Laping a CPu is it worth it.

  • Thread starter Thread starter We Live For The One We Die For The One
  • Start date Start date
W

We Live For The One We Die For The One

Laping a CPu is it worth it.

By the way i have a Xp 2400 cpu.

Just want to get the cpu cooler thats all in australia it gets 45c in
summer and it runs 60c+ at load would like it to be about 5c lower or
so.

Thanks.
 
Laping a CPu is it worth it.

By the way i have a Xp 2400 cpu.

Just want to get the cpu cooler thats all in australia it gets 45c in
summer and it runs 60c+ at load would like it to be about 5c lower or
so.

Thanks.
Click to expand...

Generally, no, it's hard enough to lap an Athlon XP without making it
worse than when you started, and it's relatively flat to begin with
except for laser etching in the corner. Unless you note an odd,
surface irregularity, it should be fine unlapped.

The heatsink, on the other hand, may benefit from lapping. Then again
only so much gain will come from a budget heatsink, if you have high
ambient temps and want a lower CPU temp than really necessary (it
only really matters if it's stable), then you may need buy a better
'sink. Thermalright SLK series are pretty good, and relatively quiet
with the right fan. Given high ambient temps you might consider a
mildly-low but not lowest RPM fan on the 'sink, perhaps a Panaflo
FBA08A12M1A. You might still wish to lap the heatsink, and note the
case ambient temps, NOT what's reported by a motherboard monitor or
bios from a chip but rather true case ambient air temp.

Also consider that motherboard temp reports can easily be less than
accurate, that sometimes even a BIOS update will result in (reported)
temp variance of ~ 5C.
 
kony said:
Generally, no, it's hard enough to lap an Athlon XP without making it
worse than when you started, and it's relatively flat to begin with
except for laser etching in the corner. Unless you note an odd,
surface irregularity, it should be fine unlapped.

The heatsink, on the other hand, may benefit from lapping. Then again
only so much gain will come from a budget heatsink, if you have high
ambient temps and want a lower CPU temp than really necessary (it
only really matters if it's stable), then you may need buy a better
'sink. Thermalright SLK series are pretty good, and relatively quiet
with the right fan. Given high ambient temps you might consider a
mildly-low but not lowest RPM fan on the 'sink, perhaps a Panaflo
FBA08A12M1A. You might still wish to lap the heatsink, and note the
case ambient temps, NOT what's reported by a motherboard monitor or
bios from a chip but rather true case ambient air temp.

Also consider that motherboard temp reports can easily be less than
accurate, that sometimes even a BIOS update will result in (reported)
temp variance of ~ 5C.
Click to expand...


This is very good advice...
have a look at your heatsink with a good light and a magnifying lens.

Although some of them might be pretty smooth I've seen some that look
pretty horrible, yet with just a little "lapping" will become nice and
smooth.


OTOH: most cpu's look pretty good as they are...and even if not "perfect"
I'd not want to take the risk!
 
The bond wires within the cpu packaging can be broken fairly easily with
pressure to the package...lapping is NOT a good idea...heat sink compound IS
a good idea and makes any thought of lapping a moot point. (Lapping
involves both pressure and vibration, neither is good for the bond wires.)

OTOH--if lapping catches on, please post here first as I'd like to purchase
AMD and Intel stock since physical damage is the users fault and not covered
by warranty. I'm sure they'd love to sell users 2 or 3 cpus for each
computer they own.
 
The bond wires within the cpu packaging can be broken fairly easily with
pressure to the package...lapping is NOT a good idea...heat sink compound IS
a good idea and makes any thought of lapping a moot point. (Lapping
involves both pressure and vibration, neither is good for the bond wires.)
Click to expand...

Not really, the package is fairly resistant to pressure, being that it
expands and contracts quite a bit every time the system is power
cycled, and simply installing a heatsink puts more pressure on it than
someone would while lapping.

I "speak" from experience lapping several T'Birds, Palominos, and
T'Breds, that it's not likely to break bond wires, but that it's also
not likely to make an appreciable difference in temp.

OTOH--if lapping catches on, please post here first as I'd like to purchase
AMD and Intel stock since physical damage is the users fault and not covered
by warranty. I'm sure they'd love to sell users 2 or 3 cpus for each
computer they own.
Click to expand...

Well since you're not even aware that people have and do lap, of
course you'd think it's news. Stick to what you know.
 
You're right, the years of engineering consulting work that I've done for
Intel, IBM, Logitech, Apple, and having been a direct employee of HP must
mean that I am totally unqualified to answer. And all those AMD 29000 bit
slice microprocessors that my test programs have caught with broken bond
wires from some heavy thumbed goon must have been a figment of my
imagination. My warning was just to help someone possibly avoid problems
(never know who has a heavy thumb) for no return on the potential risk. I'm
sure you've "tuned me out" by now, but if you're still listening there are a
couple other possibilities for damage from lapping, both from ESD. The
first, and lesser risk, is from the friction of the grinder on the
case...the second, and more likely, is from the air flow if there is plastic
nearby (can create upwards of 5kv static charge, I've measured it PCBs with
Motorola CPUs exhibiting strange behavior as a result).

Personally, I don't care if you play handball with your CPUs...but for
anyone who cares, the above represents some of my experience that may be
applicable in your situation. YMMV.
 
You're right, the years of engineering consulting work that I've done for
Intel, IBM, Logitech, Apple, and having been a direct employee of HP must
mean that I am totally unqualified to answer. And all those AMD 29000 bit
slice microprocessors that my test programs have caught with broken bond
wires from some heavy thumbed goon must have been a figment of my
imagination. My warning was just to help someone possibly avoid problems
(never know who has a heavy thumb) for no return on the potential risk. I'm
sure you've "tuned me out" by now, but if you're still listening there are a
couple other possibilities for damage from lapping, both from ESD. The
first, and lesser risk, is from the friction of the grinder on the
case...the second, and more likely, is from the air flow if there is plastic
nearby (can create upwards of 5kv static charge, I've measured it PCBs with
Motorola CPUs exhibiting strange behavior as a result).

Personally, I don't care if you play handball with your CPUs...but for
anyone who cares, the above represents some of my experience that may be
applicable in your situation. YMMV.
Click to expand...

Perhaps i should have written that, based on my 100% success rate, I
personally do not see any risk, or rather, any negative effects when
it's done properly. Certainly as with anything else in life, if
there's a way to cause a fault, no matter how remote the chance,
someone, somewhere, will find a way.

This followup you made, listing some issues involved, is a more
accurate and helpful description of the situation, it IS helpful to
know. On the other hand until I see a single instance of a "normal"
procedure for lapping causing a problem, I have nothing but evidence
against this risk you suggest.
 
I agree...until you know who you are dealing with and what they know about
handling ICs, I generally advise against it as the learning curve can be
expensive. Sounds like you either know how to handle ICs or have been lucky
doing a limited number of lappings.

The numbers can catch up to you on success rates. While you've been
successful with all of your attempts, had you done more you might have found
out that your rate was 99.9%...while that sounds great, if I'd had that sort
of rate when I was testing products for some of my past customers (I am out
of that business now) that would mean destroying 10 CPUs per day.

Flatness should be a moot point if you're using a good thermal transfer
(heat sink) compound anyway.

George
 
Back
Top