Is lapping a core i7 worth it?

[GMdoubleG]

Hello all, I hope this is in the right section since I am talking about temps while using a Water Cooled setup. Anyways, is lapping a core i7 920 worth the risk and effort? I hear that the i7's run relatively hot with all the power they consume and I'm wondering if I will see just a degree or two difference, or will it be more. Also, after seeing video demonstrations of how to lap a CPU, I noticed that most CPUs have an uneven surface and lapping the CPU will aid in the mounting of the CPU block.

I direct these questions/concerns more towards those who have experience with lapping because I have done plenty of research on this and all I seem to get is theoretical positives and negatives on lapping, but as always, everything is welcome.

PS. Any guides on how to lap and what to watch out for?

Thanks

[Albaholic]

Depends on the condition of your IHS. Here's a lapping guide (http://www.ocforums.com/showthread.php?t=220200)right from the intel cpu section. A little old but still valid. Lapping may hurt or help the resale value of your cpu depending on the quality of the work you do. I haven't lapped my i7 yet, but i eventually will. I've lapped plenty of other cpu's before and always seen 5-10c improvement.

[GMdoubleG]

Wow, 5C-10C! That seems like quite the improvement. What do you mean the condition of my IHS? Are there differences between the IHS' on the same type of CPU?

[Daddyjaxx]

I believe what he means is that some are more concave than others. The more concave, the better the temps by lapping. The cooler touches more surface area of the IHS.

[GMdoubleG]

Oh I see. That does make sense. Does lapping remove a sizable amount of metal? Could there be a possibility of the CPU water block putting too much pressure on the lapped surface and cause it to crush down?

[hokiealumnus]

It won't remove as much as you're thinking unless you go crazy with it. I've seen one guy on XS that lapped his Q6600 all the way down to the dies, so it's possible.

Before lapping, do a razor test - put a razor's edge across the CPU from multiple angles and hold it up to a light. If you see light under the razor, the IHS is not flat and you'd most likely benefit from a good lapping. If it's flat to begin with, there will still be a benefit, but potentially not significant enough for the trouble (meaning 1-2c as opposed to 4-5c), unless of course you want every single degree you can get and would be satisfied with a relatively small drop.

[Boulard83]

^^^^ +1

[Metallica]

So, it says to use 220 till you see the copper? Does that apply to the I7?

[hokiealumnus]

Yes, it does (http://forums.guru3d.com/showthread.php?t=285414). It seems that gentleman gained about 10c when he lapped.

[Daddyjaxx]

You guys have some serious...uh parts of the male body....to do this. I couldn't consider doing it even if it resulted in sub ambient temps. In my mind I may as well be taking sandpaper to two crisp 100 dollar bills. I have no problem building watercooled rigs or messing with computers but my mod skills are really off. I know if I tried this, I would be out either a CPU or a finger.

[Metallica]

Found this guide: http://www.youtube.com/watch?v=sVXuZTuoEuE&feature=channel_page

Think I'm going to try this next week :)

[Boulard83]

if your CPU is concave ... you NEED to lap it. if not your temp will just suck all the time !

[downer]

Lapped my Q6600 and temps decreased by about 5C. At some point I'm sure I'll lap my 920. It feels too new at this point to lap it.

[Daddyjaxx]

if your CPU is concave ... you NEED to lap it. if not your temp will just suck all the time !

I think Intel would disagree with that statement, but then again, overclocking and not using their great HSF voids the warranty, but you can hide those facts. I would think lapping the IHS most defintely will void the warranty. I think they would know. :bday:

[hokiealumnus]

Well, overclocking in the first place voids the warranty. No need to worry about that facet of lapping; if you're at that point, it's already voided!

[Metallica]

Well, overclocking in the first place voids the warranty. No need to worry about that facet of lapping; if you're at that point, it's already voided!

I don't believe they can tell if you have overclocked a CPU at some point :beer:

[hokiealumnus]

Their engineers are smart; they can figure out if you killed a CPU from overclocking. Not to mention it's not exactly ethical to RMA a CPU killed by OC'ing.

[GrimX9]

It won't remove as much as you're thinking unless you go crazy with it. I've seen one guy on XS that lapped his Q6600 all the way down to the dies, so it's possible.

Nice and shiny!
http://www.overclock.net/attachments/cooling-experiments/100364d1236402270-lapped-x3220-q6600-core-resized.jpg

[Boulard83]

I think Intel would disagree with that statement, but then again, overclocking and not using their great HSF voids the warranty, but you can hide those facts. I would think lapping the IHS most defintely will void the warranty. I think they would know. :bday:

As its been told. OCing and many other thing VOID your CPU warantie. Ive lapped in the past, not yet on my Q9550 but ill do it for sure in the next month when ill drain and refill my loop. Ill check its flatness and LAPP if needed. i just forgot to check when it was new ... but my temps are fine.

[downer]

Their engineers are smart; they can figure out if you killed a CPU from overclocking. Not to mention it's not exactly ethical to RMA a CPU killed by OC'ing.

I kinda doubt their RMA process is that involved. The people taking care of RMA claims, probably just check for functionality and any visible signs of physical damage caused by the user.

[Metallica]

Their engineers are smart; they can figure out if you killed a CPU from overclocking. Not to mention it's not exactly ethical to RMA a CPU killed by OC'ing.

If you RMA a dead CPU, I'm pretty sure they can't tell if it was from OCing. And if you RMA if for another reason other than being dead, I still don't think they pay that much attention, as long as it works.

[Ben333]

They don't know weather or not you overclocked it, nor would the waste the time or money to find out if you did. I'd guess they maybe pop RMA'd chips in a computer to test them, or they go directly into the trash/recycling. (after inspecting them for physical damage) Lapping however will definitely void your warranty. Also, I don't think saying that anyone NEEDS to lap their CPU is right if it is slightly bowed one way or another. If it is to the point it won't run and has horrible temps with a proper heatsink mounted on it I'd say that's grounds to RMA it but otherwise thermal paste does the job of filling the non contact areas. If you don't care about your warrenty and want some lower temperatures then lapping is an easy way to do it. Make sure its flat though, you can lap it and have it look great but you must lap it evenly - be sure to use equal force so you don't round the edges or lap it at a slant. I don't know why lapping guides suggest starting with such high grits, I usually start at 100 or 200 and then 400, higher if I have the sand paper on hand to finish with.

[downer]

I start at 150 grit, then move to 400, and finish at 600. I don't know how people can start at 400. It takes ages to lap starting at 150.

[Voidn]

Yes, it does (http://forums.guru3d.com/showthread.php?t=285414). It seems that gentleman gained about 10c when he lapped.

Impressive.... but I'm scurd. I'm going to practice on a true and maybe some old hardware first.

[GMdoubleG]

These are two videos I've found of a guy lapping his E8400 using 600, 800, 1000, 1500 and 2000 grit sand paper. Am I wrong to think that his way is safer because of the exactness of how he uses each grit and there isn't much guessing or "feeling"?

http://www.youtube.com/watch?v=sVXuZTuoEuE&feature=PlayList&p=9A6EDD3F6443769A&playnext=1&playnext_from=PL&index=1
http://www.youtube.com/watch?v=5L1s1-nXj0o&feature=PlayList&p=9A6EDD3F6443769A&index=0

[Albaholic]

You guys have some serious...uh parts of the male body....to do this. I couldn't consider doing it even if it resulted in sub ambient temps. In my mind I may as well be taking sandpaper to two crisp 100 dollar bills. I have no problem building watercooled rigs or messing with computers but my mod skills are really off. I know if I tried this, I would be out either a CPU or a finger.

It would take some serious talent to lose a finger while lapping with just plain ol' sandpaper

These are two videos I've found of a guy lapping his E8400 using 600, 800, 1000, 1500 and 2000 grit sand paper. Am I wrong to think that his way is safer because of the exactness of how he uses each grit and there isn't much guessing or "feeling"?

http://www.youtube.com/watch?v=sVXuZTuoEuE&feature=PlayList&p=9A6EDD3F6443769A&playnext=1&playnext_from=PL&index=1
http://www.youtube.com/watch?v=5L1s1-nXj0o&feature=PlayList&p=9A6EDD3F6443769A&index=0


It's really not that hard & you don't have to go all the way up to 2000 grit. That mirror finish isnt going to net you much improvement. If you're really that worried. Pick up some free old hardware off CL to practice

[Spawn-Inc]

i started off with old hardware and then moved to my celeron 420, then to my q6600. it's easy, just time consuming. i taped off the back of the chip and used Styrofoam to make it easier as well.

q6600 thread (http://www.ocforums.com/showthread.php?t=564355)

celeron thread (http://www.ocforums.com/showthread.php?t=529941)

[orion456]

Is temperature really a limiting variable for i7's? It seems frequency and voltage are the main problem. Especially water cooling, the i7 stays at reasonable temps before it stops booting.

If so, then lapping won't produce much improvement in OC even if you get a good temperature drop; which at best would be around 5c, and worst might be nothing.

Lapping doesn't seem worth the bother and loss in CPU value.

[GMdoubleG]

I haven't heard anything about the i7 having shut down problems due to high temps but I'd like to keep the temps as low as possible for ease of mind.

Now that everyone has convinced me that I should lap my i7, how long should I run/test the CPU to make sure that it won't fail on me from faulty production? Is having to RMA a CPU very common?

[Voidn]

Is temperature really a limiting variable for i7's? It seems frequency and voltage are the main problem. Especially water cooling, the i7 stays at reasonable temps before it stops booting.

If so, then lapping won't produce much improvement in OC even if you get a good temperature drop; which at best would be around 5c, and worst might be nothing.

Lapping doesn't seem worth the bother and loss in CPU value.

Some of the D0's are high temp low vcore, like mine, so every degree counts, 5-10c drop in temp would mean I could run 4.3 or 4.4 ghz 24/7 at 1.4vcore which is bordering on the unsafe but I see people going higher all the time. Or I could stay at 4.2 and never go over 70c, which would make me all warm and fuzzy inside :).

Though it is probably a good idea to find your chips limits before lapping, to see if 5c will do you any good. If you already have good temps but it needs too much vcore/qpi to clock higher then lapping wouldn't matter.

[GMdoubleG]

My Apogee GTZ has a few stains on the surface that makes contact with the CPU so I want to lap it. Should I use the same procedure as lapping the CPU? Or should I try to take as little off as possible?

[Conumdrum]

GM, I'd not lap it. The spots are wayyy less than 1/1000 thick. Ketchup and a good scrub with a pad that has the plastic brillo on the bottom is enuff. The kind of pad that billions buy at the grocery store.

Don't worry, it's not going to make a diff in temps.

[GMdoubleG]

Ok, I've tried using ketchup and the pic above is after leaving ketchup on the copper for about 5 min. At the moment, I am trying to diagnose why my Q9450 temps are so lousy before I go to the i7. This is just another thing I wanted to rule out.

[Albaholic]

What kind of temps are you getting?

[GMdoubleG]

In mid-day, with ambient temps below 80*F, load temps at stock are around 52*C and load temps OCed to 3.2Ghz (400FSB) hits highs of 60*C. As in my sig, I have a dedicated loop to the Q9450 with a XSPC 360RX rad. I am now thinking that there isn't enough air being passed through my rad. The pic below shows that I have two MCP655, some hosing, and lots of wires that might be inhibiting air to flow though.

[GMdoubleG]

I have three Yate Loon fans at 1650 RPM on the XSPC 360RX.

[Spawn-Inc]

leave the ketchup on the block for 30 minutes and it should be gone, did for me.

[hitokiri_808]

Before lapping, do a razor test - put a razor's edge across the CPU from multiple angles and hold it up to a light. If you see light under the razor, the IHS is not flat and you'd most likely benefit from a good lapping. If it's flat to begin with, there will still be a benefit, but potentially not significant enough for the trouble (meaning 1-2c as opposed to 4-5c), unless of course you want every single degree you can get and would be satisfied with a relatively small drop.

+1. Lap both the hs and ihs for maximum contact.

[GMdoubleG]

Will leaving ketchup on the GTZ for 30 min cause any unevenness?

[Spawn-Inc]

it won't mess with the flatness or anything, it doesn't work that fast. try 15 min first if you worried about that. then go for more if it's not removed. i personally went out and forgot about my dtek in the ketchup for a good 2 hours. still have great temps.

[GMdoubleG]

So I have a friend who finally wants to get into water cooling and I was going to sell him my unused parts and have him buy an apogee GTZ for his i7 setup. Is this still the best CPU block for the i7? I know there are plenty of threads around here discussing this but I haven't been able to find any data comparing the GTZ, Heatkiller 3.0, etc.

[ascl]

Heatkiller 3.0 is the best block atm I think, however the GTZ has a nice mounting system, which was enough to sway my decision (as a noob at WCing, I figured an easier mounting system was worth the loss of a couple of degrees).

Detailed info here:
http://www.xtremesystems.org/forums/showthread.php?t=224976

[GMdoubleG]

Am I reading those results correctly? Are they getting the i7 920 temps into the low 30*C while being overclocked to 3.7Ghz?

[GMdoubleG]

Nevermind, I just realized that it is the temperature difference between the CPU and the water. Now that I figured that out, can someone tell me how that translates? If there is a smaller difference between the temps then the Water Block is removing more of the heat from the CPU?

[hokiealumnus]

Yes, that's correct. If the water-to-CPU delta is 30C with a HK block and it's 33C with a Swiftech block, then the HK is performing 3C better than the Swiftech (those numbers are out of the air and have no bearing on actual performance).

[Daddyjaxx]

I take those reviews with a grain of salt. If those few degrees celcius is that important to you, you are running the chip too hard anyway. It's not the heat that kills CPU's neccessarily, it's the voltage. I'm sure plenty of them have been killed on even phase with a lot better temps than we would ever see with water.

[GMdoubleG]

It seems the heatkiller 3.0 is quite hard to set correctly and not worth the extra 1-3*C. I'm not going to be pushing my system very hard but I would like to have a stable OC 24/7.