Printable View
Well I am not sure how everone is getting there 4400+ overclocked to 2.7ghz mine pretty much doesnt even like 2500+ I am watercooled,
I recently removed the heatspreader. then applied AS5 and replied the Heatspreader, that did help with the temps 2-3c. and I just received some OCZ PC-4800 Plat Limted Edt Ram. didnt help. rest of my specs are in sig
Just remove the heatspreader completly and try again.
Sounds about like my 4400, core 0 is much weaker than core 1 and doesn't like going much over 2.5ghz, at least not prime stable, water cooling here as well.
I can boot into Windows fine at 2.7ghz but it won't even prime for a minute before stopping, not that prime stable is even 100% stable but if it can't pass prime at the very least something isn't kosher, seems to hit a stability wall around 2.5ghz.
I've been very tempted to remove the heat spreader completely but I've also been thinking about selling it on ebay and picking up another 4400.
going to remove the IHS in a sec.. I will results in a abit
mine runs 2.7ghz in windows but if I primes 5secs crash
hmm so I'm not the only one ;) one of my cores doesn't like to even go above 2.4ghz
here to . core 0 run out at ~2500 mhz and core 1 at 2800 mhz 1.55v vcore
i cant test higher , it get to hot.
EDIT
i have change bios from 711 to 602 .
and is now stable at 2600 mhz 1.55v vcore
http://valid.x86-secret.com/show_oc?id=28885
thats weird. that kinda sux actually, getting a rele good core then the other one being complete junk
That doesn't make me wanna get a X2 soon :(
ok.. 300htt*9 (2700MHZ) Running Stable with IHS Removed.. Full Load Temps 47C
I think the problem is the IHS doesnt make good contact with both cores.. but at anyrate I AM STABLE 1hr prime on both cores !
ok here is my CPUZ Validate
http://valid.x86-secret.com/show_oc?id=28887
Damnit, you're gonna make me pull the IHS!!!! :slobber:
I've been thinking about the posibility that my IHS isn't making proper contact somewhere. I went from a XP120 air cooler, to a basic Swiftech h20 120 water setup, to now a much more powerfull water setup (Storm block, D5 pump, BIX2 radiator, 1/2 plumbing) with temps STILL staying in the same range, tired AS3, AS5, & ceramique!
The water and radiator temps are hardly even noticably warmer than anything else yet cpu temps under full load ALWAYS get right to 50C+ with all three different cooling setups I've used.
I think its time to pull the IHS after hearing your results! :thumbsup:
so this is without the IHS? ...stable?Quote:
Originally Posted by X2beach
yep.. Stable as a rock.. I havent done the HARd Core 24hr prime.. but all games work pime ran for 1hr. SuperPI running.. anything else you like to me run?
highoctane: dont be afraid to remove the spreader, its very easy, if you need some help send me a pm. it really helps with the temps.
my cpu is 40C idle and 58-60C load at 2600 mhz 1.55v vcore.
and it take 2 sek to go from 40 to 60C :)
my "water cooler( I dont now to english word for it)" on the cpu is 30C on the outside.
maybe the IHS have to come off.
fine: yeah remove the ihs spreader, its holding you back
Damnit from 2500 to 2700 :slobber:
You're not going to believe this but after removing my IHS my full load temps, dual priming, dropped from 50-53C to 34-36C, everything else the same other than removing the IHS.
WOW is all I can say, was a little nervous trying to get the razor between the die and ihs but once I got the feel for how deep to cut and how much pressure to use it went alright, a little nerve racking still.
Anyways, hopefully I'll see similar gains as you, I definitely had a contact issue with the IHS!!!! :eek:
EDIT: Ok set things to 10x270 and 1.6vcore to start with, been priming for 3 minutes now, thats a record for this core to keep priming at 2.7ghz, max load temp is 37C at 2.7ghz and 1.6vcore, priming each core with affinity set to run on each core independently, up to five minues now from me typing so slow! :D
ok ima warn yall in advance but im a tard when it comes to this but if i dont ask ill never get it but here it goes.
i order a 4400+ or the 4000+ and its on the way but can you tell me what
(IHS) is because im going to have to do this as well.
maybe a pic of one of yall actually removing it.
p.s. should i have just gotten the 4000+ over the 4400+ im asking because all i do is play cs 1.6 and source but anyways thx.
Removing the ihs was the first thing I've done with all of my 3 a64 cpu's. The temps always dropped around 8-10°C under load. It very easy to remove it, just use a very sharp razorblade. Cut about 2-2,5mm throught siliconglue that is used to fix the ihs on top of core. To remount it perfectly, for example with as5 on the core is very difficult. Just forget about that an use your a64 without that good cooling killer :D
4400+ is good if you want to multitask alot
also its really is amazing how much IHS Sucks!
yeah, I've "only" a 4200+ @ 2904 :D ihs removed :p:
the system is answering to demands much better in general. Really cool. I would have bought a 4400+, but it's very hard to get it in germany. Only the 4200+ and the 4600+ are available in quantities.
Integrated heat spreader over the cpu die, you might want to leave it on. Depending on your particular heatsink you have a high likely hood of cracking and ruining the core and the mounting height for the heatsink must be lower to accomidate the lack of the heat spreader.
I've been on the fence about removing mine for a while now and finally decided to take the risk, a $600 risk at that.
Anyways I'm happy I did in the end but I accepted the risk, been priming for over 50 minutes now at 2.7ghz 35-37C reported full load temps. :D
I'm probably going to let it prime over night to see if it keeps going but so far so good. If it makes it through the night I'll start dropping the voltages some as I normally don't run my ldt at 1.4 and it probably doesn't need to be that high and the vcore may likely be able to drop just the same.
I'm guessing there must have been very poor heat transfer from core0 to the heat spreader while core1 had adequate contact, that might explain the large discrepency with core0 tapping out at a much lower clock than core1.
cracking of the core is no problem for me so far. I use little bumpers, look like the ones amd used for the athlon xp cpus. Works perfect with my nexxos watercooler, the thermaltake big thyphoon and with a modified mounting with the zalman 7000 too. I've used 10 cpus, all without ihs and without problems during the last 6 months. Just be careful an it's no problem to use it naked :)
highoctane...
Just in case this helps, here's my info. With help from 'xxmartin' and 'Major' I removed the IHS and it was the best thing I could have done :clap: Thank's gentlemen!
I've still want to dirve down my memory some, so I have another 'overnight' session planned with MemTest86+.
When I executed my "final" 2xPrime95 test, I used 'Custom' and 800MB on each instance...due to the 2GB of memory.
Enjoy,
Lam
With my FX-55 I boot windows fine at like 2.8
lambusser
our system are very close, same case, powersupply, and watercooler. except i run sli 7800 :-) and only 1gig of ocz 4800
4 hours strong now, time to let'er burn overnight and hopefully in the morning she'll still be going.
If I wouldn't have checked out this thread tonight I might not of ever removed that damn IHS and kept on thinking I had a bum x2, tommorow I think I'll try pushing it on up to 2.8ghz instead of trying to lower the voltages! :D
Its still amazing to me that my load temps dropped a solid 15C!
Welcome to club of uneven core :) I am so tempted to take my heatspreader off but I am on phase change so cracking on core might be... bit easy.
Oh my god, you people are making me do this to my X2. Perhaps someone can do a simple guide on this with pictures? That would be much appreciated.
Thanks.
Take the thinnest and sharpest razorblade you can find. And cut around the IHS. Start with the corners then the sides. Cut often but gently. This way you can always try if you can lift it up already and be sure not to damage anything.
Yeah but thats hardly amazing since stock is 2.6... :pQuote:
Originally Posted by cvlive08
you think this would work?
http://home.megapass.co.kr/~rhgkswhd/razor.jpg
There is some more information here.
Hmm.. ok. Maybe that won't work then. I will have to go out and get a really thin blade.
http://home.megapass.co.kr/~rhgkswhd/newblade.jpg
EDIT: Just ran out and got these. Wish me luck! :slobber:
Yikes - lapped his core by the looks of it..
http://www.xtremesystems.org/forums/...934#post968934
How do you know how much each core can take?
Prime95 can do the job.
I some prog. you can choose which core the prog. have to run on
http://www.dfi-street.com/forum/showthread.php?t=16446
Okay, I just did it. Previously 2.6ghz wasn't possible on this board (MSi Neo2 Platinum). And it is running 2.6ghz and running prime so far without errors (15mins) without the IHS. Amazing. The thermal compound that was in IHS was hardened, maybe solidified? And as you see in the picture, the compound wasn't completely spread on the die. QC issue, perhaps?
Temp has dropped 8-10C approx.
EDIT: I just bumped it to 2.7Ghz using clockgen. Prime runs fine!!! Temp @ 48C. This is incredible!
EDIT/2: I've tried to set it to 2.7ghz in BIOS and won't boot windows. Tried 2.6 and will crash at bootup. However, if I boot 2.5ghz and overclock using Clockgen, it will work fine till 2.7ghz?!?! Could this be Bios problem?? (MSI Neo2 Platinum v1.9 NFM modded bios)
http://home.megapass.co.kr/~rhgkswhd/x2_004.jpg
http://home.megapass.co.kr/~rhgkswhd/x2_006.jpg
Hm that sounds really great! Anyone running a decapitated X2 with phase ? :slobber:
Edit: h4nc0 make sure you read the clocks with CPU-Z in windows
Yeah, I did that. Clockgen set 2.7Ghz is prime stable, but BIOS set 2.6ghz won't even boot Windows. :confused:
Almost 16hrs of dual prime goodness at 2.7ghz, hasn't missed a beat! :D
i have just take the IHS off my 4400+.
and the load temp are drop to 52C from 60C. so it give me 8C.
but i am not sure. the temp on my Dfi nf3 is jump up an down all the time.
for now it is from 18C to 48C .so i say 48 C is the temp.
maybe there is a bios update for that
Sounds like you have bad contact with core.Quote:
Originally Posted by fine
But 18c sounds to low, try other bios.
I would like to know also, since i am awaiting on a unit and would either getQuote:
Originally Posted by sai
this chip or just jump on an FX57.
I have a Dfi NF3 ultra D. there is 3 bios 602. 705 and 711.
i have testet 602 and 711 and 711 dont work werry well.
i win and bios my vcore is 1.58 to 1.60v i now the board undervolt.
idle temp is somewhere between 20C to 40C.
Edit.
I can only use 602 bios . 705 and 711 lock my fsb at 269 mhz.
This board is asap on it way to DFi for rma. i dont wont to fry a $740 cpu in a motherboard that dont work.
is the msi neo2 work with a x2 4400+ cpu ?
am I missing something or what? its been sid that maybe there wasnt good contact with one of the cores, but the naked pics of the X2 I have seen make it look like its just 1 big core, not 2 separate ones.....
explanation??
you have right about that . is one big core.
... which includes 2 cores, because it was planned that way from the beginning of the K8 core. Not like Intel, who just put two seperate cores in the Pentium-D with some crossbar switch.
Its not that hard to understand and nobody said there where two physicaly sepearated dies with the X2's, not sure where you got that.
A quiz for you, why is it important to spead thermal compound evenly, is it true that the X2 have two processors on the same die side by side, if so what happens if one side of the die has better contact than the other?
please someone take a few pic of the operation of the X2 before and after for us new ppl. :)
All I can say is WOW.
Great job guys. :up:
I have done it so it is to late.
here is a (how to remove ihs from a amd64)
It in danish but the pic is good.
how to .
here is the tool i use.
Do you guys have any recommendations on whether a Hyper6 would be safe to use with the IHS removed? Before you flip out, I know it is heavy, but I am using it vertically (in a desktop style case), not in a tower, so there should not be any leverage issues on it, just a decent amount of weight evenly dispersed from the top.
To MMIQHW, if you see this, where did you get your bumpers from for the CPU? I would appreciate a reply or a PM, thanks man!
only if it was in english.
p.s. do you need to wear gloves or anything while doing this?
I agree with the helping in temps (it makes quite a huge difference with watercooling) and OC by removing the IHS, sure did work great for me, running 10x285 @ 1.5v stable so far, need better memory to go higher with the fsb and low multi ;)
@rsGTS
One set from an old Alpha Cooler. The other ones used ontop of my venice and my winchester are from creative labs speakers. You could buy things like this for example from tesa, the producer of the legendary tesafilm.
Greetz,
Erik
The swiftech one that screws down. You don't want to use a big heavy clip down type of sink on the exposed core. Too much pressure on the corners when mounting and remounting.Quote:
Originally Posted by rsGTS
In posts 3-6 they mention one core going higher than the other. How is this possible? Both cores on my x2 4800 have the exact same clock.
Are you using some special version of clockgen made for the x2?
I am using bios 702-2.
They just set both cores to 2,7ghz for example but test them one on one
hmm, I wasn't aware I could test the cores seperately
Yeah we only talked about a 1000 times about it. :p:
If you end up having perfectly matched core... something is wrong... you should go buy a lotto man! In general though it seems about 150-250mhz difference between cores (which could certainly have been better matched with associated cost by AMD I bet)
..and mostly the difference is caused by a bad connecting IHS
I am just going by what CPU-Z says 2905.5 for both cores. What thread talks about testing cores seperately...I have been out of the loop.
You set the affinity to specify one core or the other.
No, the issue was one core would overclock higher without crashing. They are both set to the same o/c all the time, but testing could isolate each core at the same o/c via the afinity thing and see which one was stable.
I will have to research setting affinity. I haven't seen this.
You can assign affinity of any process from task manager and prime has an option already integrated that allows you to set affinity.
I see now. Thanks.
OK, so screw down ones are better? Like the Zalman 7000cu?Quote:
Originally Posted by xgman
I got a 7000cu mounted on my XP 2500.. screws are down as far as they go (not sure if i was supposed to do that lol) but no probs with that so far.. so i think you're right...
But what i'm wondering though is.. how about the 9500?.. I'm planning to get that with a 4400+ when both are available here, but seeing the results in this thread i really want to remove the IHS as well.
Will that be dangerous considering the 9500 is taller?..
In a tower case the screws will be horizontal (on the DFI board)
i'd think it would be safer with a heavy sink if the screws were vertical.Quote:
(x) socket (x)
I guess i better find some damn good bumpers/washers... or is it really not that fragile?Quote:
(x)
socket
(x)
Not a challenge but I haven't seen seen anything that clearly proves this..... :)Quote:
Originally Posted by sai
A solid stable 200mhz gain on core0 and a 15C overall load temp drop in my case is all the proof I need. :)
Glad you got the bump, but your success hardly supports a challenge that AMD’s manufacturing process is faulty or that a poorly mounted IHS is the norm rather the exception. If this was the case we would have heard much more ‘bout this long before the X2 was born……. :)
Well, it looks like one side of the die, or one core, or whatever is getting good contact with the ihs, and one isn't, therefore one clocks higher than the other regularly. When taking it off, it seems both will do higher rate because of better contact. I don't see how this could be seen before X2 was out, as there would need to be two cores to show this happening. As for the ihs making bad contact on single core cpus, people have done this ever since 754 iirc, and have been getting better results in some cases. It looks like the IHS just sucks in general, but can definately effect the x2 as cooling every part of the die is important, and the ihs doesn't make good contact with all parts on the x2, especially one of the cores.
As for proof...I think 4 or 5 people have posted in this thread about substantial gain from this, on the avg 200mhz or so. That's enough proof for me.
Oh yeah, and to those guys who risked destroying their brand new expensive X2's to find this out: Much respect and mad props. :clap:
I bet this would really help me out, but I just can't risk it
I removed the IHS on my 2.4C a couple years ago successfully. Not a bit scared to do it on my Venice....however, I'm worried about contact with the core. Will the stock heatsink for example, make perfect contact with no IHS? :confused:
I'm thinking of removing my HS on my 4400+ too... I'll post results soon.
still shaking in my boots about taking the IHS off my 4800...gonna find my max overclock and then decide if I think it is worth it. I sure would like to see the other core be a better performer
It is risk-free if you do it properly.Quote:
Originally Posted by Ender17
right ! Only a really sharp razorblade is needed. Just cut about 2.5 mm towards the core. Cutting not to deep is important.
After removing the ihs use some athlon xp style bumpers and everthing is fine. The core isn't more fragile as the athlon xp core was.
I suppose an XP-90C would be way too heavy for this, or? What a bout an alu XP-90 or is the Cu version ok too? How about the Storm blocks (G4, G5) can they be used without worries? And what about fitting,shouldn't there be some millimetres of extra space with conventional seating?
Like MMIQHW and rsGTS were talking about, what bumpers would be good to help take pressure off the IHS-less X2 core? MMIQHW were speaking about some Tesa, any actual product from them? I only find the usual masking tape. And does anyone where you can buy in Europe (or Sweden) preferably?
Please, don't recommend US shops before reading ppl's "Location" texts. :)
example
Tesa bumpers
I'm sure there are shops in sweden selling the tesa products too.
I've got an athlon xp shim...would that work?
If you mean using the Athlon XP bumpers, sure.
I have this http://img296.imageshack.us/img296/8016/shim1mq.jpg I am wondering if it will even help. Those bumpers look thicker.
The xp bumpers compress and are fine.
Anyone thinking about removing the ihs read this thread in full.
http://www.xtremesystems.org/forums/...ad.php?t=69978
My results tonight :Quote:
Originally Posted by TuKo
Idle 38°C, Full 59°C (with HS)
Idle 33°C, Full 43°C (with HS removed)
Primestable at 2.75GHz
Stepping: CCBWE 0519UPMW
any better OC?Quote:
Originally Posted by TuKo
4200+ watercooled
264*11=2904 MHz @ 1,52V
Load = 41-42°C
Idle = 33°C
With ihs I don't know. Removing it was my first step before mounting the CPU.
I could be wrong of course, but it looks like the avg with the ihs on is 2.5-2.7, and avg with it off is 2.7-2.9. On top of that, a few degrees off the idle temp, and a crapload (~10c avg) off the load temp. The second number us surely a key to the higher clocks. The lower load temp allows for the higher clocks. IOW: the IHS sucks at it's job, and is hampering clocks by not dispersing heat correctly during load. I think enough people have tested it to call it a fact.
Once more, awesome job guys. Thanks for info like before and after clocks, and temps before and after. They really help getting to the bottom of the riddle...and these are making X2's look very appetizing! Can't wait for someone with a 3800+ to rip off the IHS and give some results... :slobber:
My Core1 went stable at 2.75GHz, nothing else...Quote:
Originally Posted by Ref
Got my core 0 to 2.94GHz tonight. Nothing stable.
http://valid.x86-secret.com/cache/29999.png
http://valid.x86-secret.com/show_oc?id=29999
Took the HS out... on my single phase... finally got dual prime95 stable 3ghz rig @ 1.7v v-core (gasp).
Not sure I want to keep it there but I will find my better 24x7 setting between memory and CPU that requires much less voltage... 2.9ghz needed only 1.6v -_-; Now bench stable probably is another 50-250mhz vs current speed but not much into it (if I wanted to bench, would have gotten FX-57 to begin with or something alike ;P) Just out of curiousity, I ran single prime with this voltage and got to 3060mhz prime 95 stable... on weaker core.
this is wicked stuff mate. Keep it up!