If and when to lap a CPU.

[DrMrLordX]

IF you're going to lap, use something better than AS5, like Ceramique or . . . Shin Etsu x23.

I lapped my CPU and I am also in the "stop at 600/800" camp. Getting the mirror finish just isn't necessary. All you really want is a flat surface.

I put together a simple put crude video of CPU/HSF lapping here:

http://one.revver.com/watch/193739

or better yet, just search for posts by GalvanizedYankee here on Anandtech. He'll set you on the right track.

 

[cmdrdredd]

Originally posted by: DrMrLordX
IF you're going to lap, use something better than AS5, like Ceramique or . . . Shin Etsu x23.

I lapped my CPU and I am also in the "stop at 600/800" camp. Getting the mirror finish just isn't necessary. All you really want is a flat surface.

I put together a simple put crude video of CPU/HSF lapping here:

http://one.revver.com/watch/193739

or better yet, just search for posts by GalvanizedYankee here on Anandtech. He'll set you on the right track.

There is nothing wrong with using AS5 at all...

I would not lap a CPU unless your temps are really high, I mean beyond what is even considered normal. I would not do it just because it looks nice. I lapped mine because my temps were crazy with an IHS that was not flat.

 

[Keysplayr]

Originally posted by: DrMrLordX
IF you're going to lap, use something better than AS5, like Ceramique or . . . Shin Etsu x23.

I lapped my CPU and I am also in the "stop at 600/800" camp. Getting the mirror finish just isn't necessary. All you really want is a flat surface.

I put together a simple put crude video of CPU/HSF lapping here:

http://one.revver.com/watch/193739

or better yet, just search for posts by GalvanizedYankee here on Anandtech. He'll set you on the right track.

Actually, there is an advantage to mirroring the IHS. When you go to clean it off and replace the thermal compound, there will be no residue left in the micro scratches left behind as if you end your lap at 600/800. So you be the judge. Sure, it "looks" clean, but is it really? hehehe.

 

[Rock Hydra]

lol, lapping. i love the mirror finish. i went crazy doing it when i first learned how. i lapped my p4 2.6c and zalman heatsink, my northbridge heatsink, and the heatsink and gpu heat spreader on my geforce 5900 ultra.

 

[Comdrpopnfresh]

and did it do anything for you?

 

[DrMrLordX]

Originally posted by: keysplayr2003

Actually, there is an advantage to mirroring the IHS. When you go to clean it off and replace the thermal compound, there will be no residue left in the micro scratches left behind as if you end your lap at 600/800. So you be the judge. Sure, it "looks" clean, but is it really? hehehe.

That's only an issue if you're going to apply TIM more than once for the life of the chip. If you're constantly swapping HSFs, yeah, that might be an issue. Still, if you're using the same TIM every time you apply it, I don't think it would be a big deal unless there's excess TIM building up on the IHS.

 

[DrMrLordX]

Originally posted by: cmdrdredd

There is nothing wrong with using AS5 at all...

There isn't all that much right with it either:

http://forumz.tomshardware.com/hardware...-Interface-Shootout-ftopict221751.html

Sure, it's better than generic TIM that comes with stock HSFs and the like, but there are better products out there (like Ceramique). Some folks prefer STG-1, some folks like MX-1/Shin-Etsu G-751, some like X23, some want liquid pro. Bottom line is they're all at least a bit better than AS5. If you're going to take the time to lap the IHS, you may as well squeeze out a bit more cooling performance by using a better-than-average TIM. I can't blame anyone for avoiding liquid pro, though, because it can cause rapid oxidation of aluminum (which is bad) and can ruin motherboards (which is also bad).

 

[imported_bungee91]

I am deciding to lap or not, and if not I think I am running out of things to do (well buy a new C2D, or HSF) or run a hot stock speed chip.

Now I am @ 333X9 (E 4300) I am seeing about 54C at idle and upwards 80C+ at load.
Voltage gain is up 2 points which is around 1.34V

I didn't really look at the flatness of the IHS or HSF, just noticed the finish of the HSF wasn't near mirrored.
I have a Ninja Plus Rev. B

I have ran out of ideas to get the temp down, and have reseated and cleaned/reapplied AS5 once with little difference.

I started a thread about this in the wrong forum earlier today.

My question is if the temps are what they are now, and it turns out my HSF and IHS are pretty unflat then the temps could drop quite substantially right?.. I don't want to waste my time if the max would be say 5C because it is WAY too hot for that to be my end all solution.

 

[Comdrpopnfresh]

Originally posted by: DrMrLordX

Originally posted by: cmdrdredd

There is nothing wrong with using AS5 at all...

There isn't all that much right with it either:

http://forumz.tomshardware.com/hardware...-Interface-Shootout-ftopict221751.html

Sure, it's better than generic TIM that comes with stock HSFs and the like, but there are better products out there (like Ceramique). Some folks prefer STG-1, some folks like MX-1/Shin-Etsu G-751, some like X23, some want liquid pro. Bottom line is they're all at least a bit better than AS5. If you're going to take the time to lap the IHS, you may as well squeeze out a bit more cooling performance by using a better-than-average TIM. I can't blame anyone for avoiding liquid pro, though, because it can cause rapid oxidation of aluminum (which is bad) and can ruin motherboards (which is also bad).

I don't care about one review posted in a forum. I'd like to see multiple repeats of this. I hate to say it, but it is most likely Ceramique's easier application that makes it "better." But looking at the thermal characteristics of it vs as5, as5 is better. It has less thermal resistance, is more thermally conductive. And you're all wrong when you say it is electrically conductive- it is not. These are all scientifically proven facts, not a case-by-case scenarios. How do we know the order the TIMs were placed on the chips? Most of what I have read has said that Lapping only gives improvement in repeat setups because it is th only way to completely ensure all of the previous paste is gone, giving fresh microgrooves for the new paste to set in. Most of the results of thermal paste are in how well you apply it. Do you use a paper towel soaked in alcohol to do it? You should use a lint-free rag. And the proper amounts are important too. AS5 is much thicker than Ceramique, making over application easier. I don't think any weight should be placed on this. But I'd like to see dailytech do some tests on this subject.

"My question is if the temps are what they are now, and it turns out my HSF and IHS are pretty unflat then the temps could drop quite substantially right?.. I don't want to waste my time if the max would be say 5C because it is WAY too hot for that to be my end all solution."

From what I've read, unless Tony Hawk can use either your ihs or hsf to do tricks, 5-7 is the max you'll see. Whether you're going to stick with this processor or not (obviously if your mentioning getting another a bit more $$ won't kill you), get the thermalright 120 extreme.

 

[aigomorla]

Originally posted by: WoodButcher

Originally posted by: aCynic2
I hear one should lap the heatsink and CPU to get the best surface contact, however, I'm certain that lapping a CPU voids the warranty.

So, should I run it first to make sure it works before considering lapping it? If I do that, should I not bother with a HS until I verify it at least works?

Any tips on how to proceed are appreciated.

YES! Test before you lap, You may want to see if it will OC worth a damn before you spend the time lapping.
Personally I lap for flat only,600 / 800 grit, I don't care If I can see my ugly mug or not
I've read arguments for both sides, mirror finish and not. Some claim the micro grooves aid in transfer of heat because you get more metal to metal contact as the grease has channels to get out.
I have no clue.
I get good temps.
I'm not pretty like aigo so I avoid looking in mirrors.
Only the first sentence in this post was worth reading!

Nice work aigomorla. no joke, the extent you go to amazes me.

My comp is my GF. So i tend to spoil her rediculously.

I have also hard about this arguement about the grooves. However, the grooves would need to allign perfectly, with the heat sink. This is almost impossible to do without a machine. If the grooves dont line up perfectly, you get bubbles which the TIM Would need to fill up.

Mirror finish = microgrooves where air can get stuck. Which means a more flat contact and less tim being needed to cover the same area.


And you honestly can't say 15 min of extra work and 2 sheets of sand paper ISNT worth it


@ sandpaper price:

i pay 4.95 for an assorted pack at autozone which has 400 -> 1000 2 sheets each grit

Then i pay 2.95 for an additional 5 sheets of P1500 and another 2.95 for 5 sheets of 2000


Total price ~ 12 dollars after i get my machine towels and whatnots.


My E6600 Lap job. Ended it on 1500 Grit. Ran out of 2000 crit
http://i125.photobucket.com/albums/p73/aigomorla/IMG_0569.jpg

Her best temps.
http://i125.photobucket.com/albums/p73/aigomorla/ApogeeGTX.jpg

Yeah your going to say OMFGWTFBBQ. Its on water incase you want to know.


Delta between lapped and non lapped. My IHS was bad: 8-10C on load.


Can you still say its not worth it if you OC?

 

[DrMrLordX]

Originally posted by: Comdrpopnfresh
[
I don't care about one review posted in a forum.

Did you care enough to check their testing methodology? It's one of the most, if not the most intensive and well-thought-out TIM tests ever conducted by hardware enthusiasts (be it on a forum or on the front page of a tech site).

I'd like to see multiple repeats of this.

You won't, because most sites stopped testing TIMs years ago under the blanket assumption that they all work about as well as anything else, usually based on less-than-thorough testing methods. Though, if you can get VinDSL interested in the conversation, I'm sure he could turn up a few tests showing the superiority of MX-1 (and it is pretty good stuff). In fact, I'll do it for him:

http://www.systemcooling.com/images/rev...rcticCooling_Freezer_64Pro/image17.gif

I hate to say it, but it is most likely Ceramique's easier application that makes it "better." But looking at the thermal characteristics of it vs as5, as5 is better. It has less thermal resistance, is more thermally conductive.

How is AS5 difficult to apply? All you do is apply a grain (or in the case of dual-core processors, a line) in the middle and let the HSF spread it out. Stuff like MX-1 and liquid pro can be devilishly difficult to apply, yet people get better results with those two products frequently.

I remain skeptical of AS5's actual performance versus its rated thermal characteristics. In most of the half-arsed tests I've seen AS5 perform identically to or worse than Ceramique, despite thermal resistance and thermal conductivity ratings. Then DaClan did their wonderful review which showed Ceramique in positive light.

And you're all wrong when you say it is electrically conductive- it is not.

Who said AS5 is electrically conductive? I've been harping on people over that for years. It's electrically capacitative, not conductive. Duh.

How do we know the order the TIMs were placed on the chips?

You could always ask him, or read the thread following the review.

Most of what I have read has said that Lapping only gives improvement in repeat setups because it is th only way to completely ensure all of the previous paste is gone, giving fresh microgrooves for the new paste to set in.

. . . uh huh

Most of the results of thermal paste are in how well you apply it.

. . .

You should use a lint-free rag. And the proper amounts are important too.

Which is exactly what we've been saying on this forum for years.

AS5 is much thicker than Ceramique, making over application easier.

AS5 has easy-to-follow instructions on their site. They have had these instructions for years. Nobody's bagging on AS5 because it's "hard to apply" (it isn't).

But I'd like to see dailytech do some tests on this subject.

Dailytech? You mean Anandtech? They swear up and down that their bulk silver paste is as good as anything else they've tested, though I remain skeptical of their claims seeing as how they haven't published any tests as thorough as DaClan's review. Dailytech just doesn't test stuff like this. They're a news site.

 

[zylander]

Thats pretty interesting, Ive never even heard of doing that before. I assume you can lap any type of CPU, like a 939 Opteron? Just asking because it seems like all the lapping info I can find is of people doig it on Intel chips.

 

[DrMrLordX]

You can lap anything with a copper or nickel-on-copper IHS at the very least. You can even lap a naked die but . . . eh let's not go into that.

 

[zylander]

Originally posted by: DrMrLordX
You can lap anything with a copper or nickel-on-copper IHS at the very least. You can even lap a naked die but . . . eh let's not go into that.

Cool thanks. Im thinking of picking up an opty, I might lap it, Id like to try this.

 

[996GT2]

Hmmm...all this lapping talk has gotten me interested in lapping the base of my Freezer 64 Pro...it looks like this right now with a bit of MX-1 residue left on it from my fresh CPU removal
http://i78.photobucket.com/albums/j120/996gt2/DSC03603.jpg

My 3600+ X2 is coming next week along with my G.Skill HZs. I think I'll lap the Freezer 64 first and see if I get any temperature improvements before lapping the IHS too.

Since I'm out of MX-1, I was going to buy AS5 to put on the IHS...I always thought that Ceramique was just a tad worse than AS5...I know that MX-1/G751 is generally accepted as the top performing thermal paste, but that stuff's pretty expensive with shipping factored in...can anyone else confirm that Ceramique is better? I can get that at the local Fry's for only 4 bucks

 

[Comdrpopnfresh]

my Freezer 64 Pro looks exactly the same on the base. far from mirror finish. the machines grooves are deep enough I think lapping my have an effect.

 

[graysky]

If price is any guidance you'll see that AS5 is more expensive than ceramique. I'd just stick w/ AS5.

 

[DrMrLordX]

Personally I'd rather use Ceramique over AS5, especially on a lapped surface.

Shin-Etsu G-751 (aka MX-1) is great stuff but works best on rougher surfaces. For lapped surfaces, X23-7783D is the best you're gonna get.

 

[graysky]

Originally posted by: DrMrLordX
Personally I'd rather use Ceramique over AS5, especially on a lapped surface.

Shin-Etsu G-751 (aka MX-1) is great stuff but works best on rougher surfaces. For lapped surfaces, X23-7783D is the best you're gonna get.

Interesting advice... what are you basing it on? (Genuine interest, not meant to be sarcastic).

 

[cmdrdredd]

Originally posted by: DrMrLordX

Originally posted by: cmdrdredd

There is nothing wrong with using AS5 at all...

There isn't all that much right with it either:

http://forumz.tomshardware.com/hardware...-Interface-Shootout-ftopict221751.html

Sure, it's better than generic TIM that comes with stock HSFs and the like, but there are better products out there (like Ceramique). Some folks prefer STG-1, some folks like MX-1/Shin-Etsu G-751, some like X23, some want liquid pro. Bottom line is they're all at least a bit better than AS5. If you're going to take the time to lap the IHS, you may as well squeeze out a bit more cooling performance by using a better-than-average TIM. I can't blame anyone for avoiding liquid pro, though, because it can cause rapid oxidation of aluminum (which is bad) and can ruin motherboards (which is also bad).

Who cares? 2-4 degrees? common

and you linked to Toms hardware? don't make me laugh

Anyhow AS5 is available everywhere for very little money. It works fine.

 

[DrMrLordX]

Originally posted by: graysky

Interesting advice... what are you basing it on? (Genuine interest, not meant to be sarcastic).

Mostly word-of-mouth from people who do a lot of work with water blocks. Lots of google searching was involved.

Originally posted by: cmdrdredd

Who cares? 2-4 degrees? common

2-4 degrees is a pretty big deal when you're already gone to the trouble of lapping your IHS and HSF to drop your temps by another 5-10C. It's just one of many steps that can be taken to lower temps that should be taken if you're serious about overvolting beyond the famed "10% safe limit" or if you're dealing with a processor that runs too hot at stock vcore for whatever reason.

and you linked to Toms hardware? don't make me laugh

It's not a post by one of Tom's staffers. Read the post and the ensuing discussion of methodology. They are quite thorough. The review is of better quality than most.

Anyhow AS5 is available everywhere for very little money. It works fine.

Sure, it works fine, but it isn't the best. There are better products out there, often available at a similar cost. Hell, Arctic Silver's own Ceramique has proven to be superior to AS5. The X23-7783D I use costs about the same as AS5, and there are even HSFs out there that come with MX-1 included. Zalman's HSFs come complete with STG-1. You can get better results with all of those compounds.

AS5 isn't the only game in town. If you've got a choice between TIMs, there isn't much reason to choose it anymore.

 

[cmdrdredd]

Originally posted by: DrMrLordX

Originally posted by: graysky

Interesting advice... what are you basing it on? (Genuine interest, not meant to be sarcastic).

Mostly word-of-mouth from people who do a lot of work with water blocks. Lots of google searching was involved.

Originally posted by: cmdrdredd

Who cares? 2-4 degrees? common

2-4 degrees is a pretty big deal when you're already gone to the trouble of lapping your IHS and HSF to drop your temps by another 5-10C. It's just one of many steps that can be taken to lower temps that should be taken if you're serious about overvolting beyond the famed "10% safe limit" or if you're dealing with a processor that runs too hot at stock vcore for whatever reason.

and you linked to Toms hardware? don't make me laugh

It's not a post by one of Tom's staffers. Read the post and the ensuing discussion of methodology. They are quite thorough. The review is of better quality than most.

Anyhow AS5 is available everywhere for very little money. It works fine.

Sure, it works fine, but it isn't the best. There are better products out there, often available at a similar cost. Hell, Arctic Silver's own Ceramique has proven to be superior to AS5. The X23-7783D I use costs about the same as AS5, and there are even HSFs out there that come with MX-1 included. Zalman's HSFs come complete with STG-1. You can get better results with all of those compounds.

AS5 isn't the only game in town. If you've got a choice between TIMs, there isn't much reason to choose it anymore.

The point is...there's nothing wrong with AS5. It's easy to find, easy to use, and cheap. The reason I'd use it over anything else is because I've used it for a long time and I have tons of it around here. And 4c is not a big deal...4c isn't going to get you more overclock.

It's all nit-picking at that point.

 

[Rock Hydra]

Originally posted by: Comdrpopnfresh
and did it do anything for you?

well, nothing really for the cpu as far as load/idle temps, but the time to go from load to idle temps noticably decreased. the northbridge decreased a couple degrees, and my video card saw like a 10 degree decrease in temps at idle and about 2 at load. something around there...it was a long time ago. anyway...i did notice my energy bill decrease roughly $3.50 after that and it seemed to level off around there...not sure if these two things are related.

 

[aigomorla]

GUYS!

DO NOT APPLY AS5 IN A SINGLE DOT ON A C2D CHIP!

That only works on optys and AMD. The proper method is a skinny verticle line in the middle if your looking at a mounted motherboard directly.

Remember the way the core is setup behind that ihs. Its not a square, its a box straight down the middle on the chip!

ALSO!!! another important statement:

If your running a waterblock, apogeeGT, or apogee. You need to make sure your barbs are on top of each other. Water needs to run down the cores, not sideways. It will give you a 3-4C improvement.

OO <-- BAD on C2D

O
O <--- good.

 

[Comdrpopnfresh]

I have a asus a8n32-sli deluxe. I wanted to lap the three heatsinks (nothbridge, southbridge, and the final fins between the socket and the case exhaust) as well as apply as5 to them. but the stupid pink thermal tape they put on there is so ingrained it didn't come off the copper heatpipe, and I ended up scraping it with a screwdriver- so now it is far from mirror. lol. Anyhow, my one usb port no longer works, which I thought might warrant a replacement rma and then I could take off the new motherboard's heatpipe and easily peel off the thermal tape and the pressure relief pads. But when i emailed to inquiry about the rma I got this back:

"solution:

Dear Valued Customer:

If you need immediate assistance, please feel free to contact the RMA Department directly at 510-739-3777 Ext. 5105 to obtain your RMA#.

Please REPLY back to this email to receive your RMA#/Replacement Part.

If you have already received an RMA# for your ASUS product, you may disregard this email. Thank you for supporting ASUS.

Please provide the following information so that we may process your request for warranty repair service. Once we have obtained that information from you, we will issue an RMA # and provide the proper shipping instructions. Please read and provide all of the information below. We cannot complete your request, if all of the information below is not provided.
Thank you,
ASUS Computer International

PLEASE PROVIDE US WITH:
-THE MODEL & SERIAL NUMBER OFF OF YOUR PRODUCT
model (name of product) serial# (10 digits/characters long, no dashes) Serial # located on either the top of the purple parallel port (printer port), side of PCI Slots, side of Brown CNR Slot, or the backside of the product itself)
-YOUR FULL NAME/NAME OF COMPANY (Please provide name of business if shipping to a company for shipping purposes).
-YOUR SHIPPING ADDRESS (no PO boxes please)
-YOUR DAYTIME PHONE/FAX NUMBER
-BRIEF DESCRIPTION OF PROBLEM EXPERIENCING

ASUS Computer International (USA) is a warranty repair service center. Please contact place of purchase for credit, refund or upgrade. Asus does not provide these services under any circumstances.

ASUS Does not cover physical damage (physically broken components on the product). Please refer to page 2 of your users manual. There is a $15.00 - $60.00 fee to replace a broken socket. There is a $25.00 - $60.00 fee to repair all other physical damage. If a product is not repairable, the product will be sent back to the customer. If a product is sent in with physical damage, the product will be rejected and customers will not be reimbursed for shipping charges. A payment can be made by check, money order (payable to ASUS), or a credit card#. The payment must be sent in with the damaged product. Customers from Canada must make payments with a credit card#. ASUS warranty covers all malfunctions that are not caused by physical damage to the product.

Thank you,
ASUS CSR Department
E002 "

if a usb port stops working, when for the life of the product I have left one wire in it for the entire duration (so it is not simple wear+tear) will they still charge me? Will my whole replacement warranty (newegg quotes the warranty to be replacement only) be voided if they realize I have removed the heatpipe and put on a different thermal paste? The two are completely unrelated, and regardless of whether they believe me about the timing, my action would no cause the disturbance of a usb port. Do you think they'll charge me? Will they simply repair it, or replace it?