Lapped IHS or Stock IHS poll?

[Martinm210]

It appears from some recent testing that you basically aquire the same net benefit from bowing by simply lapping your IHS.

I've got a brand spanking new Q6600 sitting here calling to me for some thermal testing, but the question I'm struggling with is.

Lap or don't lap for testing..:shrug:

If it was just for my own use, I would lap it without question. I don't care about losing the warranty, I'm going to Overclock the crap out of it anyway, but I want the test results to match what most people would experience, so I'm questioning lapping.

If you don't lap the IHS it will show benefits to blocks with bowing.

If you lap the IHS it will show maximum performance from all blocks regardless of bowing or very minimal differences.

[oerekum]

I say lap it. A flat block on a concave IHS suffers more than a bowed block on a lapped cpu. I think..

[Martinm210]

Great!

I'm liking this so far, getting the 1000 grit ready...:)

[oerekum]

Lapping is fun indeed. There's something about rubbing tiny really expensive things on sandpaper that gets me all tingly.

[adpr_02]

Lap it for sure, but U may need more than 1000. I used 1000 on mine and it didn't come out as mirror-finish as I would have liked it, even after sanding for pretty much 1 hour with multiple sheets of 1000...

Only reason I left it like that was because I couldn't find anything higher than 1000 at the local canadian tire:down:.

[Martinm210]

Yeah, I've got 1500 grit too, although I don't think there is anything to gain beyond 800 grit. The real gain there is making it flat and true.

Cleaning off enough TIM after time will actually polish it too.

Nothing like shiny copper...:)
http://www.overclock.net/gallery/data/500/LappedE6600.jpg

[mcoffey]

Depends on the CPU. A 200.00 Q6600 sure, what the heck. A 1400.00 QX9770, not hardly just in case I have to return it. And to be honest, with the attention being paid to flatness of the IHS now on newer CPU's, not needed. The last Q9300, QX9650 and QX9770 I just got was flat enough where it wasn't needed IMO. And with the clearence between WC CPU blocks and the CPU hold down bracket being so close, I don't need to file off another .25-.50 mm closer.

So I voted no. Good poll and topic for discusion,

andyc

[003]

You won't see any temp difference if you lap with 400 grit vs 2500 grit. The temp difference does not come from a mirror finish, it comes from making the convex IHS flat. Mirror finish is purely aesthetic.

[adpr_02]

Wow, now THAT is nice! Good job, how long did it take you?

Have you tested that quad core nozzle on the Fuzion vs. stock yet? I can't wait to get my hands on that V2.

[أشرف]

For thermal testing? And you have only one Q6600? I voted no. The majority of people don't have lapped CPUs (or don't want to lap their CPUs) compared to a few people that lapped their CPUs. :shrug:

[Martinm210]

For thermal testing? And you have only one Q6600? I voted no. The majority of people don't have lapped CPUs (or don't want to lap their CPUs) compared to a few people that lapped their CPUs. :shrug:

That's why I hesitated, I think most people would run a stock IHS although it's really not a hard thing to do.

I'm wondering since the sample variation changes if you had a really concaved IHS, wouldn't leaving it stock give preference to the bowed block unfairly if the majority of the IHS samples are actually fairly flat?

I was also thinking that testing a block on a truely flat surface is better as it removes any manufacturing error of the IHS itself and it would also maximize consistency in mounting. It appears the mounting variance is the biggest hurdle to good testing, so anything that makes that more consistent is a good thing.

[Boogerlad]

lapping it beyond 600 grit does nothing. Infact, polish makes it worse because the polish fills the microscopic grooves and replaces the tim with lower thermally conductive polish.

[Nanometer]

hard to say. Most people will be using the stock spreader application. But it would be nice to see the full potential of the blocks. I say both ;)

[nettwerk2006]

I lapped my Q6700 and got a decrease in temperatures on all 4 cores... 2 of the cores had a decrease from 70 to 50 at load because how bad the IHS was with my TRUE.

Best $20 I ever spent, and it really is easy and takes only an hour to do.

[Martinm210]

It's looking like my IHS is more convex than anything, so it'll probably do fine with flat blocks stock. I might just have to mount a few times and see if I'm getting a good TIM spread.

[Sparda]

Umm...did I the only one saw those in the circle. Usually 400 grit should make it dissapeared before going upper grit. Maybe you lapped the IHS in a hurry...:).Though maybe wont effect the temp since both core are at the middle and it already flat.

I vote yes. I think majority people at intel section that push and priming the systems for hours and want 24/7 settings bother to lapped their cpu.

I seriously has temp drop just by lapping Q6600. (400, 800, 1000, 1500, 2000, 2500 and no polish). I probabaly skip a grit between 1000-2500.

78797

[biftek.]

Umm...did I the only one saw those in the circle. Usually 400 grit should make it dissapeared before going upper grit. Maybe you lapped the IHS in a hurry...:).Though maybe wont effect the temp since both core are at the middle and it already flat.
78797

Shouldn't be too much of an issue since he uses spring mounting. if it were being air cooled and hard mounted I would say that's a problem, but springs tend to even everything out in the end.

[ZOMGVTEK]

The issue here for thermal testing would be that not every intel IHS is the same, some are evenly concave, some are convex, and mine was a little bit of both in different parts, wich lead to massive temp drops. I would say that the BEST thermal testing would be on a lapped IHS. Since your are only going to be testing on one processor, you couldnt assume yours is like everyone elses. Your supposed to be testing the waterblocks performance under ideal conditions, right?

[MpG]

I don't think we're looking for 'ideal' conditions as much as 'repeatable' and 'normal' conditions. Since no two IHS's are truly identical, a block that works good on one processor could theoretically be quite mediocre on another, even if they're technically the same processor. Lapping takes a variable out of the testing process, since a block that works well on one lapped CPU is more likely to work well on another lapped CPU. Granted, there are still other factors at play, but anytime you can take a variable out of the equation easily, it's probably worth doing so.

In all fairness, I understand that there are a lot of people who would like to know the best possible waterblock to use on their non-lapped CPU. But the strongest contenders these days are ultimately all within a few degrees of each other. And if you're that concerned about an extra degree or two... break out that sandpaper! :D

[Martinm210]

Umm...did I the only one saw those in the circle. Usually 400 grit should make it dissapeared before going upper grit. Maybe you lapped the IHS in a hurry...:).Though maybe wont effect the temp since both core are at the middle and it already flat.

I vote yes. I think majority people at intel section that push and priming the systems for hours and want 24/7 settings bother to lapped their cpu.

I seriously has temp drop just by lapping Q6600. (400, 800, 1000, 1500, 2000, 2500 and no polish). I probabaly skip a grit between 1000-2500.

78797

The picture I posted above was my first time lapping my E6600, it has since then been laped well beyond that so all the low spots were removed, although it shouldn't make any difference at the perimeter.

I don't think we're looking for 'ideal' conditions as much as 'repeatable' and 'normal' conditions. Since no two IHS's are truly identical, a block that works good on one processor could theoretically be quite mediocre on another, even if they're technically the same processor. Lapping takes a variable out of the testing process, since a block that works well on one lapped CPU is more likely to work well on another lapped CPU. Granted, there are still other factors at play, but anytime you can take a variable out of the equation easily, it's probably worth doing so.

In all fairness, I understand that there are a lot of people who would like to know the best possible waterblock to use on their non-lapped CPU. But the strongest contenders these days are ultimately all within a few degrees of each other. And if you're that concerned about an extra degree or two... break out that sandpaper! :D

My thoughts exactly...:up:

Great discussion, and it's looking like the majority say lapped, and I really like the idea of testing on a truely flat surface the same that anyone else can get just by lapping.

Thanks!:clap:

I better take some pictures of this poor stock IHS, I'm going hunting for copper...:D

[71 (Bryan)]

i say lap,
but one thing that i was thinking about was is you have it flat but leave it a bit rough, wouldnt it perform better since you have more surface area?

[RealRedRaider]

Lap, please....

[smee]

i say lap,
but one thing that i was thinking about was is you have it flat but leave it a bit rough, wouldnt it perform better since you have more surface area?

+1 on lapping :up:

As others have said, it's not necessary to go further than 800 grit, stopping at 800 grit leaves it a tad rough, but not any kind of shine to it.
Here is my 800 grit lapped QX6700 ES:

http://i254.photobucket.com/albums/hh106/parkman14/578d21a9.jpg?t=1211034579

http://i254.photobucket.com/albums/hh106/parkman14/0f33bd37-1.jpg?t=1211034567

Only took about 30 minutes to do on a 1/2" piece of tempered glass and some sand paper. But best of all, It performs great! :up:

[Repliquant67]

Some results (http://www.cooling-masters.com/articles-45-7.html)

I preffer to keep the waranty of my qx9650 box and have 1-2° more :p:

[Martinm210]

Some results (http://www.cooling-masters.com/articles-45-7.html)

I preffer to keep the guarantee of my qx9650 box and have 1-2° more :p:

That's an awesome article, showing exactly that.

The bowed Apogee still gained a very small amount from the lapping, but minimal compared to the non lapped concaved IHS. More importantly it shows you there was no downside to lapping, it was a performance advantage for all blocks.

The guys at cooling-masters are awesome, their articles are always full of great details.

http://www.cooling-masters.com/images/articles/apogeegt/images/fullc2d.png

[mcoffey]

That's an awesome article, showing exactly that.

The bowed Apogee still gained a very small amount from the lapping, but minimal compared to the non lapped concaved IHS. More importantly it shows you there was no downside to lapping, it was a performance advantage for all blocks.

The guys at cooling-masters are awesome, their articles are always full of great details.

http://www.cooling-masters.com/images/articles/apogeegt/images/fullc2d.png

So with a flat block like the AC Cuplex DI, you'd see the most improvement by lapping, correct? As opposed to a ultra bowed block like the EK Supreme? Also, would lapping increase or decrease mounting consistantcy on a bowed block?

andyc

[nikhsub1]

My thoughts are this, lapping will SOMETIMES yield you nice results, SOMETIMES it won't. It depends on how your IHS started out in the first place. If your CPU unlapped is as flat as it would be lapped, I would say the gain would be nothing. If your IHS is whack, the gains will be very nice. While David (Rosco here) is one of the smartest guys I know regarding thermal testing etc, I think he would agree with what I am saying. If you took 100 CPU's and tested each one lapped and then unlapped the results would be non uniform from CPU to CPU. Unless Intel has gotten very stringent on their IHS specs, this will continue to be the case.

My test CPU is UNLAPPED but is very flat. If I were to lap it, all prior results would be meaningless to newer results and this is why I do not lap it. FWIW, Quad core IHS seem to be way more whack than dual core for some reason

[mcoffey]

My thoughts are this, lapping will SOMETIMES yield you nice results, SOMETIMES it won't. It depends on how your IHS started out in the first place. If your CPU unlapped is as flat as it would be lapped, I would say the gain would be nothing. If your IHS is whack, the gains will be very nice. While David (Rosco here) is one of the smartest guys I know regarding thermal testing etc, I think he would agree with what I am saying. If you took 100 CPU's and tested each one lapped and then unlapped the results would be non uniform from CPU to CPU. Unless Intel has gotten very stringent on their IHS specs, this will continue to be the case.

My test CPU is UNLAPPED but is very flat. If I were to lap it, all prior results would be meaningless to newer results and this is why I do not lap it. FWIW, Quad core IHS seem to be way more whack than dual core for some reason

Good stuff,

From the last 3 45nm quads I've gotten, the IHS seems much flater than the older Q6600's. I can't see a gap when placing a straight edge across the IHS on the newer 45nm quads. Maybe I got lucky, not sure.:shrug:

andyc

[Martinm210]

So with a flat block like the AC Cuplex DI, you'd see the most improvement by lapping, correct? As opposed to a ultra bowed block like the EK Supreme? Also, would lapping increase or decrease mounting consistantcy on a bowed block?

andyc

I think it's like Scott says, it all really depends what you get for an IHS. If it's wavy or concaved, you'll gain a good amount from lapping it, similar to bowing, but if it's convex then you may not see much of anything.

It probably also depends on the type of block. Something with very focus and isolated cooling in the center may loose more performance by poor center contact, than a block that has a larger cooling area.

It looks like I got lucky with my quad, it's actually fairly convex:

I took out one of my precision ground parallels for the milling machine and snapped a few pictures.

http://img177.imageshack.us/img177/6190/q6600ihsrq8.jpg

[nikhsub1]

yeah, must quads are totally rippled. The edges start out high, then there seems to be a mote about the cores and then at the cores it is high again. At least this is my experience with Quad IHS's. Like I said, dual core IHS's seem to be in much better shape in terms of being flat.

[the monk]

you fellas got to understand that production stampings are crap......production anything is crap.....compaired to Accurate machining tolerancing....

[rosco]

My thoughts are this, lapping will SOMETIMES yield you nice results, SOMETIMES it won't. It depends on how your IHS started out in the first place. If your CPU unlapped is as flat as it would be lapped, I would say the gain would be nothing. If your IHS is whack, the gains will be very nice. While David (Rosco here) is one of the smartest guys I know regarding thermal testing etc, I think he would agree with what I am saying.

Sure, I agree.

Only one answer, a Norman one : that depends :D. Some IHS are fairly "good", but most of them are quite bad (when there's a 0.1 mm difference measured between edges and center with a comparator, results are truly disastrous using a flat WB/rad...). Dual dies processor are a bit less hit because the IHS is more sustained and less prone to warp, I saw that too. It's not so easy to make the package flat at reasonable cost due to the process, I already talked about that with an Intel thermal engineer saying their IHS were a shame. Don't talk about lapping IHS at the end of their process, they don't like it ;)

Convex IHS is a better option than concav IHS anyway, at least the WB touches it near the center where max heat flux is ! It's a "reversed bowed" situation.

That's a real problem when we have to test some parts. What situation to choose ? The most common ? But on what side ? Guys interested by cooling systems, checking on forums with no fear to watercool/lap CPU/etc. OR the mainstream people who don't know much about that and keep the IHS stock + Intel radbox ? The flat IHS giving low variance to achieve good repeatability ? A convex IHS ? A wavy one ? Each possibility could be defended in a way or another.

Results are dependant of this choice, but unfortunately we can't test all of them, it's very time consuming... It's a pity because cooling products are originally made to be used on a flat surface (or they have a stepped base to avoid IHS edges), otherwise why spend time to flatten and polish them ? There's a severe discrepancy unsolvable and we have to accept a certain variance. What we need is only good trends, not absolute ranking because that's not universally achievable anyway, forget the 0.01 °C ranking, that's BS.

[mcoffey]

I think it's like Scott says, it all really depends what you get for an IHS. If it's wavy or concaved, you'll gain a good amount from lapping it, similar to bowing, but if it's convex then you may not see much of anything.

It probably also depends on the type of block. Something with very focus and isolated cooling in the center may loose more performance by poor center contact, than a block that has a larger cooling area.

It looks like I got lucky with my quad, it's actually fairly convex:

I took out one of my precision ground parallels for the milling machine and snapped a few pictures.



Thanks man, one more thing.

When you mount your block, is that with or without the CPU holdown bracket attached? I've always wondered if you could get away with the CPU block providing the presure to seat the CPU instead of the holdown bracket. From what I've noticed, the tolerance is very close sometimes, and seems like the bracket intefers with a good clean mount. I've noticed some dings and scratches out towards the edges of some blocks, which leads me to believe the block bottoms out on the bracket.

What your thoughts?

andyc

[nikhsub1]

This is precisely why I think that a FLAT IHS and a FLAT stepped block is the ideal situation. You have flat to flat and pressure where it's needed and no where else. It's a win/win for everyone.

[mcoffey]

This is precisely why I think that a FLAT IHS and a FLAT stepped block is the ideal situation. You have flat to flat and pressure where it's needed and no where else. It's a win/win for everyone.

Yup,

I remember thinking that when you came up with the concept. Whatever happened to that idea? Did anyone ever come up with block based on that design. I remember seeing OCZ talking about doing one, but I haven't seen it get off the ground yet.

andyc

[Martinm210]

Thanks man, one more thing.

When you mount your block, is that with or without the CPU holdown bracket attached? I've always wondered if you could get away with the CPU block providing the presure to seat the CPU instead of the holdown bracket. From what I've noticed, the tolerance is very close sometimes, and seems like the bracket intefers with a good clean mount. I've noticed some dings and scratches out towards the edges of some blocks, which leads me to believe the block bottoms out on the bracket.

What your thoughts?

andyc

You can run without the retention bracket, and I did on my E6600 for quite a while. The only problem is that during mounting and moving the block around, it's very abusive the the CPU pins (I actually bent a pin after a while). The retention bracket as mentioned in coolingmaster's article is a problem in itself and will artificially bow the processors IHS concaved making the problem even worse.

I took my retention bracket and ground it down thinner and straightened it out a bit to avoid all that pressure on the center points. Now the retention bracket has just enough pressure to hold it in place without too much force.

I've been playing with the new quad stock and finding the differential between cores is more than I would like, so I'm going to lap it anyway even though it's a pretty good stock IHS.

Maybe some of these manufacturers will pick up on the stepped base idea too. I think quads are even more sensetive and better served with a stepped flat base as opposed to bowing.:up:

[smee]

yeah, must quads are totally rippled. The edges start out high, then there seems to be a mote about the cores and then at the cores it is high again. At least this is my experience with Quad IHS's. Like I said, dual core IHS's seem to be in much better shape in terms of being flat.

Hehe, yep... I had a little mote around my cores... :rolleyes:

http://i254.photobucket.com/albums/hh106/parkman14/88957069.jpg?t=1211034950
Sorry for blurry pic....

[Jimmer411]

I didn't vote because I don't believe the either one is definite. If its already flat, no point. Tho if its not flat its very much worth it.

Personally I prefer to remove the ihs when possible. Not hard to glue back on if you need to rma... I uncapped my 4400 x2 s939, but didn't bother even lapping my 3800 am2.

I personally wouldn't lap a new cpu untill I ran it long enough to determine that there are no issues, like if you q6600 had a bum core that can't do more than 2.8

[Digital Nirvana]

Personally if you want any results to be accurate I believe you are going to want to lap it, there is no way to guarantee that you are going to have a relatively close temperature to person B, who could have a different IHS wave or what-not on his. If you lap it, that means that if person B wants to truly have as close of temperatures to you as he can, he should lap it, that way your results can apply to as large of audience as they can. No one has the same IHS as the next guy, which is the problem with not lapping yours.

[Martinm210]

I lapped it, here are a few in-between shots of the leveling that took place.

I think it's actually very difficult to keep from rounding the edges down a little. After taking my good straight edge to it again, there is still a very very sight convex shape to it which I think it good, but I removed the slight dips that existed between the center cores areas and the IHS perimeter.

I didn't take a final shot, but I quit just after removing the last of the nickel coating. I suppose some of the gain from lapping is also removing this coating, it's extremely thin, but copper does have better thermal conductivity than nickel.

http://img255.imageshack.us/img255/682/ihslappingis3.jpg

It's done, no going back....now for finding a nice solidly stable OC.. I'm not one to put up with any instability, so I'll probably run it backed down a bit.

[the monk]

fellas, i quit lapping when i figured i exposed enough surface......when i get time im gonna finnish the job, and do a 100% clean up, peace randy.


http://i301.photobucket.com/albums/nn54/rjarois/005-2.jpg

[mcoffey]

I lapped it, here are a few in-between shots of the leveling that took place.

I think it's actually very difficult to keep from rounding the edges down a little. After taking my good straight edge to it again, there is still a very very sight convex shape to it which I think it good, but I removed the slight dips that existed between the center cores areas and the IHS perimeter.

I didn't take a final shot, but I quit just after removing the last of the nickel coating. I suppose some of the gain from lapping is also removing this coating, it's extremely thin, but copper does have better thermal conductivity than nickel.


It's done, no going back....now for finding a nice solidly stable OC.. I'm not one to put up with any instability, so I'll probably run it backed down a bit.

Are you still using the 680i, because if so you'll have your hands full.

gl,

andyc

[Martinm210]

Are you still using the 680i, because if so you'll have your hands full.

gl,

andyc

Yeah, I'm thinking I need a bios update. It's being testy even at 3.0..currently running P28, so I might have to cross my fingers and try the new P32.:eek:
So much for keeping my bios and having the ability to test more on the dual core...oh well...I'm sticking with the quad...runs alot smoother multitasking at 2.4 than my E6600 did at 3.3.

[Giannis86]

you should be able to hit 3.6 on a quad on some 680s (eg evga)..sadly enough my previous abit went only up to 3.0 because of fsb barriers..anyway you can always just raise the vcore to about 1.4-1.5 safely at stock speeds for some good testing

[RealRedRaider]

I took my retention bracket and ground it down thinner and straightened it out a bit to avoid all that pressure on the center points. Now the retention bracket has just enough pressure to hold it in place without too much force.



So if I read this right, you lapped the CPU retention bracket ???

If so, GENIUS... :up:

[ANP !!!]

I used to have a E6400, lapping got me 4C lower temps, I will surely lap my E8200.

[mcoffey]

Yeah, I'm thinking I need a bios update. It's being testy even at 3.0..currently running P28, so I might have to cross my fingers and try the new P32.:eek:
So much for keeping my bios and having the ability to test more on the dual core...oh well...I'm sticking with the quad...runs alot smoother multitasking at 2.4 than my E6600 did at 3.3.

You'll still be able to run the e6600 and test when you want with another bios. Can't imagine it would change things any.

Gl,

andyc

[Martinm210]

you should be able to hit 3.6 on a quad on some 680s (eg evga)..sadly enough my previous abit went only up to 3.0 because of fsb barriers..anyway you can always just raise the vcore to about 1.4-1.5 safely at stock speeds for some good testing

I've been running at 3.3 with the E6600, but I'm hitting some sort of barrier around 3.0 on this quad, random BSOD stuff..:(

I heard they made some improvements to the bios after P28 specifically for quads, so I'm going in for the plung. I hate bios updates after reading soo many dead board stories.

So if I read this right, you lapped the CPU retention bracket ???

If so, GENIUS... :up:

More of a I ground it down on the belt sander...:) I had just noticed on occasion that it appeared to be high enough that it could impact the base of the block and make for a bad mount. I also straightened out the retention plate a little so there wasn't soo much force at the one point. Not sure it makes that much difference but the cooling masters article points out that the retention bracket itself causes some reverse bowing of the IHS, so I figured..why not..

You'll still be able to run the e6600 and test when you want with another bios. Can't imagine it would change things any.

Gl,

andyc

I'll have to run a validation test first, but considering I have 8 blocks sitting here, it'll be a while before I have my Quad testing done.

Off to Evga for a new bios...

[Talonman]

I lapped my Q6600. ;)
It was not even close to flat to start with...

http://img85.imageshack.us/img85/6036/pic036zd1.jpg (http://imageshack.us)

[Martinm210]

Are you still using the 680i, because if so you'll have your hands full.

gl,

andyc

You weren't kidding. This 680i is a total PITA for the quad. I've tried several different bios versions and still having trouble getting anything decent. I even had the bios corrupt on my three times now. It's almost like Vcore doesn't matter. I can get to about 2900 at a 1.26Vcore and as soon as I start playing with 2950 or 3000...it doesn't matter what voltage I throw at it..:(

I'm seriously considering a MB upgrade now...now just what to get that I don't have to upgrade ram at the same time..:shrug:

[smee]

I'm seriously considering a MB upgrade now...now just what to get that I don't have to upgrade ram at the same time..:shrug:

EVGA 750i FTW!! (http://www.newegg.com/Product/Product.aspx?Item=N82E16813188026&Tpk=750i%2bftw)

The board rocks with quads, all reviews on it are A+++, I'm eyeing this board myself. :yepp:

Or check out the ASUS P5E X38 (http://www.newegg.com/Product/Product.aspx?Item=N82E16813131219), another good one I'm looking at. :up:

[iowamoe300]

You weren't kidding. This 680i is a total PITA for the quad. I've tried several different bios versions and still having trouble getting anything decent. I even had the bios corrupt on my three times now. It's almost like Vcore doesn't matter. I can get to about 2900 at a 1.26Vcore and as soon as I start playing with 2950 or 3000...it doesn't matter what voltage I throw at it..:(

I'm seriously considering a MB upgrade now...now just what to get that I don't have to upgrade ram at the same time..:shrug:

Martin, upgrade to the p32 bios. I did it, and I have no issues @ 3.150ghz with my q6600, and I have a crappy b3 stepping. There was a tutorial on the evga forums about it, and I followed it for every bios update and they went flawless. Do a full reset of your bios by removing the battery and unplugging the psu for 10 min. Boot into the bios, and change only the memory voltage to the oem recommended voltage. Pop in your floppy and do the bios update and your gtg.

[Martinm210]

EVGA 750i FTW!! (http://www.newegg.com/Product/Product.aspx?Item=N82E16813188026&Tpk=750i%2bftw)

The board rocks with quads, all reviews on it are A+++, I'm eyeing this board myself. :yepp:

Or check out the ASUS P5E X38 (http://www.newegg.com/Product/Product.aspx?Item=N82E16813131219), another good one I'm looking at. :up:

I'm looking. I'm a little peeved with EVGA at the moment, so I'm not sure I want the 750 although the reviews do look good.

Martin, upgrade to the p32 bios. I did it, and I have no issues @ 3.150ghz with my q6600, and I have a crappy b3 stepping. There was a tutorial on the evga forums about it, and I followed it for every bios update and they went flawless. Do a full reset of your bios by removing the battery and unplugging the psu for 10 min. Boot into the bios, and change only the memory voltage to the oem recommended voltage. Pop in your floppy and do the bios update and your gtg.

I had P32 in there, it's the one that was giving me random BSODs, P31 seems to be doing better although I didn't remove the battery, I just did the defaults, flash, clear cmos with the jumper for 30 seconds thing. I litterally flashed the board 6 or more times today. 3 times with the P32 after instability crashing it no longer would boot into windows noting an error with the bios, so I had to keep flashing it. I havn't had that with P31 yet.


I'll keep trying for a while, it seems like the quad I have has really awesome temps though. It's freaking hot today with ambients still at 26C right now and at 2.9/1.26 Vcore my worst core idles at 33C(Load 40C), so it's a heck of alot cooler running than my old E6600 which suprises me. It also runs clear up to 2.9 Ghz on 1.26Vcore (after droop), so I would think there's plenty left in it. For whatever reason though, I get all sorts of wierd random stuff occurring much higher than that and added Vcore doesn't seem to do anything.:shrug:

This old 680i is one of the first revisions and I heard there were quad issues with the first ones..:confused:

[mcoffey]

You weren't kidding. This 680i is a total PITA for the quad. I've tried several different bios versions and still having trouble getting anything decent. I even had the bios corrupt on my three times now. It's almost like Vcore doesn't matter. I can get to about 2900 at a 1.26Vcore and as soon as I start playing with 2950 or 3000...it doesn't matter what voltage I throw at it..:(

I'm seriously considering a MB upgrade now...now just what to get that I don't have to upgrade ram at the same time..:shrug:

You might be hitting a FSB hole. Try going str8 to 3.2 or above @ 1.4V
.
EVAG 780i OC's like a champ. Had no problems getting my quads up to speed. To me it was everything the 680i should have been, you can SLI and it takes DDR2. Any of the ASUS X38 board do very well also. I have the Maximus and P5E.

I was hoping you'd have better luck, sorry to hear it didn't work for you.

andyc

[Sparda]

My 2 cent is when you dont use Raptor hd , Areca card or SLI....intel p35 or x38 is the best to go for quad because of the mature BIOS. I jump from NF4 to p35 after reading about yummy matrix raid although it already been there since ICH7R.
http://www.ocforums.com/showthread.php?t=467848

Used p5k-e and dfi lt p35(gtl:)). Both are great board and can push my old quad(crappy batch & VID) to 3.3ghz with default voltage & with good ram. But the pcie slots too close to dimm slots. Now I eyeing for maximus because I read it can do 8gb better and can crossflash BIOS with rampage.

My vote for intel:
budget > p5k-e wifi , ip35(dont have any idea)
premium > dfi p35 , maximus.
Of course there are other good mobo like dfi x48 p5k-premium and rampage but I´m not swimming in money.:)

Btw my raid 0/5 with 3x500gbAAKS lost it snappy feeling a bit after loading all the windows services and antivirus....lol:(

[Xilikon]

I have 5 quads in my house and the best board to overclock them is indeed a P35 or X38. My good mix is Abit IP35-E (cheap but overclock well), Asus P5K-E/Wifi-AP (great all-around performance for a great price, ran my Q6600 at 3.6 GHz for a long time, even up to 3.8 GHz for a short time, Asus P5E-VM (mATX but overclock greak, known to get 3.8-4 GHz with Q9450) and Maximus Formula (a favorite among the WC crowd).

As for the initial question of lapping or not, rosco summed it very well. Anyway, the issue with the stock IHS is that it's too variable from cpu to cpu so using a stock one as reference is useless. Lapping is the only way to make the results more consistent.

[71 (Bryan)]

thx smee,
Martin, i cant wait for your results, is your Q6600 a B3 or a G0?

[Martinm210]

Thanks guys,

My Q6600 is a G0, I think it's time to retire the 680i. It was a great board for my e6600, but I'm not convinced my current OC limit is with the chip. I just want a good stable quad overclocking board. I'm tempted by the 750i, but I've never had an ASUS board.

Off to reading reviews...:)

Unfortunately I can't do any thermals with the quad until I settle on an overclock. I want to find a good hard but 24/7 stable OC to stick with. Currently with the 680i, it likes 2880 and stock volts the best, but that's just not much heat and I'd like to push it harder. I'd be really happy with 3.3... or even 3.0.

I've also abused the old 680i for a while now, over a year old now so it's time to give the new quad some matching realestate..:)

[ZOMGVTEK]

Thanks guys,

My 6600 is a G0, I think it's time to retire the 680i. It was a great board for my e6600, but I'm not convinced my current OC limit is with the chip. I just want a good stable quad overclocking board. I'm tempted by the 750i, but I've never had an ASUS board.

Off to reading reviews...:)

The P5K Premium is a good choice for the Q6600. And its super cheap these days. I think $180 new on the egg.

[ripken204]

The P5K Premium is a good choice for the Q6600. And its super cheap these days. I think $180 new on the egg.

ya i bought it when it was like 230, lol. it's a great motherboard tho, now only if my Q6600 G0 was good enough..

[NaeKuh]

You might be hitting a FSB hole. Try going str8 to 3.2 or above @ 1.4V
.
EVAG 780i OC's like a champ. Had no problems getting my quads up to speed. To me it was everything the 680i should have been, you can SLI and it takes DDR2. Any of the ASUS X38 board do very well also. I have the Maximus and P5E.

I was hoping you'd have better luck, sorry to hear it didn't work for you.

andyc

Bah... i had raid problems. :\ Already on 1 RMA.

The second one isnt that bad however i majorly beefed up the cooling on it.


Martin the best board i got to play with is the DFI LT X38. Im thinkn of picking up a UT X48 soon after a friend of mine tells if its good or not. But you know me and how much hardware i go though. Only Look at DFI LT or UT series. Trust me on this bro.


Also i think you missed me when i was going around telling people how crappy the 680i was. Remember i pulled one out of my rig on the 5th RMA. You need to uber voltage the northbridge for good quad overclocking. Then you need to reprep your mosfets so they have good contact cuz now those are going to get super hot.

The board cant handle high load for long periods of time either. Its not the best WCG/F@H board.

[bigslappy]

EVGA 750i FTW!! (http://www.newegg.com/Product/Product.aspx?Item=N82E16813188026&Tpk=750i%2bftw)

The board rocks with quads, all reviews on it are A+++, I'm eyeing this board myself. :yepp:

Or check out the ASUS P5E X38 (http://www.newegg.com/Product/Product.aspx?Item=N82E16813131219), another good one I'm looking at. :up:

if yer thinkin' P5E go with a Rampage Formula X48 it's a much better board ! WAY easy to O/C

i got both ! both are great boards but the R/F X48 rocks & the Bios is better than the P5E

[Xilikon]

Forgot to say that it's very important to outline that you have a lapped cpu in all your thermal testing reviews. The last thing we want to do is if someone see a great review about block X, run to buy one then use it on the stock IHS to be disappointed by the results (a block can work great with a lapped IHS and perform poorly with a bad IHS).

About the motherboard, since we learned to know you enough to say you love to tinker with settings, I guess a Maximus Formula, Rampage Formula or the DFI X38 ones would be a great pick. They have a huge amount of settings, even to the point to be too much for a professional tweaker :p:

[Helfarch]

if yer thinkin' P5E go with a Rampage Formula X48 it's a much better board ! WAY easy to O/C

i got both ! both are great boards but the R/F X48 rocks & the Bios is better than the P5E

However still VERY similar and the P5E can be loaded with a Rampage BIOS.

I have both too.

Forgot to say that it's very important to outline that you have a lapped cpu in all your thermal testing reviews. The last thing we want to do is if someone see a great review about block X, run to buy one then use it on the stock IHS to be disappointed by the results (a block can work great with a lapped IHS and perform poorly with a bad IHS).

About the motherboard, since we learned to know you enough to say you love to tinker with settings, I guess a Maximus Formula, Rampage Formula or the DFI X38 ones would be a great pick. They have a huge amount of settings, even to the point to be too much for a professional tweaker :p:

Personally I'd lap it and specify in the revies after all it's a test for the blocks, not the quality of the CPU and or it's IHS so as long as all the blocks are tested with the CPU in the same state the comparison is valid. A 'bad' IHS will adversly effect temperatures more often than not.

[Martinm210]

Thanks for all the great MB notes, I spent most of yesterday looking at all the suggestions made in this thread and still havn't made a purchasing decision yet. If nothing else I'll just increase the volts on the 680i where this quad runs good and solid and leave it be.

This quad runs way cool though, never thought I'd be dissappointed in a chip running too cool!...lol!