Slashdot Mirror
Surface Mapping Athlons For Fun And Knowledge
ryemax writes: "'Surface Mapped Athlon - Is Lapping Required?' is the title of today's bit of insight from FrostyTech. Basically we took an Athlon and measured to within 0.0005" how out-of-whack the surface of the aluminum plate is. With that knowledge and a few hundred values we made a nifty image map of the surface features. With that done, the picture was overlaid atop an image of the processor so all can see where surface flatness may be a problem. Why, you ask? Because I get really annoyed when expensive heatsinks/cold plates get suck on unflat surfaces - and thermal performance gets kicked in the arse. So, rather then just say "it's unflat", I decided to quantitatively prove it using a dial micrometer. Bottom line -- lap that Althon plate." Wow.
Forget the dial micrometer, these guys should try a Zeiss confocal laser scanning microscope for some interesting surface maps.
Combine that with an air balanced table and the software that ships with the scope and you'll get some pretty fly images of the surface
without all the fuss of a finicky scanning electron microscope.
Besides, I'll bet that glob of thermal goo would make it pretty hard to pull a proper vacuum.
Egc4ever@carolina.rr.com
Most of the lapping sites I have come accross warn that the grease actually acts as an insulator but is still better than a gap. Your best off lapping and using the thinest layer of thermal compound possible.
Firstly, the "goo" is there to deal with this.
Secondly, why not lap the heat sink too? This is massive over-kill and totally inappropriate if you're using the correct thermal grease.
Again, the grease is there for a reason, and that reason is to create a better contact than crude "grit grinding" ever will.
Why not fuse the heatsink to the chip in production? It would allow for maximum heat transfer and the chip would be more reliable. If they need to fuse a aluminum back on the thing in the first place I see no reason not to put a real heat sink on the chip instead of a heatplate that can warp.
but having quite a bit of experience dealing with extreme tolerances (Fixture Builder/Machinists) for 6 years I can say with confidence it really wont matter with this. if you can imagine the heatsink from a horizontal perspective (and if i read that diagram right.. ) it would teeter off on either side about 6/10ths of a thou (just about 1/2 a thou.) Not a big deal, but to put some perspective on the situation, if you consider a dowel pin is undersized half a thou, it should be press-fit (push it with fingers),,. if its over half a thou, ya might have to force it in with a hammer of some sort, nice and tight.. this is apples and oranges though, dowels are cylindrical and this is two horizontal planes.. (did lots of this ugly work with key slot machining) problem is, it costs more and more money to get the perfection these guys are asking for. Calibrating the machines, paying the operater enough so he doesn't slack off for 90% of the day (cus i know it doesn't pay very well for operators),fix fix fix, replace, lots of external costs... and besides, even with it out of tolerance half a thou, i believe its still reasonable for plate to plate heat transfer... those big red areas on their diagram trick the casual reader into thinking *BAD* when it really isn't.. a piece of paper is about 3 thou(sanths) thick, if you take the alumunim foil from a marlboro pack and burn off the one side that has the thin paper on it, thats one thousanths/inch (ya, i machining can be boring sometimes :) and anything under one thousands/inch adds 5X as much money to build :-) Trust me on that. at an old tool&die shop that was working at, some young brilliant know-it-all mech. engineer fresh out of university did some design work for the company.. First project he designed was a battery holder for some Nissan vehicle (i think, something like that, maybe ford)... of course he had to make sure the die was within a .00005 tolerance... ya, a battery holder.. shows ya how much college does , no replacement for experience.
The key issue is to make sure you have a *no* air between the surfaces. Air and other gases have extremely poor thermal conductivity compared with solids and liquids. Therefore air acts as a insulator (the same principle behind styrofoam and sleeping bags).
)
If there is any air between the surfaces, it should be in as thin a film as possible. Better yet, replace the air with a thermal paste. The paste doesn't have as high a thermal conductivity as metal or ceramic, but it will be much better than air.
Thinner films of thermal paste are better, thus (I assume) the concern about flat surfaces. But if I had to pick, I would take a thick film of thermal paste over a thin film of air.
If you want to do a thought experiment on this, the dimensions of thermal conductivity are
Energy/(Time*(Length^2)*(TempDifference/Length)
i.e. BTU/(hour*ft*DegF)
Using these units, thermal conductivity is approximately:
Vacuum = 0.000...
Air = 0.014
Water = 0.36
Steel = 26
Aluminum = 117
Copper = 224
(Source: Perry Chemical Engineering Handbook)
Thermal conductivity does not take convection into account, but convection can be ignored in thin films.
It's true that a non-flat surface will radiate heat better due to increased surface area. But when you slap a heat sink on top of the chip, you're not trying to radiate heat away at all! You want to conduct the heat from one item (the chip) to the other (the heat sink) so that the heat sink, with its comparatively enormous surface area, can radiate the heat away. The idea is to increase the efficiency with which heat transfers from the casing to the heat sink, and assuming that the heat sink's bottom surface is flat, the best way to do this is to make the chip casing's surface as flat as possible so that the "contact patch" (to use an automotive term) between the heat sink and the chip casing is as large as possible. Essentially, the more direct contact there is between the casing and the heat sink, the larger the cross-sectional area of the surface through which the heat gets transferred and therefore the faster the heat transfers. Simple physics. And my Athlon, formerly a 550, is all the faster for it.
Well, maybe not so but there sure are a whole lot of people like that. Some are just perfectionists, like this guy. Some are fatuous. My favorite was someone who bought these little plastic rings that were applied to CD discs to make them sound better.
Bruce
Bruce Perens.
He should just get access to a Coordinate Measuring Machine (CMM) and do it right. Any decent manufacturing plant will have one.
Your password has expired, please login to change it.
Ok, AMD can make a chip accurate to 0.18 microns. Granted, that accuracy doesn't apply to the top surface of the chip, but the packaging tolerances probably aren't very loose, and I have a hard time believing that Joe Overclocker and a Brillo pad can make matters any better. And how about the heat sink? Is that chunk of aluminum squirted out of a Taiwanese extruding machine really gonna be any flatter than the Athlon?
:)
Whatever.
---
Calling all materials physicists! What would be the net effect of using some other material for the heatsink itself? Aluminum is used because it's cheap not because it's a terribly good heatsink material. Can anyone suggest some other material that some smart lad here could manufacture a heatsink out of and then test and record the results, both from a temperature perspective and from an overclockability perspective?
Just when you thought overclocking had gone far enough, some one comes along to prove you completely wrong. :-)
While not an overclocker myself, I have to admire the lengths to which people will go to pursue an extra few MHz... (Never mind that actual system performance on non-Quake-3 workloads won't feel ANY different, and you'll feel like you're in a sauna if you're in the same room as the machine....) ;-)
--Joe--
Program Intellivision!
These kind folks showed one way to improve conduction. IOW, make sure the contact surface area is improved. I can just see it: a true geek hand crafts one's CPU using high-grit sandpaper.
Is there a substance with a high heat conductivity that is also very flexible so that it clings to the CPU?
While air is a good insulator (hey, check out your double plane windows), the distances describe here are extremely small. Heat will be tranfered somewhat effectively. Right?
My take on this is that the heat conduction will not be uniform across the CPU. Then the weak link will cause failure. I wish that I didn't open my last beer before answering this question.
OT: Concerning /. answer to MS: You go girl.:-)
Are you calling me a hothead?
The last time I built a 500 Watt Amplifier from a kit, the instructions emphasized polishing the heat sink to a mirror finish - he claimed this was very important, and to use polishing compound and a wheel in a drill, which I did untill I could almost see reflections in it. It was in no way perfectly flat but that's what the thermal heat sink compound (available at Radio Shack) is for, to fill the gaps between the device and heat sink. (and a 500 watt amp comes with a BIG heat sink with 2 4" fans on top!)
try { do() || do_not(); } catch (JediException err) { yoda(err); }
Yes. The word you're looking for is "anal."
Cheers,
ZicoKnows@hotmail.com
Plenty of people love to work on their
boats, cars etc and have precision
machining equipment at home. As for
material choice, gold is expensive,
whereas copper is not. Gold plating
your piece may be a good idea to
prevent oxidation.
If you are going to such lengths as lapping
then why not just machine the plate and
heatsink yourself as one piece? And while
you are at it, why not make it out of copper
which is better than aluminum at heat
conduction (use annealed copper - it is
soft enough to conform to the insides of your
processor quite well)?
Some people have way too much time on their hands (says the Slashdot poster)..
In all seriousness, this is impressive. But am I the only one that sees a globe in the surface mapping?
It looks like he's getting the entire heatsink as close to the board as possible, so he sands out holes for everything he needs and gets it closer in.
He's probably STILL using thermal conducting grease, but that isn't as important as getting a good connection in the first place.
Thing I did with MY Athalon was get rid of that
;) hehehe.....
:) can't wait for the ;). :) I'll be in heaven.
cruddy heatsink plate altogether...
Much MUCH more efficient to mount the heatsink
directly to the face of the chip...
the CHIP is FLAT
Miniscule amount of heat transfer paste...
rock solid stability...
I know I love MY Athalon
Durons too
Now if only the manufacturers would get off their
butts and finalize a chipset and m/b layout for
dual/+ Athalons
(Light travels faster than sound. That's
why most people appear intelligent until they
speak.)
Friends don't let friends buy Compaq's. (Dell/Gateway... same same) You want a good computer? Build it yourself.
I read this stuff. I read it again and again and again.
I am willing to bet these guys are audiophiles. No, not guys who like their music to sound good, but guys who spend $600 on a power cord and talk about "richer bass and higher highs because we have a $600 6 foot power cord running from the wall outlet to our amp - never mind the crappy wiring in the house, out on the poles, etc."
Yes, there may be some improvement in the execution of the processor... probably in the order of one or two instructions per second. In other words they can't claim it to be so.
It is like the folks who make the Splitfire spark plug and Slick 50 oil treatment. They claim "get a 15% increse in gas mileage!" Well, the thing they forget to mention is that your gas mileage varies by 15% all the time. It is based on weather, stop and go driving, etc. etc. etc.
So, these guys are spewing forth nothing but snake oil. Take my advise and ignore them, and instead of spending 1/2 hour grinding your Athlon, spend it doing something productive like watching TV or cleaning out your sock drawer.
This was an interesting post, with interesting follow ups. I can see the benefit of "flat" contact between the heatsink and thermal plate but I also can see the point of the critics w.r.t temperature differences and screw-down mounting deformations.
What I'd really like to see is before and after temperature measurements. I think that's the only way to gauge the effectiveness of any lapping procedures.
I suppose you could also solve the heat-transfer equations for various partial/full contact ratios to get a feel for how much improvement is possible but I'm just not up to it right now.
--
Moderation in everything, including moderation.
I think you mean thermal conductivity there guy, not specific heat.
If I had no sense of humor, I would long ago have committed suicide. -Ghandi
1) The stated value of measuring accuracy is bogus. For one thing, if he did it with a micrometer, you are getting thickness of the plate, not flatness of the plate. It could be totally flat on the heat sink side, and unflat on the other. Plus stupid things like if his body heat raised the temperature of the plate, I bet that whacked off the last measurments from the first by a good .001' atleast, etc., etc. If you are going to say something like .0005' accuracy or whatever it was, put error bars on it, or just don't say it. Misleading.
2) After lapping the processor, then letting it run to operating temperatures, I bet the processor warps some due to differential coefficients of expansion and just cranking the processor on to the motherboard, and the heat sink on top of that.
3) I _assume_ that the the little numbers next to their psuedocolor matrix is a color bar, which shows all of about 1.2 mm of difference in surface relief. When you take in to account the crudity of their measurments, the error bars are probably about that big.
4) Do the math. Heat flux is inversely propotional to the distance between the two surfaces and proportional to the thermal conductivity of the medium and the temperature gradient (q = -(k dT)/l). If you change the distance between the heat sink and processor only a little tiny bit, you only change the heat flow a little tiny bit. It's linear. As long as you increase k by putting heat sink compound, and the distance l isn't huge, then dT mostly takes over.
In other words: Use a good heat sink that stays cool and will make a large dt. Use a thin coat of thermal compound. If your processor still gets too hot, turn down the clock speed a little! Geez. Talk about too much time on your hands.
If I had no sense of humor, I would long ago have committed suicide. -Ghandi
Seems to me any processor is going to generate heat. The better the design, the less heat. But any processor is going to get hotter when you crank up the clock speed, and conversely will not run so hot at lower clock speeds. At some clock speed, the processor will be adequately cooled by ordinary convection - if you cranked that 450MHz Athlon down to, oh, I don't know, 100-200 MHz, it wouldn't need a heat sink either.
--
--
Do I look like I speak for my employer?
What I've done on my OC'ed Celerons...
I took the heatsink and chip, and applied a thick slurry of Comet cleanser (the abrasive kind) and water. Then swirl the chip and heatsink in a circular pattern, the Comet will grind down the high spots on the CPU and not mess with the low spots. After a few minutes, wipe it clean and see how much shiny surface is showing. Repeat until you get about 20-30% area polished, thats good enough to clean the whole thing up and use a THIN layer of the thermal grease, swirl it into place just like you did the Comet to get the thinnest layer possible.
Those of you absolutists can go for 100% mirror finish on the CPU, then it is perfectly matched to your heatsink. You still need the grease, but just the slightest amount. The heatsink black oxide coating will not be removed, it's harder than the Comet particles.
Oh yeah, DONT do this to a CopperMine processor or PowerPC, you dont want to put that much stress on the bare silicon back of the die.
Starman97@Gmail.com (bring it on spammers)
This may certainly be overkill but the theory behind the practice is sound. The "goo" in question has a higher thermal resistance than aluminum. So, the more aluminum you can cause to meet, the better the heat transfer you'll get. Without thermal grease however there will be small air gaps in the porous surface of the metal. Unfortunately air has a higher thermal resistance than both "goo" and aluminum. To solve this problem you add a little thermal grease, which will improve heat transfer where there is an air gap. If you apply too much thermal grease however, you risk the possibility of increasing the thermal resistance of your heat sink instead of decreasing it.
Not only that, but it's "Insightful"
aargh
I have seen the future, and it is inconvenient.
hmmm ,so your telling me that if you air cool a brain you can overclock it? well seems that you arent the first to do that....
nmarshall
#include "standard_disclaimer.h"
R.U. SIRIUS: THE ONLY POSSIBLE RESPONSE
nmarshall
The law is that which it boldly asserted and plausibly maintained..
--Colonel Burr 1783
Interesting article. However, the lapping advice really sucks!
You can't lap the heat sink mounting plate much flatter with sandpaper on your granite countertop (or any old glass plate for that matter). As a woodworker who tunes his handtools, I can tell you that lapping requires a VERY flat surface. These can be had relatively cheap, but you have to go out and look for them. I got mine from Lee Valley Tools. Then mount the sandpaper very carefully. It's best to use plastic backed abrasives, rather than cloth or paper backed stuff. Or simply use loose grit.
You also have to hold the tool to be lapped at a very consistent angle and apply consistent pressure/force over the entire surface to be lapped and over the entire lapping stroke. No rocking, no twisting, just smooth moves.
If you use just any hard surface that appears flat to you and don't practice your lapping technique, you might end up with a severely dished surface!
My apologies, "the surface possibly" should have been "the mating surface possibly".
I will forgive you the inaccuracy of the dial micrometer. But I can't forgive the fact that the flatness of a surface can NOT be measured with a micrometer. You need a known flat surface (certified granite plate), and a dial indicator with sturdy, stable stand. Anyone who has worked in a machine shop can verify this for you. The way that you arrived at your measurement; the surface possibly, could be "perfectly" flat.
http://slashdot.org/articles/99 /11/15/1910226.shtml
It was in a quickie a little while ago.. But still amusing nonetheless.. =)
Daniel
daniel@splink.net
Out of curiosity I just Timbuktu'd into my G3/450 B&W at work (which is running PowerLogix's cache boosting utility - gotta crunch those SETI units) and checked the processor temperature with Gauge Pro, and although the surface of the G3 probably isn't all that level, it's still only 41 degrees C. So no sanding of my processor for me, thanks. :-]
I use Macs for work, Linux for education, and Windows for cardplaying.
For a start, a 5mm grid? OK, if you want to not take too long in making your mesaurements. And don't forget those measurements were made by hand and eye using some unknown micrometer.
Second, the cache chips do not get very hot anyway. If you look at an Athlon minus the plastic outer shell you can see that the raised metal doesn't even touch the cache chips, there is a 1mm chunk of heatsink goo providing all contact. The part that matters is the part right above the die of the Athlon. And according to ryemax's questionable measurements, it's 100% flat - of course it isn't, but it's flatter than he can measure.
Lapping can't really hurt if it's done properly, and not to excess, but I can imagine lots of people taking a piece of sandpaper to their plates and making it into a nice rounded lump. The Athlon heat spreader plate has great big pins sticking out of it, so you can't lay it flat on a piece of glass while lapping - obviously he didn't do that. You could glue fine-grit paper to a piece of wood, but would the wood be flat? And while I'm ranting I have to laugh at those articles that show a PPGA Celeron wrapped up like a mummy so it doesn't get wet during sanding - these chips are allowed to get wet! I used to build prototypes by hand, wash the rosin off with solvent and then lather the whole board up with palmolive and a paintbrush before rinsing in water and drying in hot air.
If you do lap a chip don't bother with 1500 grit, you'll be there all night. And don't expect this to help very much. If a chip won't POST or boot at a certain clock speed then it's not going to do it lapped either, all you're looking for is to keep heat down once it builds up, possibly avoiding a crash once you're up and running.
A word on Athlon heatsinks: the standard one generally works fine. Two great alternatives are the Alpha 7125, and the Arctic Circle. I suspect other Alpha heatsinks like the PPGA or FCPGA versions would also be good on an Athlon, but the mounting hardware is wrong. The heatsinks on offer at http://www.coolwhip.dk look like they are also good but fit in a more constrained space. They also have peltier, slot 1 and socket 370 models.
Overclocking is about more speed for less money just like putting a big rear spoiler on your Honda is about performance - it is a bit, under certain conditions, but mostly it's just about looks. It's about having shiny flat surfaces, a big shiny copper or milled aluminium heatsink, or big dual fans. No wonder there is a market for cases like the submarine. (http://www.nikao.net/) It's also about feeling like you're sticking it to a big company. Well, as long as you ignore the fact that a top heatsink costs up to $60, and a "golden fingers" voltage and multipler switch can also cost up to $60. But remember, it's not AMD or Intel that got the money.
I do it - I have a Celeron 400 that's happy at 600 with a nice Alpha heatsink on it, as a third machine. I've extended that machine's useful (not operating) life a little.
Most of us hardcores use copper, aluminum, or silver grease now :-) No dry out, and I'd like to see silicone beat those for heat transfer.
:P People will sand these things down to 2000 grit or higher, until you can get perfect reflections in the copper.
As for those of you who aren't sure about lapping, it's grinding down the rather large amount of extra metal on certain processor cores such as the Socket370 Celerons, flattening it to an incredible amount. Most people equally lap their heatsinks, and a few have gotten them so flat and clean that they got impact welded
Actually, looks like he is the real bruce perens.
Umm, its a joke, the chip never went that fast.
Please see my site regarding that.
Contact resistance is the problem here. Yes, making the surfaces "flatter" on the macroscopic level (.0005") will help, but there are still microscopic variations in surface height. You don't get enough ACTUAL contact. This is why you use a thermal grease or paste. Ideally you have a layer just thick enough to fill the gaps. Lapping the surfaces may help a little, but without that grease you are not going to get very far.
-Jeff Albro, BS in Mechanical Engineering.
Here's some actual honest to goodness research on the exact subject, cooling electronics. Imagine that.
-------- This space intentionally left blank --------
Sorry, you're correct: specific heat is energy required to raise a certain amount of a substance a given temperature. Thermal conductivity is the amount of heat that can pass through a substance in a given time. The two are related, but not the same. It's been a long, long time since I've taken chemistry. :)
kugano
Standard-issue thermal grease has a higher specific heat than the metals used on the heatspreader and the heatsink, so less heat is transferred between the processor and the heatsink. More/closer metal-to-metal contact is always a benefit.
kugano
The thing not mentioned in the article is that flat surfaces aren't needed, as you've noted with flexible materials, but matching surfaces. Two flat surfaces will do, but so will two spheroidal surfaces with the same radius of curvature. Or perhaps I should say opposite since one will need to be concave and the other convex.
The Truly Devoted (or Truly Mad) could go as far as abandoning the micrometer and going to optical tests and optician's methods for working the surfaces if they were sufficiently careful.
Myself, I'd use the heat sink compound and not go that far, but that's just me. There is a point of diminshing returns. Amateur Telescope Making Book I, Book II and Book III, edited by Ingalls, have much detailed information on shaping optical surfaces. I recall that Books I & II would be the more likely for this. The contact will get much better with optical working techniques. This can become a problem as the two surface worked against each other can become wedged with their adhesion if the material between them is permitted to dry out. I have wedged and unwedge glass disks. I'd not want to have to do the unwedging with a relatively expensive and small processor.
A note for anyone going to check out those books: They were written/editted with a 'revised' spelling, so 'technique' is spelled 'technic' amongst other oddities.
I don't subscribe to RMS's GNUtopian vision.
Don't mistake equipment junkies for audiophiles. By definition, audiophiles love sound. The best of them are willing to _listen_ (instead of measure or speculate) to find equipment that sounds good.
They'll never be convinced by some guy who says it _shouldn't_ sound better when they've heard that it _does_, green magic markers or no.
Take this troll elsewhere. Too bad I'm out of points for the moment.
Probably an optimal solution to the cooling problem would be to make the heatsink an integral part of the processor casing, with only the fan replaceable. Of course, this would make liquid cooling difficult, and would not be popular with over-clockers.
"Freedom means freedom for everybody" -- Dick Cheney
1) The stated value of measuring accuracy is bogus. For one thing, if he did it with a micrometer you are getting thickness of the plate, not flatness of the plate.
He used a dial micrometer, not a caliper micrometer. This accuracy is quite reasonable, especially using proper techniques (granite surface plate, etc.).
2) After lapping the processor, then letting it run to operating temperatures, I bet the processor warps some due to differential coefficients of expansion...
I bet not. If the processor is expanding and contracting enough to flex a reasonably thick aluminum plate, you've got another bag of problems anyway.
If you change the distance between the heat sink and processor only a little tiny bit, you only change the heat flow a little tiny bit.
Agreed. I found the article interesting and credible, although such methods are not necessary for 99% of average users/overclockers.
Of course, an overclocked SX couldn't run Quake, no a FPU and all... but funny none the less :)
I remember seeing that in a quicky awhile back... pretyy cool though, and completly relevant...
It actually radiates heat better if the surface isn't flat, because the surface area is increased. Sort of like the villi in your intestines, which increases the surface area for food absorption.
*sigh*
Got Rhinos?
Be careful with what you mean by "power" as the ARM is typically thought of as a lower power (as in P=IV) but high performance processor.
Now the ARM is also typical RISC with really not a whole lot on the die. Sure, you've got peripherals and misc but it's also been specifically optimized for lower power consumption. Does the crusoe need a heat sink? Probably not and given that it's supposedly as fast as and compatible with a PIII 500 , now that's a chip that we should be thinking about.
before you find that you need to overclock your RAM :)
========================
63,000 bugs in the code, 63,000 bugs,
ya get 1 whacked with a service pack,
--- Grow a pair, liberals... stop letting the Republicans bully you!
Compare this with hackers, who make a daily practice of getting their hands dirty in the SW. Most hackers prefer to run an OS which facilitates this.
Now compare this with automobilia. Some people can't be bothered to pump their own gas. Until the energy crisis of the 70s, many of these automotive lusers didn't even know HOW. Some automotive users know to check the oil now and again and are responsible about maintenance, but don't really care what happens under the hood. Some people are performance drivers who like to know what's happening inside the machine and who really enjoy a fine sports car.
Then there are the automotive hackers who rebuild their engines to relieve stress, and who perform such "risky" modifications as increasing the compression ratios and advancing the timing. They know what they're doing. They're confident they won't screw it up.
Oh, tangentially related: A specific, fairly recent example of processors being marked and sold at less than their full potential: AHX-core AMD K6-IIIs. This core was marked as a 2.2v 400MHz unit and as a 2.4v 450MHz unit. They were otherwise identical. So, if you bought a 400 AHX, cranked the core voltage +0.2v and the multiplier up by 0.5x, you had a 450 AHX without paying extra. Your overclocked 400 AHX had 100% of the stability of a factory 450.
That's why people overclock. Free (as in beer) performance.
J
Yes, the flattening effect is nullified unless you also lap the bottom of the heatsink. Yes, a little thermal paste accomplishes almost (but not quite) the same thing. BTW, most people put about 10x the grease on there that's actually necessary. The layer should be about as thick as a sheet of notebook paper. Much more and you're not helping, you're hurting.
The main benefits of lapping come from two points unrelated to the flatness of the contacting surfaces. When heat's conducted through different media, each change in medium reduces the overall conductivity. Some processors (i.e. Celeron) have jacket over the casing. The jacket and the casing are two different materials. If you sand the jacket away, that's one less medium to conduct through.
The other main benefit is from reducing the thickness of the casing between the processor core and the heatsink. If you're dumb enough to sand right through the core, you were too dumb to be taking sandpaper to your processor anyway.
Yes, I overclock. No, I'm not ashamed of it. I overclock a Win9x box used for gaming and for visitors who get freaked out when they can't find a Start menu. Since the stability of Win9x approaches zero, knocking it down a few percent farther produces generally no observable effect.
Now, you're all more than welcome to flame me (and what kind of place would Slashdot be if you weren't?). Before you do: Yes, I'm aware that overclocking and lapping generally shouldn't be done in an environment where stability is important. Realize that some of us prefer a side order of hardware hacking with our RDA of software hacking.
Note to moderators of this discussion: Posting something like
'Doing this would be fscking stupid. I don't really know anything about this, and I've never tried it.'
is called trolling.
J
from what i remember of my chemistry lessons SHC - specific heat capacity is how much heat you can "absorb" over time whereas thermal conductivity is how much heat can pass through the medium over time.
.oO0Oo.
A bit like the difference between a buffer (SHC) and a pipe (TC)
There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
hmmm,
.oO0Oo.
i think you need to put your ehad in the freezer for a while.
what's an erronious error?
an error made in error
it's make my exceptrion handler freak that's for sure
try:
hello:
!"$!$!$!"£$%"
except:
goto hello
There are places where the networks are not touching,and there are places where they are-Boeing's Lori Gunter
But when it comes to heat transfer, metal on metal is better, yes? Especially compared to metal near metal with air in between.
Thermal grease was invented to fill in those gaps with a solid, heat transferring substance, rather than air, which tends to hold heat when placed between two 'uneven' slabs of metal.
If you can reduce the 6/10 of a thou that the thermal grease would occupy by even 1/10 thou, that gives more metal to metal contact, improving the efficency of the heat transfer. As you say, the fabricator in Taiwan isn't going to make it that accurate. They only make it accurate enough to justify costs, and within tolerances.
If we go and make the tolerances a little tighter, so they work better under our heavy heat stress, it's woth it to us.
"History doesn't repeat itself, but it does rhyme." Mark Twain
Happy overclocking! I've got my Athlon "700" running at 900, and it maintains 30c!
A bit of advice - always spring for a 300W+ power supply with Athlons...400W if you plan to use the Peltier!
"History doesn't repeat itself, but it does rhyme." Mark Twain
with cool eyes. that is all.
As the human head radiates approximately 75% of all our expended body heat, we could use the lessons learned from this site to efficiently cool our pointy noggins, so as to allow us to overclock our craniums to the maximum extent.
At the very least, cranial cooling will reduce the number of times we product erronious errors, or outright crash.
Damn, I forgot to take my medication again.....
Feed The Need[goatse.cx]
Since when has slashdot started posting front page articles on old-school hacks like this? Lapping is one of the oldest tricks in the overclocking books, and hardly worth such a mention. Posts like this should be reserved for sites like HardOCP and Ars and the whatnot, not slashdot. Just my $.002.
Mike Liska, Electrical Engineering Technology and Computer Engineering Technology Undergrad, Purdue University.
You don't actually want to lap the heat sink flat, you want to scrape fit it to match the processor plate. You want look for gaps between them with clamps in place, preferably at something close to operating temprature. (as both the clips and the results of heating a non-constant thickness object will change the shape) For those that haven't read "Machine Tool Reconditioning" (published in 1936, and never superseeded), you need a marking compound, and a scraping tool. Apply the compound (these days usually a blue paste) to the cleaned surfaces, and assemble, pressing them into contact with each other with the clamps only. Separate, and "read" the spot pattern. It will show you the places where stuff doesn't touch. Use your scraper to adjust. Repeat till you get a reasonable bearing surface. The surface won't be "flat" like you would get with lapping, but the two parts are matched to each other, and mate without gaps. Scrape fitting is a whole lot faster than lapping. (and with lapping, you would still have to use the marking compound to tell when you had the parts flat enough to fit together, and that you wern't lapping a bow into the surface.) I am guessing that they used a "Dial Indicator" instead of a "dial micrometer". The only micrometers commonly found with a dial on them are so-called "indicating" micrometers, used primaraly for comparison measurements. (besides, a micrometers contact spot is too big to give the number of spots they indicated)
Organizer:New England Rubbish Deconstruction Society;The NERDS,first US team in the UK Scrapheap Challenge/Junkyard Wars
Great mad hatters think alike. To stop evaporation, I wonder if you could use an amalgam, perhaps as dentists use? Maybe you could apply some gold leaf, then add a tiny bit of mercury for luck.
Your analysis is naive.
In order for the heat sink to bow it must be constrained on both sides. I have NEVER seen a heat sink that wasn't free to expand in the plane of the processor/heatsink interface.
Even if the heatsink was constrained in this manner, the expansion would then propogate as stress in the material that is constraining the sink itself. The stress on the heatsink could indeed cause the sink to bow out, but the amount of bowing is now a function of the strength of the heatsink structure. It will be _lots_ less than what you are estimating.
This is not good metrology. A dial micrometer?!? How does one get flatness (3D) with a dial micrometer (1D, no reference plane)? To do this properly you'd need to use something very accurate with known reference axes like a Zygo white light interferometer w/ stitching or long stroke profilometer w/ y indexing. You could use a CMM as well but it'd probably be less accurate unless you have a really expensive CMM.
And you need that equipment just to get the data! After that you have to pump it through some analysis algorithms that can do orthogonal least squares plane fitting.
Then, finally, you could make some inferences on the geometry of the part. Not that it matters, though. Neither the plane nor the heat sink are going to be perfectly flat. You have to count on the thermal conductivity of the heat sink compound at some point.
BTW, My phd research deals with 3D metrology. I have access to all of the equipment and algorithms to do this right if anyone wants to send me a cracked open processor.
Andre Claudet
Computational Metrology Research Group
Manufacturing Research Center
Georgia Institute of Technology
The difference is that there was not a version of Doom (Until Doom 2 anyway) that was better than Doom, sold by ID. By contrast, there is a version of the P3-550 which is better (Called the P3-600, or P3-750, or whatever) made by intel. They have no motivation to let you get the functionality of the P3-750 on a P3-550, hence, their lack of desire to help you overclock.
Repeat until learned: Hardware is not software.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Okay, I don't know if anyone will see this... I lost my earlier response to this post to network problems and my own impatience. I'll try again now, belatedly.
There are two classes of upgrade part you can buy for your car, more or less; OEM, and Third Party. If you put a Third Party (Or at least an unlicensed Third Party) addon onto your car, you void your warranty; For example, if you put a turbo on your 5.0.
However, some companies have approved add-ons. For example, you can spend $5,000 and get a Turbo Kit installed on your Boxter, making it something other than a chick car. This does not void your warranty. It is essentially an OEM part.
Car makers want to make money off of the upgrades same as anyone else. Some cars are pushed right to the line, and to upgrade them you have to replace the engine, which cannot legally be done in some states (Including California, where I live.) If you want to modify a car in California you have to either have a pre-smog (pre smog check) car (Currently I believe the cutoff is 1973) or get an exemption; I don't know if the latter is even possible any more, though I do recall something about having to get a NEWER engine (with ALL smog gear) than the original motor if you are not replacing it with the stock part.
I have an AMD k7-700. I haven't gotten around to overclocking it yet, but it's in my plans. However, if the 650 and the 700 could be overclocked to the same final clock rate, would I have bought a 700? Probably not. Of course, I'm sure someone will end up telling me the 650 is a better process and can be overclocked farther or something, but the point is, people buy faster chips pretty frequently, and pay more money for them. This is how some people have 1ghz chips ahead of time. However, to overclock the chip voids your warranty, and that stops others, and those others will pay for the 1ghz chips.
As for the assertion that you sell more cars for their upgradability, and therefore the same is true of overclockable CPUs, that is probably true. On top of that, all a chip company has to do to have multiple grades of chip is to set them to not take a higher clock rate, or only bind the pins for a certain multiplier, or both. The last prevents overclocking, the others make it less effective.
Anyway, the car company has a faster model and sells it to you for more. So does the chip company. Both of them void your warranty if you push it past the specifications they gave you. Where's the dissimilarity here?
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Well, if you really want good thermal contact, the best thing to use is indium, a soft metal. To make it work ideally, you'd remove the paint (or anodization, though Al2O3 has pretty good thermal conductivity) and use pressure. The indium fills in all the voids and eliminates a great deal of the thermal resistivity of the interface, which is probably as great as that of the thermal goo.
This is not really a flame per se, but a philosophical musing about overclocking. I am not technical enough on the topic to question your methods, so I am just pondering the motivations.
Overclocking has always seemed to me like the computer equivalent of base-jumping. People who do it either like the thrill or feeling of accomplishment (and some just plain have no idea what they are doing). To me, it has generally been rendered unnecessary by this combination of things:
(1) processor power increases frequently enough for my tastes;
(2) I do not require more processor power for my usual tasks;
(3) there is a significant learning curve involved with overclocking;
(4) new processors cost money when ignorant people like me torch them;
(5) I find no particular attraction in that area of hacking/optimization;
(6) processors are one area of computing that are generally rock-solid. Since I run a few Winblows machines and a lot of *nix betas, extra BSODs or core dumps are not appealing to me.
So I am asking (not being a dick, honestly interested) why people overclock? What is the greatest attraction? All respondents, legitimate flames, etc welcome.
-L
Good insight, thanks.
(1) so basically it is like most things in life - knowledge leads to monetary reward. Someone experienced in O/C'ing saves a buck because they are able to minimize or negate risks associated with an unskilled O/C'er "taking his chances". They get away with maximizing a processor;
(2) more power like hot-rodders get from their cars. You can buy a leading edge processor, but it costs 4x as much, and maybe does not "feel" all that fast;
(3) I will take your word for it. I hear people talking about polishing, grinding, etc. I understand the physics, but do not know the "tricks of the trade" nor do I own the specific equipment, the deficit to which is part of the barrier to entry;
(4) I probably can be satisfied, but it is more out of ignorance of what my hardware really could do if pushed. I do like to hack and massage on the software side, and I do not presume that you can't do the same for hardware;
(5) I have heard that. O/C legend says that your average Intel is 20%+ below capacity without much additional heat management. I have heard astronomical figures where good heat management is used;
(5b) Intel 1GHz had the feel to me of being a total market bluff anyway to steal AMD's thunder. Good old FUD!
(6) I definitely respect it as a hobby. I have friends who O/C everything they can on their machines, and they like it. I trust them enough to believe it makes a decent hobby.
-L
Wow.. on my old K6 all I have to do is underclock by 100Mhz and put a couple of extra fans in the case -Works like a charm!
-
air and light and time and space
I think they do. It tends to make them look bad.. I can see their point of view: "WHo cares if it is a little unlevel. Its 'designed' that way. Thats a feature. If we wanted you to be able to mount any other heat sink and be able to overclock it until it burst into flames, then we would have built it that way. But we don't, so there!"
/. is a commercial entity. goto slashdot.com
Sorry, I forgot a factor of 1/2 when I wrote that post. It was 5:40am (geek bedtime) when I wrote it, so by the time I realized it, I was in bed.
The formula should read:
B=(1/2)*Square root of ((L+sigma)2 -(L)2)
So the result should be
B=.045 rather than B=.09
.045 inches is still a very large degreee of bowing, however, so my point still stands. If bowing of the heat spreader occurs then it will cause large errors in surface flatness.
BTW, I am talking about bowing of the heat spreader not the heat sink. The heat sink is a thick piece of aluminum and it not directly fixed to the processor assembly. It's not going to deform. The heat spreader, which is attached to the processor assembly, is much thinner and depending on where it is attached it could warp somewhat.
The Eponymous Mallard
graccito ergo sum
I quack, therefore I am
My analysis would be naive if I were talking about the heatsink. But read the web page that started the discussion. They didn't measure the flatness of the HEAT SINK. They measured the flatness of the PROCESSOR HEAT SPREADER. It's an entirely different object.
The HEAT SINK is a bulky piece of aluminum and is not fixed to the processor assembly. It's not going to bend or bow. The HEAT SPREADER is much thinner and is attached directly to the processor assembly. Depending on just how it's attached, it might bend or bow under heat stress. I don't have an Athlon here I can take apart to see where the attachment points are, but it's something worth looking at.
The point of my post is that IF bowing occurs, it has the potential to be a surpisingly large effect. To see what I mean, take two pieces of paper or index cards and place one on top of another. Then, keeping one edge in place, move the other edge of the top card a fraction of a millimeter towards the other. You'll see the card bow out to a much larger distance than you moved the edge. Bowing magnifies heat expansion differences.
I don't take my numbers too seriously. They are just a back of the envelope, order of magnitude calculation, to show what effects thermal expansion can have under certain circumstances. The only way to know if this actually does happen is to measure the flatness of a hot heat spreader in situ.
I have some experience with heat deformation. I did an internship at the National Insitute of Standards and Technology (Metrology Section/Optical Physcis Division) working on computer-controlled measuring machine that required a large flat steel plate as its base The major problem was that any slight temperature gradients in the plate caused microscopic deformations that affected the precise measurements we were making. We had the plate covered in a grid of tiny thermocouples to ascertain and eliminate or compensate for the temperature gradients.
As I said, just because something is flat when it's cold, doesn't mean it's flat when it's hot.
The Eponymous Mallard
You can't accuse me of "ducking" the question.
If it's flat when it's cold, is it flat when it's hot?
.001 inches.
The coefficient of linear thermal expansion is:
Aluminum: 25 x 10-6 (C)-1
Silicon: 3 x 10-6 (C)-1
Difference: 22 x 10-6 (C)-1
If you attach a sheet of aluminum to a wafer of silicon the aluminum will expand more than the silicon. The difference in the expansion will be 22 x 10-6 of an inch per inch per degree centigrade.
For a 4 inch piece of aluminum heated ten degrees the difference will be approximately
If the plate is fixed at the ends it may bow out. The amount that the bowing will pull the aluminum away from the silicon is approximately:
B=Square root of ((L+sigma)2 -(L)2)
Where L is the length and sigma is the expansion difference. For sigma much smaller than L this is approximately:
B=Square root of (2L*sigma)
If L=4 inches and sigma=.001inches
B=Square root of (.008)
B=.09 inches
That's almost a tenth of an inch. That's on the order of fifty times larger than any surface imperfections.
Now, I've admittedly made a lot of simplifying assumptions in my calculations. Some of them, if anything, underestimate the bowing factor.(Only ten degrees above room temperature--that's a well cooled processor indeed.)
But the magnitude of the effect of any bowing due to differential thermal expansion is so large, that if it does occur it would dwarf other departures from flatness.
So attention must be paid to how and where the alumium is attached to the silicon (as well as joints between other substances) to be sure such thermal bowing does not occur when the computer is actually running.
The Eponymous Mallard
If it walks like a duck...it's the Eponymous Mallard
...the one about quality control at NASA, also posted today on /. Sure it's all fine and dandy to make something of passable quality. But wouldn't we all like to like to see AMD (and other companies) go the extra mile to ensure very good quality in all facets of their products. Sure the Athlon is a good processor, but what good is it if it can be subject to heating problems because of an avoidable design oversight?
Sure, it can be compensated for by applying a conductive paste, but someone should have seen this at the design phase.
Microsoft got really rich when it decided that passable quality was good enough. Others copied this philosophy and now we have bloatware
1) To save money.
2) Some people want more power, just like some people want to knit quilts. Why the fuck would anyone want to knit a quilt? I don't know, it doesn't matter, and I wouldn't enter a quilt knitting discusion to ask.
3) The learning curve is not that steep. What is steep is the equipment often required. But O/C'ing is a hobby, cost isn't the only factor always.
4) Without hacking/optimization you would NOT be satisfied with the performance of any machine.
5) Processors aren't just generally rock solid, they are also sold at below their capacity - at least by the reputable vendors (the old Intel never would of released that POS 1 GHz proc).
6) Like I said: it's a hobby.
What difference does it make as long as you use the thermal conducting grease that usually comes with a good-quality heat sink/fan?
--
Sometimes it's best to just let stupid people be stupid.
the last thing you want to do is to start exposing the electronics, by being a bit too enthusiastic when you are polishing it flat.
that would be a definite bummer
"It is a greater offense to steal men's labor, than their clothes"
they tried various options with good results. They had the computer running in the freezer
Towards the end, things got hazy. To quote:
some very interesting photos too
"It is a greater offense to steal men's labor, than their clothes"
There are powerful processors which don't need heat sinks at all (see the ARM series, for example).
My two cents.
-- javaDragon is an instance of JavaDragon.