Couldn't take it any longer - lapped my E6400

[[timko]]

I was just looking at a few lapping guides before I put sandpaper to the bottom of my Ninja-Plus and found this...

http://forums.techpowerup.com/showthread.php?t=19

Do I want a mirror finish?

If you start lapping your surface beyond the 600 grit range your gaps in the heatsink will be very small. Thus you MUST apply a very very small amount of thermal paste for proper thermal transfer. In most cases where you have a mirror finish your final result will be like this:



There's no contact between the two metals because of excess thermal paste. The heat transfer of this is worse than with a surface that is only 600 grit because of the boundary layer of TIM between heatsink and CPU.

So to the guys that have gone for the mirror finish, have you remembered to reduce the amount of AS5 etc you use after lapping?

[mesyn191]

What you think (and speculate) doesn't matter more than what i wan't and it absolutely doesn't make my argument null and void. Fact is Intel is selling some chips with uber crappy IHS and i saw a bunch of XS fellas that got pissed when they got theirs.

The problem with this arguement is your definition of "crappy IHS" and it is YOUR definition which makes it irrelevent, so no its not a "fact" at all that the IHS is crappy.

Does your argument about running in spec at stock make the crappy IHS any better? No, so i'd say this one is null and void

Straw man argument.

[mesyn191]

Meanwhile this thread is at XtremeSystems where 99% of us give a about an issue like this.

Doesn't matter, you're running thier product out of spec if you OC so you have to expect to run into issues like this, as I said before Intel/AMD/etc. do not cater to overclockers.

If you don't want to hear it from me you can always email Intel and get an official stance on how they view these IHS "issues", make sure to mention that you're OC'ing while you're at it too, wouldn't want to lie now would you?

[Fred_Pohl]

Just curious, but what methods of measurement are all you amateur machinists using to make the determination that your C2D IHS's are "concave" when a HS is attached with the minimum clamping pressure specified by Intel?

[LowRun]

The problem with this arguement is your definition of "crappy IHS" and it is YOUR definition which makes it irrelevent, so no its not a "fact" at all that the IHS is crappy.


Straw man argument.

Well, the purpose of the IHS is to transfer the heat from the core to the HSF, for that it must be FLAT, which is very easy to achieve nowadays, so if only the edges of the IHS make contact with the HSF for like 30% of the IHS surface then crappy seems totaly adequate.

[Eastcoasthandle]

Some of you guys are forgetting something when you have a true IHS and true HS surface any thermal compound you place on it will push its way out through the sides. A perfect contact between IHS and HS surface requires little TC. That's not to mean only apply little TC, you have to actually apply TS first and let the compressio between IHS/HS surface do the work for your.

[LowRun]

Just curious, but what methods of measurement are all you amateur machinists using to make the determination that your C2D IHS's are "concave" when a HS is attached with the minimum clamping pressure specified by Intel?

Use the search function, there are threads allready regarding the Concave IHS with means to control how bad it can be.

[Fred_Pohl]

Use the search function, there are threads allready regarding the Concave IHS with means to control how bad it can be.

Why waste my time? The best way to mitigate this alleged IHS concavity is to apply a HS with sufficient clamping pressure. That's how Intel designed it to work and it works fine for me.

I used to waste my time lapping sinks and IHSs but the benefits have always been minimal. IIRC the best improvements I ever saw were 2C by lapping and 4C by removing the IHS.

[mesyn191]

Well, the purpose of the IHS is to transfer the heat from the core to the HSF,

Actually its keep the core from getting crushed, remember all the issues people had with socket 370 and socket 462 chips?

for that it must be FLAT,

It only need be flat enough to ensure the temp. be kept within specificed range, it doesn't have to be perfectly flat at all.

which is very easy to achieve nowadays,

Easy and cheap are 2 different things, perhaps it saves Intel a significant amount of money to loosen the IHS tolerances?

so if only the edges of the IHS make contact with the HSF for like 30% of the IHS surface and the temp is within Intel's specified limit for that chip then that is totaly adequate.

Fixed that for you.

[mesyn191]

Why waste my time? The best way to mitigate this alleged IHS concavity is to apply a HS with sufficient clamping pressure. That's how Intel designed it to work and it works fine for me.

Yea this seems to be the solution to this "issue". I noticed my E6300 has raised edges on its IHS when I was installing it but made sure to tighten the HSF down (I have a Tt BigTyphoon) thoroughly and I've never had any temp issues at all, even when OC'd.

[Fred_Pohl]

Yea this seems to be the solution to this "issue". I noticed my E6300 has raised edges on its IHS when I was installing it but made sure to tighten the HSF down (I have a Tt BigTyphoon) thoroughly and I've never had any temp issues at all, even when OC'd.

This "issue" is a classic example of people making a mountain out of a mole hill. Start with a grain of truth, add some ignorance, rumors and very poorly controlled test results and stir until mixture acheives myth consistency...

I bought into this same "concave IHS" myth when I got my 1.6A a few years back. I began by carefully lapping both my HS base and IHS. My CPU load temp was lowered by 1-2C. Next I removed the IHS and my load temp was lowered by another 2C. Net gain was 3-4C lower load temp and 0% overclocking improvement. Same thing with my 1.8A. I didn't waste my time on my 2.6C or my 3.0C and I won't be wasting time lapping my 6400 either.

[775911]

Seems to me Intel may have done this on purpose, the cpu will work at stock on the stock cooler and they dont really want people running a 6300 at 6800 speeds, that messes their whole price plan up

An IHS which only makes proper contact on the edges helps to keep people in their place (though its effect is not so dramatic really.. ) and if people bypass this, they lose their warranty immediately.
They wont care about us minority of overclockers but they hardly want overclocking done enmass by system builders which would bring their profits down..

If they can make a cpu this complex I think they can make a straight peice of metal, obviously they did not want to particularly

[mesyn191]

Seems to me Intel may have done this on purpose, the cpu will work at stock on the stock cooler and they dont really want people running a 6300 at 6800 speeds, that messes their whole price plan up

An IHS which only makes proper contact on the edges helps to keep people in their place (though its effect is not so dramatic really.. ) and if people bypass this, they lose their warranty immediately.
They wont care about us minority of overclockers but they hardly want overclocking done enmass by system builders which would bring their profits down..

If they can make a cpu this complex I think they can make a straight peice of metal, obviously they did not want to particularly

Oh come on now, we've got enough BS in this topic as is, do you really have to start pushing consipiracy theories? If Intel wanted to stop OC'ing they would and using loose tolerances on the IHS is probably not the way they would go about doing it.

[dissident]

Seems to me Intel may have done this on purpose, the cpu will work at stock on the stock cooler and they dont really want people running a 6300 at 6800 speeds, that messes their whole price plan up

An IHS which only makes proper contact on the edges helps to keep people in their place (though its effect is not so dramatic really.. ) and if people bypass this, they lose their warranty immediately.
They wont care about us minority of overclockers but they hardly want overclocking done enmass by system builders which would bring their profits down..

If they can make a cpu this complex I think they can make a straight peice of metal, obviously they did not want to particularly

Agreed. Also, the stock heatsink supplied with the processors is circular, small, and does not reach the edges of the IHS, which is where the problem is. I think they did this intentionally to encourage those who use "custom" heatsinks to have to lap the IHS, thus voiding the warranty and saving them a bit of money on people RMAing the chips because of excessive overclocking or whatnot.

his "issue" is a classic example of people making a mountain out of a mole hill. Start with a grain of truth, add some ignorance, rumors and very poorly controlled test results and stir until mixture acheives myth consistency...

I bought into this same "concave IHS" myth when I got my 1.6A a few years back. I began by carefully lapping both my HS base and IHS. My CPU load temp was lowered by 1-2C. Next I removed the IHS and my load temp was lowered by another 2C. Net gain was 3-4C lower load temp and 0% overclocking improvement. Same thing with my 1.8A. I didn't waste my time on my 2.6C or my 3.0C and I won't be wasting time lapping my 6400 either.

Lapping the IHS does make a big difference. This is not "poorly controlled test results" or anything of the sort. Lapping reduced the load temps of this watercooled system by 14 degrees load. There wasn't any bad mounting, or anything like that. Thermal compound was applied evenly to the heatsink and securely mounted, then removed, and observed... 80% of the surface in the middle of the CPU was not making contact.

[Iconyu]

Oh come on now, we've got enough BS in this topic as is, do you really have to start pushing consipiracy theories? If Intel wanted to stop OC'ing they would and using loose tolerances on the IHS is probably not the way they would go about doing it.

We all know, Intel included, that these chips are very, very underclocked. Its like a overclockers wetdream, crazy high IPC and possible 4ghz on air, with mainstream software hardly pushing these cores at 2.4ghz nevermind higher.
So while I doubt Intel would go to the trouble, if they did I wouldn't blame them, better than clock limiting them I'd say.

[toledo]

If the chip runs advertised speeds at default voltage within thermal tolerances and is stable with the stock heatsink - how exactly is Intel "letting quality by the wayside?"

Don`t forget a lot of people don`t use retail heatsink . It is a matter of robust design , quality is not just "work under specs " , i am not saying that they should support overclock but at least an IHS that supports other than the stock heatsink .

[momomo6789]

mine was backwards from yalls mine was i never got to the outside like still some
of the metal around it when i frst got cpu i put a razer ontop ad i could see the diff from the sides to the middle @@ thats how bad mine was but now its 100 % flat

[JayG30]

Hey guys quick question. How much is everyone lapping there CPU? Till it shows copper over the whole surface? I've lapped mine for a while but am not getting to much copper to show. Should I keep going? I had to take it apart to do some work so I figured I'd go all out this time since last time I only did minor lapping to the CPU.

[mesyn191]

We all know, Intel included, that these chips are very, very underclocked.

Actually you and/or we do don't know that. Just because a chip OC's well doesn't mean its underclocked. If anything current C2D's remind me of the Celeron 300A's which overclocked like crazy too.

[mesyn191]

Don`t forget a lot of people don`t use retail heatsink . It is a matter of robust design , quality is not just "work under specs " , i am not saying that they should support overclock but at least an IHS that supports other than the stock heatsink .

AFAWK it does, its only all the OC'ers here that are running into heat issues.

[Iconyu]

Actually you and/or we do don't know that. Just because a chip OC's well doesn't mean its underclocked. If anything current C2D's remind me of the Celeron 300A's which overclocked like crazy too.

Maybe underclocked was the wrong word, but Intel picked a very low thermal envelope, as these cpu's have deep pentium M roots. When you build to such a low watt range, but the chip can handle more you create head room.
Core 2 has a lot of head room, but Intel plans to stay within these self inflicted limits to stop AMD just releasing high voltage chips.

[mesyn191]

Maybe underclocked was the wrong word, but Intel picked a very low thermal envelope, as these cpu's have deep pentium M roots. When you build to such a low watt range, but the chip can handle more you create head room.
Core 2 has a lot of head room, but Intel plans to stay within these self inflicted limits to stop AMD just releasing high voltage chips.

A low TDP doesn't guarantee a good OC "head room". C2D OC's well because its a new arch. that is designed to scale in clockspeed over a period of several years and processes, much the same reason why the 300A OC'd so well which was why I held it up as an example. Watch what happens when the EOL Core arch. chips are released, I bet they'll OC about as well as the 1Ghz PIII did.

[Turtle 1]

I really found it rather hard to believe, I was reading a thread like this at XS.

My feeling on this topic are very simple . If you build your own pc which should be better than the same hardware if you bought a FNwest pc . Lapping all heatsinks and the Cpu should be automatic. I can't even imagine doing things any other way. Its the same with I really don't need Pci x slots . Well using sata add in card to run 4 hardrives in Raid0 is better performance by a lot. It won't make any differance's in super pi. But in the real world . Gaming /video editing / ect. ect. ect. Its a high performance gain that is measurable signifently. A high performance PC has all the little time consuming tricks completed Lappping / wire management./ All the hardware added that takes a PC to its maxable performance level . Which would include Sound card/ Sata card./ Top of the line GPU. At least 4 harddrives running raid 0 and a tweaked OS. Watercooling as a base. Other wise their is nothing Extreme about your home built DELL!!!!!!

[Turtle 1]

Dang double post!!!!!

[JayG30]

I honestly just look at this as more of a discussion on how well Intel is making there chips...not that people should or should not need/should/could lap there CPU's. Simple that Intel really did a poor job here and should/could have done a better job but they decieded not to and that is just sad. Maybe not everyone is having tremendous issues with there CPU's but I can tell you that I sure did. When a CPU runs sluggish even at stock speeds thats unacceptable...and mine was getting better contact stock then most that I saw here. After smoothing out the edges a bit it ran much better...honestly more along the lines of what I expected to see at stock.

After a few hours now both my CPU and heatsink are flat as can be. I went through to the copper on the CPU...now it's time to see if it actually makes a difference.