I am interested if you would list yourt speed and the actual vcore under load. Also ambient temp and then coretemp temps....I would like to see what your coefficient is with that system...

we can also by looking at the deltas find the point at which it goes above its rated heat dissapation wattage...

 

[Elfear]

Originally posted by: Duvie
What speeds have you hit on that? I am interested if you would list yourt speed and the actual vcore under load. Also ambient temp and then coretemp temps....I would like to see what your coefficient is with that system...

we can also by looking at the deltas find the point at which it goes above its rated heat dissapation wattage...

I'll get some numbers sometime this week for you. I didn't want to start overclocking until I figured out the temp problem.

Ambient was about 70F or 21C.

 

[Duvie]

Originally posted by: Elfear

Originally posted by: Duvie
What speeds have you hit on that? I am interested if you would list yourt speed and the actual vcore under load. Also ambient temp and then coretemp temps....I would like to see what your coefficient is with that system...

we can also by looking at the deltas find the point at which it goes above its rated heat dissapation wattage...

I'll get some numbers sometime this week for you. I didn't want to start overclocking until I figured out the temp problem.

Ambient was about 70F or 21C.

well we are about identical there....

the key is lower the vcore. You have a retail and perhaps you will be able to hit my speeds at lower vcore and thus you will have less heat to dissapate. That will be the only hop of keeping a standard H2O system close to my numbers....

MOst of the testing on these water coolers are done with dual core chips and are usually tested at 100watts....That is a C2D at about 3.2ghz with 1.375v...Not really extreme but the coefficent (C/W) is ually .11 for the good swiftech stuff...meaning your load at 100watts would allow your temp to be only 11c higher then ambient air just outside of intake fans. That is what they say though I am skeptical they are using an internal die temp like that of the C2D and may be taking calibrated Tcase temps...which in most case are running 10-20c lower the what coretemp reports...

Asus is the biggest crack smokers when it comes to this...PCprobe is off24c to core temp for me though on my Gigabyte boards it is 10-13c depending on the board. Asus has always been the lowest temp reporters for years with their boards.

 

[zephyrprime]

Originally posted by: bobsmith1492
Uh, I think your initial setup had some problems like not having good contact with the heatsink... 90C sounds ridiculous. I say you should attribute 95% of the increase to reseating your CPU and 5% to lapping. Lapping was not the only variable there.

Yeah, 90C is ridiculous. Even the stock air cooler can do a lot better than that.

 

[StopSign]

Originally posted by: Duvie
Asus is the biggest crack smokers when it comes to this...PCprobe is off24c to core temp for me though on my Gigabyte boards it is 10-13c depending on the board. Asus has always been the lowest temp reporters for years with their boards.

It helps with their sales.

 

[bobsmith1492]

Originally posted by: Elfear

Originally posted by: bobsmith1492
Uh, I think your initial setup had some problems like not having good contact with the heatsink... 90C sounds ridiculous. I say you should attribute 95% of the increase to reseating your CPU and 5% to lapping. Lapping was not the only variable there.

I reseated the waterblock a couple times. I'm no guru by any means but I have properly seated a heatsink or two. The reason I credit lapping with the huge drops in temps is because of the pattern of the Ceramique when I pulled the MCW6002 off. The only contact the waterblock had made was a razor thin line around the outside of most of the chip. I should have taken a picture because it was almost ludicrous how concave the IHS was. Even if I had tightened one side down a little too much I would have seen more of a contact patch than just the outside of the chip.

Ok, I'll give you that. Maybe you needed a lil' more paste?

Did you lap the waterblock as well?

 

[Noubourne]

Originally posted by: Duvie

Originally posted by: Elfear

Originally posted by: Duvie
What speeds have you hit on that? I am interested if you would list yourt speed and the actual vcore under load. Also ambient temp and then coretemp temps....I would like to see what your coefficient is with that system...

we can also by looking at the deltas find the point at which it goes above its rated heat dissapation wattage...

I'll get some numbers sometime this week for you. I didn't want to start overclocking until I figured out the temp problem.

Ambient was about 70F or 21C.

well we are about identical there....

the key is lower the vcore. You have a retail and perhaps you will be able to hit my speeds at lower vcore and thus you will have less heat to dissapate. That will be the only hop of keeping a standard H2O system close to my numbers....

I'd be surprised to find out that was true.

Your TECs are adding more heat to be dissipated on top of the chip itself, and as proven elsewhere in TEC testing, selecting a highly undervolted TEC (or in the case of the Freezone, six of them) to cool a high end CPU often results in zero extra cooling performance.

A well configured water kit for $250 or less is highly likely to match the Freezone in performance based on what I have seen. The key is that water is so highly effective at heat dissipation that is capable of handling a much, much higher heatload than what a QX6700, even overclocked to very high levels, is able to put out. This is evidenced by the fact that water is often sufficient to cool not only highly overclocked CPUS, GPUS, and anything else that can be thrown into the loop, but on top of that it is also capable of handling very high powered TECs, sporting hundreds of watts of power (well over 400).

If your TECs are undervolted, as I suspect they are, then they are likely to be doing very little to actually cool the CPU above what a decent water kit could do, except maybe suck a bunch more power than a water kit would normally need.

Your only advantage is ease of setup. Needless to say, at similar price points, I too am interested to see his results.

I have oft been convinced that the Storm or MP-05 Pro at $75+ are the only viable options for high-performance waterblocks, so it will be interesting to see what a sub-$50 block is capable of on the QX6700. Paired with the top performing PA 120.3 rad for $125, a high end $80 pump, and some tubes and clamps, it seems to be a very viable option in comparison. Of course, a $50 120x2 radiator is likely to provide plenty of cooling power without the price premium for the best gear around. Nevermind the ability to add blocks to cool other components in the future...

 

[Elfear]

Originally posted by: bobsmith1492
Ok, I'll give you that. Maybe you needed a lil' more paste?

Did you lap the waterblock as well?

I'm usually pretty careful with the amount of paste I put on. I put a BB size dab in the middle of the IHS and spread it evenly with a flat razor. I make it so it is just opaque all across the IHS. More paste would have helped but only because the IHS was so concave. Temps would have been less than ideal with the big goober of paste.

I didn't touch the waterblock. I figured D-Tek had done a decent job with that.

Originally posted by: zephyrprime
Yeah, 90C is ridiculous. Even the stock air cooler can do a lot better than that.

The stock heatsink would have been screaming for mercy unless it has a smaller base that would have fit inside the raised sides of the IHS.