a64 - heatspreader = slight problems

C

(cf)Eclipse

Freelance Overclocker
Joined
Feb 18, 2003
Messages
30,027
ok, to follow up on tenchi's post on his process of removing the heatspreader from his a64, i planned out the same. i have about half of my data, but i can't get any further for one reason:
my core isn't flat
and i don't have sandpaper here in my dorm to fix it.

now my thought was that this would explain why some people have better temps given the same setup than others. the chips with flat cores will definitly get better contact with the core than the ones with a slight round in it.
and a word of warning for those who take the IHS off, it's difficult to remove fully. i was able to get about 1/2 of the silicon type glue off with the knife i had. ironically, i used a sharp edge of an ecplise gum package to cut the rest of the glue.
then i bent the metal "clamp" that tightens my zalman heatsink down to compensate for the missing IHS.. it seems to be giving the proper amount of pressure, but i definitly don't feel safe moving my case around too much ;)

once i get ahold of some sandpaper, i'll update this and give you the results of my heatspreader removal, cause right now, my load temps are like 2ºc higher than before (after taking ambient into consideration), and that's at the low power setting (2ghz, 1.3v = ~52 watts) i'm afraid i might kill something if i pump up the voltage without full contact

:p
 
If by “core” you are talking about the actual CPU core itself, I’d have to suggest you think again about “sanding”. The core is the actual processor part; even a slight chip can destroy it.

Sanding the top just doesn’t sound like a good idea at all. It’s entirely possible that just under the surface you would be sanding silicone. Working silicone. Sanding working silicone would no doubt render it non-working silicone.

Just something to think about.
 
nah, no pictures.. i'll try to remember next time i have it apart. and the roundness seems like the edges are about 1/10 mm lower than the middle.. very slight, but enough that my heatsink will spin in it's mounting bracket at the slightest touch
and i'm aware of that possibility bill.. hence why i'm slighty hesitant about it, however, the core is a really shiny metal that is pretty rough looking. nickel i think. and i don't plan on sanding it too much.. maybe like 2-3 passes on 1000 grit with a few more on 2000 to extra smoothness... if that doesn't do it.. i'll think of another solution, though i do recall some people lapping the old k7 cores without ill effect, so..
 
(cf)Eclipse said:
nah, no pictures.. i'll try to remember next time i have it apart. and the roundness seems like the edges are about 1/10 mm lower than the middle.. very slight, but enough that my heatsink will spin in it's mounting bracket at the slightest touch
and i'm aware of that possibility bill.. hence why i'm slighty hesitant about it, however, the core is a really shiny metal that is pretty rough looking. nickel i think. and i don't plan on sanding it too much.. maybe like 2-3 passes on 1000 grit with a few more on 2000 to extra smoothness... if that doesn't do it.. i'll think of another solution, though i do recall some people lapping the old k7 cores without ill effect, so..
Click to expand...

I know you are a pretty “aware” guy, I just had these bad visuals of someone sanding away something important, like the FPU part…..;)

Luck
 
The important stuff is on the underside of the silicon. You could take off a few tenths of a millimeter without killing it (not that I recommend the idea).
 
BillR said:
I know you are a pretty “aware” guy, I just had these bad visuals of someone sanding away something important, like the FPU part…..;)
Click to expand...
haha, that would be so funny if someone managed to have a cpu that couldn't do any fpu work, but otherwise ran fine
:D :D :D
 
What now people are lapping their cpu cores..........oh man... :D ;) :p
 
darktiger said:
What now people are lapping their cpu cores..........oh man... :D ;) :p
Click to expand...

Oh I've seen this done, even back in the Coppermine days - not that it ever helped much . . .

I wouldn't do this unless you are 100% sure that A) there's a low chance of damaging the processor and B) you absolutely have to. Or of course there's C, which reads "to hell with it, I wanna try this!"

Which BTW, I completely understand :D
 
interesting. no lapping and the load temp at 2.3ghz has gone from 52c to 42c under full load. i didn't change a single thing. i blame the as5 settling :D i expect to drop another degree or two over the next couple of days as it goes through it's last few setting thermal cycles
but anyway, now i have some reasonably valid data to give, and i must say, while it looks good, it's confusing

dT.PNG


and for easy reading, i made a chart of the change in temp between idle and load with and without the heatspreader so that the point of this is really clear

chart.PNG


basically the temps drop a bunch without the heatspreader. i was expecting the difference to be most noticable when putting out the most heat, but for some reason, at the lowest speed i tested, the biggest change in temperatures between the ihs and w/o the ihs showed up. not really sure as to why this happened. i'll run the tests again for a third time sometime tomorrow after finishing up my english paper ;)
 
WOW!

interesting about the core. you should have taken some pics of the non-flat core.

basically the temps drop a bunch without the heatspreader
Click to expand...
You're damn right they do, thats why I did mine so long ago. I can only imagine with water cooling. Mine went fwent from 38ºC idle with IHS (70ºF ambient) to 32ºC. I rarely see 40ºC (except on hotter days) under full load. My Newcastle still runs 2.31ghz (210x11) on the IHS-free MCX6400V setup.

Well welcome to the club, now I wonder who will be the first to f*** up their core b/c of posts like ours :D piquing their interest. Good luck with the sanding (if you still have to do that).
 
Tenchi4U said:
interesting about the core. you should have taken some pics of the non-flat core.
Click to expand...
yeah, i'll do that when i get my hands on some really really find sandpaper to take a wee bit off the core. it doesn't seem to noticable on the core itself, but if i look at the ihs carefully, the stock thermal paste is thinner in the middle, and there is a point on each corner where it looks like the corners turned up. here, i took a picture for you


(click for bigger if you wanna see in detail)
 
I'd be careful about speculating about the flatness of the core by looking at the IHS. Remember, the IHS is manufactured by making it shit hot and smacking it with a die... it's very possible that the IHS *itself* isn't perfectly flat.

Also, the X pattern that the interface material leads me to believe that there's more to it than flatness - it looks to me that the X pattern is a result of the IM being squeezed from the center, and the thinner edges a result of gassing out in as it dried (that's why the corners are thinnest...the corners don't have that blob of excess goo like the edges do, so it has more ready access to the surrounding atmosphere to dry out...)

Remember - the procs are made from 99.9999999999999% pure silicon wafers, and you can rest assured that those babies are unbelievably smooth. I've said this before in a similar thread long ago - if AMD can make a transistor gate 90nm wide, they can make a fucking silicon wafer flat.

The only way you could tell, would be to hold a razor blade edge-on against the core. I'll put money on it that you won't see a curve.

Just leave the core be.
 
when i have the heatsink off, i'll take a picture of the wee scratches in the center of the heatsink that weren't there before i took off the ihs. i still think the core isn't quite round because of those scratches, though what you say is true.
and remember, this isn't direct copper -> silicon contact. amd puts a layer of metal on top of the cores to strengthen them some. (nickel i believe.. dunno why though)
 
I think those scratches are simply caused by slight rocking on the edge of the core as you mount/ dismount the HSF...most of my cpu's have done that.

lol @ blark..Why didn't you just call yourself Eclipse?

You should get a razor blade and hold it against the cpu edge... best way to see if there's a contour.
.
 
check eclipse's post count. i figured, why not, i'll leave it there :D

and when i get my sandpaper today, i'll have everything apart so i can really have a good look at it. i know what you're talking about with the scratches though, my old barton did that too. however, this is different, it's not at the edge, it's dead center. *shrug*
either way, we'll all find out in about 2-3 hours ;)
 
Sounds like your IHS removal project is turning out to be much more eventful than mine :p

Mine was somewhat boring by comparison:

remove HSF, remove CPU, cut off IHS, clean core, apply AS5 to core, install cpu, remove nylon spacers from IHS setup to compensate, install HSF, plug in fan, power-on, test, close case, enjoy IHS-free livin' :D

no weird anomalies (concave core), no sandpaper (except when lapping HSF contact area)...nothing.

sometimes boring is good I suppose ;)

Keep us informed...don't f*** up your proc :cool:
 
(cf)Eclipse said:
haha, that would be so funny if someone managed to have a cpu that couldn't do any fpu work, but otherwise ran fine
:D :D :D
Click to expand...

THIS actly happed to my Athlon 700 Slot-a it had 2 fans on the HS and one died and just the FPU burned up no clue how it was just FPU but it booted windows and ran any thing that didnt need FPU fine the sec. i tryed to do some thing that used the FPU it would lock up drove me nuts for a week >.<
till i tryed playing and MP3 with winamp with FPU opts off and that worked then turned them no BOOM freeze wird :eek:

Edit: that CPU is now a key chain lol i pulled it apart to you can see the core and PCB its on
 
all right. i was hoping the sandpaper would get here today. i have between 220 and 400 grit, but that's not fine enough, so i had sent out for some 1000 and 2000 grit (thanks mom! :D )
and personally tenchi, i dunno what's better. it's kind of a pain to go through all this not really knowing what's safe and what's not. but hell, at least i'm man enough to do it, and now people will know if it's safe or not if done properly (i would hope so, i'd rather not have to buy a new cpu, even if the one i have oc's terribly) ;)
since i have class from 12-2 tomorrow, i'll probably get it smoothed out after that, but before 4 (i'll try to find a razor and take a picture for you phyber)
 
Dude, your going to hose your CPU. Don't lap a CPU core. Wether or not you think its flat it is. If you messured it its probably within .00001 at any given point on the die.

On a side note if anyone else is planning on removing the heatspreader. Just buy a mobile CPU they come without one stock.
 
at least i'm man enough to do it,
Click to expand...
are you questioning my manhood b/c my IHS removal came out to be event free (just by chance, not my doing)? :D j/k

hope all goes well...be sure to post pics and result of the final product...
 
haha, nah.. you did it first.. i'm just having annoying complications :cool:

and king: i'm not too worried if i kill it. sure, i'm gonna be out of $200, and a computer for a week, which will suck... but it's all part of the learning process, and letting others know what can and can't be done with the appropriate set of tools ;)
 
I agree, life is too boring without risks.

"It seems to me that people have vast potential. Most people can do extraordinary things if they have the confidence or take the risks. Yet most people don't. They sit in front of the telly and treat life as if it goes on forever." - Philip Adams
 
Tenchi4U said:
I agree, life is too boring without risks.

"It seems to me that people have vast potential. Most people can do extraordinary things if they have the confidence or take the risks. Yet most people don't. They sit in front of the telly and treat life as if it goes on forever." - Philip Adams
Click to expand...

You two are like the people from scary movies that hear a noise and have to go and investigate the situation. Then everybody dies b/c you guys need to take risks. Me im that brotha thats a'ok with his uneventful, risk free, extra 7C IHS cpu

Like the Russians say, "If its not broke dont fix it" :D
 
haha, to sum things up quickly, i'm a mechanical engineer major.
i see something, i have to fiddle with it. ;)
 
thats pretty funny, but not an entirely accurate metaphor.

I gauge my risks. I'm pretty sure nobody is going to die if I remove my IHS; if I'm in a scary movie and something makes a sound, I'm gonna find a gun and start running.

I'll risk my own property, but I will not risk others lives or property (w/o their permission). :D
 
so if someone says "hey what was that scary noise that sounded like a knife slicing flesh(or a gun being locked and cocked(nothing else worked here. i didnt want to rhyme)), lets go check it out!" youd go even though you BOTH thought you might be killed :confused: ;)

hell id remove my IHS if i had the balls to do it. but the second reason (next to speed) im going A64 is im sick of cracking AXP cores. and this would just bring my problem back :eek:
 
I diddn't realise CPU's became a life or death issue lmao

If you do go foward with it, best of luck....a downed computer for a week or two sucks.
 
exactly!

just a quick update for you guys, i have to go to the hearing therapist today after class (long story, i'll explain in a pm is anyone really has to know :rolleyes: )
so i dunno if i'll have the final data by tomorrow or the next day. (i've decided to give the as5 a few thermal cycles to set before actually trying to get data)
 
Do you have any special types of breathing methods or a pre meditation thread to prepare our minds before we attempt this task?

I plan on doing it to my ghetto week 51 3000+ to practice. If I beat anyone to it, i'll have some pics up on this :)
 
hehe, nice bump..

however, do not drink/eat anythign that has sugar or caffine. still hands are key.
and have a foam pad to stick the cpu in to avoid bent pins.. i had to straighten out quite a few when i was done.
i actually ended up cutting off the ihs with a piece of plastic due to no suitible blade, but it took a really long time.. and pissed me off. haha. just make sure you don't go in too far and cut the caps off the package.
 
estimated time of completion? how far do you cut in?

check out this lap job, non core lap though, just thought i'd add this.

lapped.jpg
 
BillR said:
If by “core” you are talking about the actual CPU core itself, I’d have to suggest you think again about “sanding”. The core is the actual processor part; even a slight chip can destroy it.

Sanding the top just doesn’t sound like a good idea at all. It’s entirely possible that just under the surface you would be sanding silicone. Working silicone. Sanding working silicone would no doubt render it non-working silicone.

Just something to think about.
Click to expand...
You think they don't lap the cores at the fab? They come off the packaging machine flat and shiny like that.. :rolleyes:
 
heh, they could definitly do a better job at putting on the coating..
at least now i know how far too far is when lapping a core. :D
 
check the thread date. this has been pulled up from the depths. this is a continuation of this thread with some real data/graphs and a whole bunch more babbling from me about why my cpu sucks
 
Yeah, my ClawHammer was definetly better. Then again, I was much more voltage-happy too :D
 
yup, voltage doesn't do much for me. to go from 2.3 to 2.4, i need to bump the core voltage .2v to 1.75. by then, the chip it putting out enough heat to make me unhappy, even if i think the sensor is reading high.
 
Hmm? Mine did 2.4 at 1.75 too. Well, actually 2.48 but I ran it at 2.4 for great justice. Oh i just remembered; you still have MSI Neo Platinum. So you're pretty much limited to 1.75vcore. Ditched that board a while ago :p
 
You must log in or register to reply here.
Back
Top