View Full Version : Lapping A CPU Core
OK I have a coppermine core in me hands and I wonder if lapping is very dangerous to a core. I mean I'm scared to crush the core.
Has anyone ever done it? How much improvement have they seen(espically with the cumine) and how dangerous is it has anyone killed a core?
donny_paycheck
12-16-01, 09:21 PM
Originally posted by tsunami
OK I have a coppermine core in me hands and I wonder if lapping is very dangerous to a core. I mean I'm scared to crush the core.
Has anyone ever done it? How much improvement have they seen(espically with the cumine) and how dangerous is it has anyone killed a core?
Good question. I wonder this too as with the athlons the lettering is raised a little. I'm eager to see what people have to say.
TruckChase!
12-16-01, 09:28 PM
I lapped one of my old p3s and saw no noticeable (EDIT: worthwhile rather than noticeable.. 'bout 0.5 deg c in optimal conditions, well within the margin of error) performance improvement. The coppermine cores are pretty flat to begin with. I'd spend your time making sure the heatsink is flat, then fill in any gaps with ASII.
Originally posted by TruckChase!
I lapped one of my old p3s and saw no noticeable performance improvement. The coppermine cores are pretty flat to begin with. I'd spend your time making sure the heatsink is flat, then fill in any gaps with ASII.
Sounds resonable anyone else reply to this..
AntmanMike
12-16-01, 09:55 PM
You can try. I have lapped a Katmai core and found small improvement (there was improvement though)
Hello Tsunami, I haven't posted in the forums since August. I was lurking around and felt like I had to reply to this topic.
DON't do it!! I killed a celly 600 and a P3 700, both coppermine FCPGA's by lapping them. I did get about 1 degree celsius lower temps before the 700 bit the dust. The Celly 600 died on the operating table. I used wet 2000 grit sandpaper. I lapped in circular and figure eight motions for total time of approx 30 to 40 secs. rinsing sandpaper often. The first time I lapped the celly I got higher temps, which I attributed to an uneven lapping job. So I lapped a little more and it never again posted. I believe I had that little gem up to 954MHz with air-cooling only. The 700 P3 died this past week. I came home from work and noticed that we had experienced a power failure as all of the clocks TV/VCR etc were blinking. I had set the bios to not restart the pc if a power failure occured. I switched it on and...nada. The pci cards and drives powered up but no post. I lapped the chip last June or July and have been running at 924MHz/1.95VCore up until last month when I decreased to 854MHz/1.8VCore because of heat issues. (it would occasionally crash to desktop while playing Counterstrike.) I believe it would still be alive if I hadn't weakened the core by lapping it. Close examination of both cores reveals physical defects. The Celly had an almost invisible crack right on the edge and the 700 is chipped. For the second time, "I promise to never lapp another core!"
:burn:
Maximus Nickus
12-17-01, 12:31 PM
Conclusive evidence it seems, not worth it, espicially as AMD's are very delicate anyway;)
Warlord2
12-17-01, 07:16 PM
I think Colin said he lapped a couple of Tbirds without a problem
as long as you stop when the writing is gone
I think I may lap mine when I take it apart again
Wicked Klown
12-17-01, 08:28 PM
Both my Celeron 800 and my T-Bird 1.1 are lapped. I wouldn't want them any other way.
I dropped the temps a couple of degrees C by lapping my CPUs. With my adventures in super cooling I no longer find it necessary. Where it would really help is with one obnoxious 1.33 AXIA I have that has a convex core, only if it were possible to hold the CPU absolutely flat while lapping.
If you are going to take a walk on the wild side, use 2000 grit and lots of water. Check your work every 20 seconds with a 20 power loop. Do the final lap with a fresh sheet of 2000 grit and Arctic Silver as the lube.
RangerJoe
12-17-01, 09:47 PM
yes, i would not recommend trying to lap the core....i have killed 6 cpu's trying to do this....so take my word for it, nothing good comes of it
*JEREMY*
12-17-01, 09:59 PM
Originally posted by RangerJoe
yes, i would not recommend trying to lap the core....i have killed 6 cpu's trying to do this....so take my word for it, nothing good comes of it
I never would have done it again after killing just one:eek:
CrystalMethod
12-17-01, 10:09 PM
My conclusion to all of the above is simply...
... try at your own risk. Might pan out for you, or it could kill your CPU. Geuss it depends on how skilled you are and how much, errr... let's just say "how big they are in your pants".
LOL I repeat myself I'm not going to try this thing!
The devastating stories!
Intraveinous
12-18-01, 12:45 AM
Basically, I personally don't trust myself to do any exposed core... I'll lap a GF2 chip, I'll lap the IHS on Old Cellys and new Cellys and P4s, but I won't get near my athlons or durons...
Guess that's just me... I don't have the money to mess up and even 3 degrees isn't worth a hundred bucks to me...
peace
John
Maximus Nickus
12-18-01, 03:35 PM
Ok fact: Intel's are cooler than AMD's.
SO this is weird, somone took apart a P4 so they could see the core (as you know it has a shim built in!) and inbetween there was poor contact with it and the shim and a blob of therman grease!!
How they can still run cooler.....
Oh well, bring on the Thoroughbred, 1.8Ghz,1.65V and out in Feb!!!!
The P4 core has a much larger surface area than an Athlon, so it's easier to transfer the heat to the heatsink.
Maximus Nickus
12-18-01, 03:49 PM
That shim doesn't do it any justice though, why don't AMD just bring the rest of the CPU up to the same height as the core?? Or make the core lower?
It would make sense surely:)
Originally posted by nick_cw
That shim doesn't do it any justice though, why don't AMD just bring the rest of the CPU up to the same height as the core?? Or make the core lower?
It would make sense surely:)
No, not really. The core is where the heat is generated. And there are other "bits" on the processor that you probably don't want the heatsink compound (ASII) on, or getting heat spread TO them by the HSF.
So I think it makes more sense to have the core raised.
Gravity Man
12-18-01, 07:25 PM
I lapped a P 166 MMX. It wasn't a PPGA chip, either, so it took me about 2 hours to get flat (ceramic is quite resistant to sanding, and it was concave as hell).
Originally posted by Gravity Man
I lapped a P 166 MMX. It wasn't a PPGA chip, either, so it took me about 2 hours to get flat (ceramic is quite resistant to sanding, and it was concave as hell).
I don't see much point to lapping P 166 since they are fairly flat and it wouldn't make much of a difference since they don't run as hot as the modern chips today.
Gravity Man
12-18-01, 09:03 PM
it got me from 300 to 333, and no, they aren't very flat, they are actually quite concave, and the stepping code is burned in with a laser, and is anything but smooth.
Originally posted by Gravity Man
it got me from 300 to 333, and no, they aren't very flat, they are actually quite concave, and the stepping code is burned in with a laser, and is anything but smooth.
Well, many people took the chance of removing lil bumpy letters just for 0.5 C so I don't really think its worth it and worst of all people killed their chips for it.
So lapping a heatsink you would see 2-5 degrees cooler.
Lapping a core you take a risk of killing the core and get 0.5 Celius cooler and you still probably end up with a slighty rounded core.
Gravity Man
12-18-01, 10:13 PM
I would have never done that to an athlon/duron/P-III, etc.
The P 166 has the core on the bottom of the package, ather than on the to as with all currently available chips, so you can't really lap that, you just lap the ceramic case that the heatsink comes in contact with. As for temperature change, I really don't know, we didn't have thermistors on mobos back then. I can tell you that when running@333 MHz and 3.5v, it put out 34.2w of heat, 261% of the 13.1w it put out at the default settings. That would be like running an Athlon 1.4@2.8GHz with a 2.1 vcore. 34 watts may not be alot, but when the heatsink is separated from the core by 1/8" of ceramic, and the chip isn't designed to handle the same amount of heat as a modern processor, every bit helps. Sorry about the history lesson, but the P55C is an entirely different beast from a modern Athlon.
man, do I feel old
Maximus Nickus
12-19-01, 05:53 PM
Originally posted by Wa11y
No, not really. The core is where the heat is generated. And there are other "bits" on the processor that you probably don't want the heatsink compound (ASII) on, or getting heat spread TO them by the HSF.
So I think it makes more sense to have the core raised.
I stand corrected:p
How much voltage do you have to put through a P166 to get 333??
I tried a P75 @ 166 but even with a huge HSF the VCore was to much and the board blew, oh well.
What older boards have thoose settings though ??I mean I have a few 166's lying about.
Now why can't my AMD hit 2Ghz!!!!! I guess that the older chips where at the beggining of creation so they have much more room for overclocking whereas todays CPU's ARE ON THE LIMIT!!!:eek:
Lapping a CPU Core?...............Are you nuts!!!!!
Originally posted by abb1
Lapping a CPU Core?...............Are you nuts!!!!!
Well *before* I've heard it straightens out the core and gives the heatsink obviously a better contact.
Gravity Man
12-19-01, 10:05 PM
Originally posted by nick_cw
How much voltage do you have to put through a P166 to get 333??
I tried a P75 @ 166 but even with a huge HSF the VCore was to much and the board blew, oh well.
What older boards have thoose settings though ??I mean I have a few 166's lying about.
I had to use 3.5v :eek:. It was also an MMX proc., and those generally go higher than the Pentium classics (I think). The board I used was for a K6-2 (FIC PA-2013, I think) I used PC-133 SDRAM, and set the FSB to 112 and the multiplier to 3.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.