I'm wondering why dont you just sand off a mm from the IHS. Its getting lap and the heat travels a mm less. Maybe its worth trying.
And aswell it wont be as fragile as with removing the IHS.
Printable View
I'm wondering why dont you just sand off a mm from the IHS. Its getting lap and the heat travels a mm less. Maybe its worth trying.
And aswell it wont be as fragile as with removing the IHS.
I hope you know the TIM amd uses is used to FILL VOIDS in the imperfections of the core..
They use a chalky like TIM that is alot better than AS5/Ceramic for large voids "dips and so on" to level the surface of the core itself ;)
thats why some see no gains or "LOOSE" clocks from removing th IHS..
ones that will gain are ones that have perfectly flat cores so the as5 or ceramic does its job awsome..
Trust me it can "MAKE" a cpu or it can "BREAK" a cpu's max clocks. ;)
If it clocks worse you can't jsut add the IHS back on as you dont have the TIM AMD uses ;)
Quote:
Originally Posted by sandybeach
There are not supposed to be any imperfections on core(maybe there still are) as the cpu manufacturers "grow" diamond on it, and its flat like hell :)
Maybe the differences are all depending on the mood of master Gong Ciao Peng while puting(putting?) on the IHS :D
I can say from exp. that I have a machined to .0001 inch copper sink and on a few of my fx/a64 chips after you remove it you see the voids "spots" left on the core and HS from the viods ;)
the ones that clock highest and take the voltage and don't change temps quickly under load have the perfectly flat cores no voids ;)
Ive used about 30+ a64 chips almost "ALL" the IHS has came off, hehe
the 90nm winnys have the closet to void free cores :)
what do you mean with void free cores?Quote:
Originally Posted by sandybeach
and the chips that change the temps under load slower than the ones that jump up higher are better ocers? do you think the cpus that get a lot higher temps all the sudden when put under load have bad contact with the ihs and thats why the temps go up so fast?
Quote:
Originally Posted by sandybeach
Sanding the die with toothpaste ?...
Well I cielled at 2.6ghz before ..no matter the volts
Now
http://199.202.73.113/mb/0365578/IHS/IHS_P95.JPG
I STILL don't understand why my load temps are so fricking high...47°c !
Set vCore to 1.225 + 133%.
I'll check my WW for clogging .
1.225V + 133% = 1.63VQuote:
Originally Posted by nOx34
47C at load is far from high for air cooling. If you're checking temperatures with a software application, the readings are inaccurate anyway. ;)
I'm watercooled , see sig :p:Quote:
Originally Posted by conrad.maranan
Well, if you're relying on a software-based application to check your temperature readings, you might be better off just taking a guess. I recommend a CompuNurse Advanced digital thermometer. They usually run for less than $20. :)
I'll chek it out thanks buddy :toast:Quote:
Originally Posted by conrad.maranan
Here's a link to the unit that I'm talking about.
Update , 2.7ghz broken . Stable
http://199.202.73.113/mb/0365578/IHS/IHS_P95_2.JPG
This is great ! :banana:
If I could say that its mine, Mmmmm... I would be a happy overclocker :DQuote:
Originally Posted by nOx34
Hi,
i've read every thread in this forum which is about this topic.
but i still miss few facts.
1) is there anything (exept the deepth of the cut) that i should take care of?
2) is there really an improvement of the clockspeed or "just" the temp?
i wanted to know these things before i destroy my a64 :D
now it's win/game/3d stable at 2700mhz but not prime95 stable!
would it be reallistic to expect an improvement of at least 50mhz?
some said the gained 150mhz! that's an miracle, isn't it?
and sorry for this thread. but i couldn't find these answers in other threads.
mfg
The most important thing is not to scratch the surface of the chip with the knife.
I did read about people using dental floss rather than a knife. (I dont remember where)
You should get some sort of an overclock increase, its hard to say how much. It depends on how "bad" the contact is with your IHS.
On some motherboards the highest part of the CPU socket is higher than the CPU core so you might need to sand the socket down a bit to get a good contact.
Good luck and let us know how you get on.
thank's
i think the contact won't be a problem cuz i use an adjustable watercooling without this strange rentation module...
mfg
im going to do the same thing as soon as a venice replacement is available just in case hehe good luck and report on your findings
________
MEDICAL CANNABIS
yap
i'll keep you informed!
cu good n8
mfg
I dont think you understand me.Quote:
Originally Posted by killingspreez
The top of the SOCKET might be higher than the core.
The white bit that holds the cam end of the ZIF socket lever.
This is not a problem on all boards, but it was on my DFI UT and ive read of similar problems on other boards, ic 939 ones.
The height difference is only a faraction of a mm but it does affect the mount.
umm now i understand......i'll have a look at the weekend!
thx 4 advice!
mfg
It will drop your temps by about 8-12 deg c...you might see a clock increase but it is not set in stone...i my self never say a temp drop....
Did you put back the IHS? Or did you just leave it naked?Quote:
Originally Posted by IvanAndreevich
Is the core too fragil? Im not worried on messing it up doing the mod, but im worried of braking the core installing my waterblock.
i think thats a little optimistic, i havnt seen anyone achieve such drops yet, maybe 3 - 5*C at the mostQuote:
Originally Posted by FireDragon
ive seen a temp drop of 6C so about that much :woot:Quote:
Originally Posted by Jamo
Quote:
Originally Posted by Jamo
i got 10c. But i knew i would, as voltage increases @ cpu did not increase water temps. :fact:
hi all
i've made it!
http://server2.uploadit.org/files/ki...ez-yyyyyyy.JPG
no troubles at all. it was a little bit tricky but it is possible to do!
well now the default-v-core looks quiet well :)
http://server3.uploadit.org/files/ki...ez-newvolt.JPG
tomorrow i'll go for the 27xx mhz i hope!
mfg and thx 4 advices! the razorblade was really helpfull!
http://www.hardforum.com/showthread.php?t=851884
http://www.hardforum.com/showthread.php?t=857220
that was a while ago.
also, here's what my core looks like now, after lapping it :D
http://www.freewebs.com/ozzimark/core.JPG
wow, how much did you take off? also, that origami swan on your desk, I have two. About three years apart. :)
why did u lap it, was there a scratch?
my core isn't very even, too.
but 2730mhz prime/pistable is okay.........i hope the cooler fully touches the DIE.
-_-
mfg
Nice one ozzimark, tht takes some balls
:D
My 2.7G chip has died after a bios flash and I have a new chip with me now, tis a 0502 BPMW one and is currently at a prime stable ceiling of 2650mhz with 1.73 volts, watercooled.
The temps are arnd 43-45C full load, and im thinking of taking the shim off again, only problem is if im unfortunate enuff to kill another CPU with bios flash I may be able to RMA it, by removing the shim I cant.
So the question is, (and I should know the answer too this) should I take the shim off this new CPU? Will the temp drops equte to an increase overclk.
I just wished amd had designed the shim to have gaps in the side, so tht the heat building up within the shim would have an outlet to escape!!! I also thought it may be possible to pass a tiny tube between the gap in the shim (yup as u peeps know there is a small gap in the center on one of the shims sides) so tht air may be passed down here cooling the core. Very difficult or improbable to do but would be an idea if someone had the right tools
mong
eh, probably like 1/4 millimeter? not muchQuote:
Originally Posted by coop
i did it cause the core was really rounded. i could place it die down and spin it like a top. plus, my heatsink was rocking whenever i bumped into it. i was honestly surprised at how.. un-flat it was. so i took a few passes of 1000grit then a couple more at 2000 to try to get everything flattened out. sadly i was afraid to use water, so the surface isn't terribly shiny when it's cleaned off... though it seemed to work pretty well after it was flattened out. :toast:
disclaimer: if you do this too, don't be surprised if you kill your cpu. i was very lucky
That is what was a little startling was the color. Not that I knew if they were polished like an XP or not. I have seen pictures of what is inside of the chip, but do not know where all of the components are located in there. Like how far down, are they in the center or torward the bottom or layered throughout. I'm glad your works, have you been able to see any difference yet. This also brings up something I've thought about but never to the point of doing any research. Inside of a ram chip, what is in there? Is it just material or is there actual structured construction of columns and rows. All the legs on the sides, do they actually connect to a column? Is the chip separated into columns and rows or does the electrons just behave in a manner that keeps everything separated, stored, what determines the various rows?
so at the end i can#t report a huge gain of mhz!
now i'm able to clock 30mhz higher....not much but the temps/voltages dropped a lot!
http://server2.uploadit.org/files/ki...z-newclock.JPG
mfg
I was able to get another 100 mhz with same volts and same temps. I should be able to get higher, with better cooling.
http://img20.echo.cx/img20/6642/28edit8ni.th.jpg
Ive gone and done something i said i wouldnt do and tht is take the shim off
:blink:
bye bye warrenty, 2.65ghz just was not stable enuff for my liking, temps were causing the CPU to crash, removing the shim has enabled me to run 2700mhz with 1.75volts, not quite prime95 stable, but seems stable enuff for all the other stress test programs and games.
My temps have also gone down in respect to clockspeed and voltage. I can am now running full load (S&M) 38C, prime dosent go over 37C, and all this with my fans at 5V
:)
I just wish AMD sold some chips without a shim, so we could still have the warrenty, hehehee, 100mhz extra is nothing to laugh at!!
Just hope i dont end up killing this chip inadvertedly, needs to tide me over until Venice, can then shift it ebay or online forum, i know there are peeps out there who would like a shimless CPU, even without warrenty
JuanFlaiter
I left it off. My waterblock fits nicely on it. However, I sold that chip to a friend and when he brought his system to a shop to diagnose his mobo, the morons in there chipped the chip. Now it can only run stock :(
I am now running a 3200+ which does better than the 3000+ with the IHS off :) I would take it off for 2.8 GHz stable but I'll probably just go dual-core or Venice so I decided not to bother.
Ok..
Just read this whole thread and I think im taking my fx55 IHS off later..
I bought it from someone that has taken it off already and put it back on..
Right now im totally prime stable @ 275x10 with about 1.59vcore.. Temps are about 38c idle and 48-52c load running prime.. I can boot to windows alot higher (2900mhz or so.. but not totally stable.. ) 280x10 is stable with most benchies but not prime..
Im hoping for some more MHZ and mostly to lower the temps... I have a 3400 mobile on my 250gb with a slk948 also and im pretty confident in mounting the sink on it without issues since its basically the same size core and all..
Plus with the IHS off I can place my temp probe next to the core like im used to doing on my 3400m...
Well wish me luck and I'll post my results later as long as I dont chicken out..hehe..
edit..
Maybe I shouldnt have did a search about removing IHS's.... getting scared just thinking about it after reading some of the stories.. :eek:
Then again there was some success stories..
Whew.. Ok that was scary ..lol.. what a hell of a rush..hehe..
But its all working and im gona test it out tonight.. Ive been using MBM for temps and it seemed to already shave off about 5c and the AS5 isnt even broke in yet .. I also put a probe next to the core to test .. here are some pics.. :)
http://img250.echo.cx/img250/4563/pi...2medium3xy.jpg
http://img250.echo.cx/img250/4040/pi...4medium6vt.jpg
I would have shipped it with it off if oyu wanted it that way ;)
Looks to me it was on there DAMM GOOD from the way the cermaic is spread :)
Good luck with the clocking and just rem to "FILE" the socket where you see the
"SOCKET 939" down if your sink doenst make perfect die contact..
call me if you have any questions :) :woot: :woot:
nice, you planning on cleaning that up at all, kinda looks like poo but thats an aestitic issue not performance.
Hey dude ..Quote:
Originally Posted by sandybeach
Yes it was on there good....I checked the part of the socket and its all good.I read this whole thread today at work.. :)
Still testing it.. I had a raptor that died a slow death and was giving me much instablity and I didnt know that was it. But so far 275x10 is totally stable with whatever I do.. Gona be testing with my bh5 and a higher multi this week..
edit..
I didnt bother cleaning it , maybe next time its off.. I was just happy it came off fine with no issues..Quote:
Originally Posted by Disposibleteen
:banana:
sounds like fun, keep us updated with results!
Are these cores made from the same material as the ones on mobile xp's. If so, without the IHS can they be handled treated the same. In other words no special precautions need to be taken as far as mounting water blocks and air coolers? I noticed a few post above someone mentioned chipping, and I did not know if these were more brittle/fragile.
you can use huge force as with reg mobile chips just go down strait not at an angle.. Ive used coolers that weight upwards of 8lbs "4x 7700cu zalmans for reference", lol..
its when you dont go down strait you chip/crush the cores :)
Hi guys ! I read the whole thread and seems that i didn't payed attention on wich boards people were succeeded to remove the IHS and installed their HSF.
Can anyone confirm it on DFI NF4 Ultra,D,DR boards?
And maybe someone had a luck to use the Zalman CNPS-7000B HSF after removing IHS ?
Tnx everyone !
if you are really worried about cracking/chipping the core just get some little sticky foam circles like the ones on the mobiles and put them on there. You can find them at any hardware store in the same section they sell cabinet hardware (knobs and handles and such)
Yup thats spot on, ive done exactly the same thing to my CPU and GPU :). Spongy material on each corner is the way to go, as well as along the small resistors, you dont want to damage thoseQuote:
Originally Posted by Disposibleteen
;)
Heyup THunDA (from dfi-street??),Quote:
Originally Posted by THunDA
Good to see you had a successful dismount of the shim :). Takes great care and some *alls :), but once its off, dont drop your guard taking precautions, the cores do chip quite easily, so always be cautious
:)
mong
Yup thats me.. :)Quote:
Originally Posted by mongoled
Im not too worried about it because im used to mounting my other slk948 on the 3400 mobile chip thats on my 250gb.. Was just a little scarey taking it off.. I almost gave up when I started..lol..
After some testing last night the temps are definitly better..
Before.. idle 38c load 48-52c
After ... idle 33c load 43-44c
Also with the probe i placed next to the core it reads 32c idle and 40-41c load..
So it was well worth it for the temps.. :)
Ive been using the 30min test in OCCT .. And before the highest I could pass it was at 275x10 with about 1.59vcore.. Now after this its passing at 280x10.. This is also the highest it will pass when not using a high htt also..
Before prime would also error out in 3-5 min's at 280x10.. But after this it ran 4 hours last night before an error.. I know Sandybeach..hehe.. im not being one of them running prime all the time..lol.. just wanted to see if this changed anything.. It does pass all 3dmarks at 280x10 with no crashes at all.
This weekend im gona finally install some games and see how she does at 280x10..
:D Anyone tried a naked FX on single phase.......??? :confused:
:toast:
why on singel pahse why not some cascade action?
nice results :)
yeah you cant beat the TIM AMD uses between there chip and IHS :)
It gets all the tiny imperfections as its chalky/hard :)
Only put it back on in the first place for that other guy ;)
:( Pic does not work for me.........Quote:
Originally Posted by Disposibleteen
:toast:
my winnie died for removing the ISH.
I remove that of my winnie in 1min with a cutter
very easy, +50MHz average stock cooling
I had to mod
:D Think I'll give it a try - FX-57 is still far away from now......... :p:
:toast:
no IHS + dod or chiller = crackpotQuote:
Originally Posted by HARDCORECLOCKER
:confused: What do You mean with this? Sorry for askin' but english insn't my first language.Quote:
Originally Posted by cpulloverclock
Do You mean I'll damage the core or do You mean it will gimme a nice improvement?
THX !
:toast:
I think he's saying that you'll crack the core
:confused: Hm....., kind of strange here, just PMed to OPP and he told me he always removed the IHS from all of his chips.Quote:
Originally Posted by IvanAndreevich
And AFAIK he's using a cascade or promy all the time............
:toast:
of courseQuote:
Originally Posted by IvanAndreevich
Noob question, but what IS no a an IHS?
IHS= Intergrated Heat SinkQuote:
Originally Posted by YanBooth
actually, it's Integrated Heat Spreader last i checked ;)
Quote:
Originally Posted by ozzimark
Spreader.. sink.. same thing..hehe.. ;)
hehehe :toast:
Who is the first one to stripped off those IHS on Venice or San Diego? have you noticed any improvement? common....I know someone is pretty extreme and hardcore enough to do it on their new cpu!! Please show some pictures when you do as well.... :D :D :D I'll take off mine as soon as I get my hands on it.
yeah i want to see the die, so much purddier than heat spreader
i know it's not a real san diego, but the die size is the same (1mb)
http://www.xtremesystems.org/forums/...412#post822412
I wish they would start making them out of copper add least!
HS is made off copper :stick:Quote:
Originally Posted by thesaucier
Ummmm they dont look cooper to me!
In fact all the heat spreaders on all the A64s I have had are not cooper. :confused:
it is copper with nickle plated. just send it down and you'll see copper
Ummm I will always be learning!! LOL
mmmm purdy heheheQuote:
Originally Posted by ozzimark
if the ihs was soldered/welded to the core instead of just some TIM then it would actually work really well
As far as I know transistor counts for the Venice are identical to Winchester, so as far as size goes, the 512kb dies shouldn't be any different. And ozzimark linked to some pics of the 1mb Turion dies, which should be the same size as the San Diegos.
I intend to if I get one.
How do you take it off? Is it easy to damage the chip if you do? I mean in the process of doing so.
Check the sticky thread on it in Best of AMD
I cut the heatspreader off anything I get into my hands. :cool:
Don't like them.
Here are some pics of "naked" NewCastle/Winchester/Venice cores.
http://www.planet3dnow.de/vbulletin/...d.php?t=175268
(first posting in the thread)
The Venice looks like that
http://home.arcor.de/xxmartin/A64_38...nice_naked.jpg
Martin
hey martin i have your 3500+ newcastle under my prommi atm.
i can't find the hs - for myself i had no problems when using "normal" cooling without hs but a selfmade prommi kit and a cpu without hs is no good.
it's running after 3 days of working, checking evap position, a little bit less sealstring...checking position, be carefull not to crack the die, scratched an edge, still running, but man this was the hardest i ever did since i started building computer systems.
still problems with over zero temps on =/>1.7V vcore....its hard to get enough contact on that small die without destroying it but with enough isolation on an standard mach I evap...the pipe to the evap turns as it wants so you always have a pressure against the best position - really nothing for cpu's without hs...
That heat speader is hard to take off. I used a razorblade and worked at it for about 15minutes in and all still can't get the darn heat speader off. I ended up snapping the razor when trying to prey it off. Maybe more of you are more talented at the heat speader removal art form or I might be just too nervous compared to many of you. On the other hand I'm having to work at it with my chip in the socket still cause my chip doesn't want to come out. I have a good gap I cut in the one corner and cut the glue a bit along all sides but it wasn't enough yet.
Update, I've been trying to the the IHS off for about 40 minutes now still no luck. That thing is glued on pretty good.
Oh sorry I needed updated my sig. I have an AMD64 Winchester 3200+ CPU.
With my old 3500+ Newcastle the temp dropped 10 ~11 degrees loaded. On my old winchester 3200+ it dropped 8~9 degrees. With the newcastle I gain 10~20 Mhz..with the winchester 70~80Mhz....Quote:
Originally Posted by Jamo
Well I broke down and striped off the IHS. It just had to go.
It came off easy. Nice sharp razor blade and 2 mins. later it was naked. And to my horror there was no TIM contact with the cpu in the center. the center of the slug was bare. ( shakeing head )
Had to make up some shimes to remount the sink with good pressure, abit of as5 and we are back in the game. Add'ed another 45 mhz oc to it with the same 1.60 volts.
Its been folding for about 30 mins. and the temps have droped on the cpu by 6c. Was at 44c and now bouncing between 37c and 38c.
Ive been cycling the speed of the fan on the hyper 48. turn fan down to its lowist setting and saw 48c the first time. Then cranked it back up to high. Temp went down to 40c. Cycled it once more and now down to 38c. Room temp hasnt changed more than 2 tenths of a degree.
http://www.psyshack.com/a64naked.jpg
This move may very well pay off good.
i riped the heatspreader of my sandiego 3700 and before mod my temps were 56c in prime 95 and it failed in 5 minutes and after temps are 42 and it passes prime all night long. this is at 9x300 with 1.65 volts. i think my heatspreader had bad contact or something because i didnt expect that improvement!!! im using a dfi and watercooling btw
________
SATANISM, THE CHURCH OF SATAN ADVICE
New case badge!!
http://www.psyshack.com/li1.jpg
http://www.psyshack.com/li2.jpg
Quote:
Originally Posted by psyshack
Haha !!.. thats cool.. :toast:
I'm gonna try to remove HS this weekend...What have to do ?
Careful for what ? I have DFI NF4 and 7000B-AlCU.
As soon as i get some zerex for my watercooling im gonna install it and do the IHS at the same :D
Right now I can do 295x9 1:1 real stable at 1.6v but anything over that gets alittle unstable
First time I took my heat sync off my cpu, my cpu came out of the socket with my heat sync. Of course I put my cpu back in and put the heat sync back on to find my cpu is stuck in the darn socket... this is a new 939 motherboard, if I can't get my current working CPU out in the future I guess I'll need a new motherboard to ever run a dual core cpu. Also I've been trying to get my IHS off but it's pretty hard to do when my cpu is stuck to the board and there's not enough room around the socket to work a razor blade around the edges of the IHS to get it off.
I would be getting the cpu out of the socket before any razor action. Then look over you shiming options and how you going to get the hs back on.
Its not uncommon for a cpu to come out with the hs. I dont like it. but with a IHS and proper TIM application it happens
A cpu stuck in the socket,, now thats strange.
I removed my IHS.
Was i getting a good contact between CPU and IHS?
What is the best thermal paste to use? I used the one that came with my Zalman CNPS7000B. Is this a good one?
Well I got the CPU out of the socket and got the IHS off but my temps went up. I guess I just can't win... should of left mine on in my case. I put the ISH back on top of the CPU but temps are always higher with the retail or the Hyper6 I have.
Update, I shaved down a bit of the rubber left on the IHS and my temps are slightly better than they were in the first place. It looks like I need my IHS on to get the best temps but I added artic silver 3 between the core and the IHS.
my load temp is only about 1-2C lower but that is at standard voltage... well test higher voltage after. But with a load temp of 40C after running prime95 for about 10 minutes isn't bad using air cooling
The problem is my OC lowered :(
Can this be because of not enough pressure on the core?
Please look at the picture in my previous post and tell me what do you think about contact :confused:
My overclocked lowered as well but I believe this is because the heat sync isn't making full proper contact and my temps went up. I had to put the IHS back on in order to get proper cooling but I had to level up the bottom of the IHS but evening the rubber first in order to get good cooling once again.Quote:
Originally Posted by Ref
I can't believe how heat sensitive these .09um process cores are when it comes to overclocking.
I have 0.13 um core.
So you just put the IHS on the core with some thermal paste on the core? and then mounted the heatsink?
And you got better OC than before you removed the IHS ? :cool:
Hopefully they run cool :)Quote:
Originally Posted by cu2day
Yep you got it. I can now overclock on average per voltage/clock ratio about 50Mhz more. I was hoping I wouldn't need the IHS but the core is so small that the slightest weight shift from direct center makes a big difference in the temperature department. if I run about 5C hotter my overclock drops by about 70Mhz. I need actual R134 cooling, than I'll see some nice gains with how much the smallest temperture changes affect overclocking.Quote:
Originally Posted by Ref
How exactly did you put the IHS back?
is it worth it to buy some better thermal paste? e.g. Arctic Cooling MX-1 ?
I never heard of that thermal paste. I know of Artic silver 1 through 5. 5 being the best of them all. Getting the best thermal paste is always a good investment if you like your equipment to run the coolest possible.Quote:
Originally Posted by Ref
I only placed the IHS back on the center of the CPU no glue or anything with the case flat to the floor. My backet is a custom one that comes with my Coolermaster Hyper6 which allows very little side to side movement so their is little risk of moving the IHS out of place.