These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Out of Pod Experience

 
  • Topic is locked indefinitely.
 

How I fixed my Ivy Bridge with a razor blade
Author
Previous Topic Next Topic
Agaetis Byrjun Endalaust
#1 - 2013-01-20 14:04:38 UTC
First: don't do this at home! Oh well, unless you're very sure about what you're doing

Second: this is not a guide nor a suggest, and not original or new. Actually I did it last summer after my rig upgrade. I've read many forums before doing it and checked other people experiences, issues and suggests, so DO THE SAME before risking your 300 $ worth cpu because you're nub and didn't train Cold Blood Control to lvl 3 and Hands Pressure Management to lvl 4.

Basically the problem is that Intel put a crappy tim inside their cpu, but allowed us to open them. At the contrary in the previous Sandy Bridge they used a very good tim but soldered the cpu so it was not possible to remove it and change it. Don't ask me what is intentional and what not in the Intel politics, the fact is that if you want to overclock your brand new Ivy's (and you DO want to!!) then you'll face some evident problem with overheating, running prime for 10 mins on a bland 4.1/4.2 overclock will bring your cpu up to 80-90° C, thus reducing your oc'ability margins. And remember that at 105° C your piece of silicon will burn unless the thermal throttle safeties will be activated (check your BIOS and never ever deactivate them).

In any case, that's a totally absurd behaviour, and in fact if you change the tim you'll cap your temp at about 55-60°C which is an outstanding 30°C less average. Simply changing the tim!!

Now, as I told you, the process involves SERIOUS RISKS and you should inform youself about which are them before trying, but if you do it as you should, in 3 minutes you'll have fixed your cpu once for all.

When an image is more than a thousand words...... I'll give you 3 images

http://i.imgur.com/fKTwe3w.jpg

http://i.imgur.com/p0xcnqr.jpg

http://i.imgur.com/WeGMjUU.jpg

The process is not over and you'll have to clean it very well, choose an appropriate new tim for the inside (I suggest the Liquid Pro), close it and put some more tim on the top of the IHS, not the Liquid Pro here (inform yourself about why not!!) but a normal artic colling mx-4 but yes...... that's it.

__________________________ just because you're paranoid it doesn't mean they're not after you
baltec1
Bat Country
Pandemic Horde
#2 - 2013-01-20 15:52:03 UTC
Came expecting to see before and after pics of you clearing a bridge encased in ivy with a razor blade. Leaving dissapointed.
Akita T
Caldari Navy Volunteer Task Force
#3 - 2013-01-20 18:03:15 UTC
Personally I'm kind of against overclocking and even a fan of underclocking, but I'm always glad to see DIY work done by others to push the edge of what you can do with off-the-shelf gear.
Waiting for feedback of actual improvements after the fix-up-job.

P.S. I hope you did benchmark it with clean heatsink and fan before starting that, you would not want to actually see an improvement that was due (even if only in part) by doing the rest of the stuff alone without opening the CPU at all.

Akita T evewiki profile | | | Build your own EVE-ready PC
Sturmwolke
#4 - 2013-01-20 20:12:51 UTC
Did you use anything to replace the resin glue ... or just left it at that?
Mars Theran
Foreign Interloper
#5 - 2013-01-20 20:28:56 UTC
I never bothered looking into it, but I always wondered why the new Ivy's were generally viewed as running too hot. This sort of explains it. I've got a Sandy Bridge myself, so not something I have to worry about for the moment. Interesting though.
zubzubzubzubzubzubzubzub
Agaetis Byrjun Endalaust
#6 - 2013-01-20 21:12:35 UTC
Quote:
Did you use anything to replace the resin glue ... or just left it at that?

I personally left it without. Replacing the resin glue is not as easy as it seems, mostly for the kind of stuff you have to use, where to find it, how to apply it, how to spread the right quantity and many other "lil things". The intel socket has those steel brackets that just work perfectly for keeping it still and very well pressed. The only problem is that if you used the Liquid Pro you don't want to remove it too many times, and possibly never. This means that if for some reason you're one of those who likes to change the tim between the cpu and the cooler, you'll have to make it leaving the cpu in the socket/motherboard.

Quote:
P.S. I hope you did benchmark it with clean heatsink and fan before starting that, you would not want to actually see an improvement that was due (even if only in part) by doing the rest of the stuff alone without opening the CPU at all.

sure I used my pc for few days before "operating it" because I was waiting for the Liquid Pro and also because I wanted to see if it was really that terrible. It was. After the mod nothing else changed, it was full summer and in idle I was at external temp +3-4° and in full load at about 55°C, versus the 80-85° I had previously.

Quote:
Came expecting to see before and after pics of you clearing a bridge encased in ivy with a razor blade. Leaving dissapointed.

ROFL

I use my minmatar knife for that Lol

__________________________ just because you're paranoid it doesn't mean they're not after you
Shalia Ripper
#7 - 2013-01-20 21:47:54 UTC
I approve of voiding the warranty in this fashion.

Sig blah blah blah blah
Marie Hartinez
Aries Munitions and Defense
#8 - 2013-01-20 22:08:12 UTC
To me, it looks like the thermal grease was a bit dry. Also looks like they added just a bit too much. Some of the engineers I work with have said that too much grease can be just as bad as not enough.

I don't normally approve of such measures, but if it's working for you, who am I to judge.

I've had a couple AMD chips run way hotter under load then I felt was normal. Tried everything I could think of. Even had a huge third party heat sink and fan on them. Now, I wonder if AMD had similar manufacturing defects.

Surrender is still your slightly less painful option.
Rain6639
GoonWaffe
Goonswarm Federation
#9 - 2013-01-20 23:12:46 UTC  |  Edited by: Rain6639
Ignoring your claim of success, for objectivity's sake, the consensus seems to be:

benefit is negligible.
Puget Systems wrote:
What we believe happened is that the first CPU had a particularly bad TIM paste application, while the third was relatively very good. No matter how automated the manufacturing process, there are always variances from plant to plant and even from day to day. The first CPU we tested likely had a relatively bad TIM paste application, while the CPU we used for testing had relatively good paste application. This is supported by two facts. First, our pre-TIM testing showed that the first CPU we modded ran about 7 °C hotter than the CPU that we performed our extended testing on. Second: after replacing the TIM paste, both CPUs had temperatures within 1 °C of each other. So in other words, replacing the TIM paste will only cause a significant difference if the TIM paste was badly applied in the first place.


that you capped your temps at 55-60C is a huge claim, brosef.

Agaetis Byrjun Endalaust wrote:

In any case, that's a totally absurd behaviour, and in fact if you change the tim you'll cap your temp at about 55-60°C which is an outstanding 30°C less average. Simply changing the tim!!



only worth doing if it runs hotter than other CPU's in a similar configuration. then if the CPU is modified in this way, it's only to get it back to spec. I would rather RMA a CPU and cite higher temps as the reason.
Akita T
Caldari Navy Volunteer Task Force
#10 - 2013-01-21 00:03:37 UTC  |  Edited by: Akita T
Rain6639 wrote:
Ignoring your claim of success, for objectivity's sake, the consensus seems to be:
http://www.pugetsystems.com/labs/articles/Ivy-Bridge-CPU-TIM-Paste-Replacement-160
Quote:
So in other words, replacing the TIM paste will only cause a significant difference if the TIM paste was badly applied in the first place.

that you capped your temps at 55-60C is a huge claim, brosef.

Actually, that link kind of adds some credence to his claims in a slightly roundabout way.
The OP might have been indeed exaggerating a bit with the temperatures, but he might have a bit of a point there.
Keywords, "might", and "a bit".
See last quote/paragraph for overall assessment agreement.

He was talking about his CPU reaching 80-90 with overclocks to 4.1/4.2, while the article has a 4.5 overclock running at under 70 both before and after TIM replacement, and the stock at 58-61 after/before TIM replacement. Granted, that's with liquid cooling, but let's assume the OP also has liquid cooling (and maybe even better liquid cooling).
Sure, it's not the exact same chip model, but I can only assume the difference between the 3570k and 3770k is minimal as far as thermal stuff is concerned.

Looks to me like it is within the realm of "somewhat possible" that the OP had a "badly applied TIM" chip to start with (or slightly damaged in transit), and what he actually managed to do is bring it down to what was supposed to be the "usual specs".
And it can also be that the thermal sensors are not completely reliable, so he might have thought he got a certain temp in some cases but in actuality it was slightly different.

TL;DR - maybe a bit far-fetched, and slightly inflated, but considering all circumstances, somewhat possible, ballpark-wise

Rain6639 wrote:
I would rather RMA a CPU and cite higher temps as the reason.


Agreed. Most likely the much better alternative for most people.
Certainly not worth the risk of ending up with a damaged CPU just to fix a defective one.

Akita T evewiki profile | | | Build your own EVE-ready PC
Agaetis Byrjun Endalaust
#11 - 2013-01-21 00:16:09 UTC
Quote:
that you capped your temps at 55-60C is a huge claim, brosef.


Quote:
TL;DR - maybe a bit far-fetched, and slightly inflated, but considering all circumstances, somewhat possible, ballpark-wise.


http://i.imgur.com/rW5ssKD.jpg

ah btw, not only I don't have a liquid cooler, but I also removed 1 of the 2 fans from my heatsink. You know, didn't need that fan anyway

LoL

__________________________ just because you're paranoid it doesn't mean they're not after you
Akita T
Caldari Navy Volunteer Task Force
#12 - 2013-01-21 00:23:35 UTC  |  Edited by: Akita T
Well, to be fair, that's peaks at 60-65 not 55-60, most likely at a lower ambient temperature, and we don't know for sure just how well calibrated those temperature sensors are (or if they operate like they should after the DIY job).

Akita T evewiki profile | | | Build your own EVE-ready PC
Agaetis Byrjun Endalaust
#13 - 2013-01-21 00:34:07 UTC  |  Edited by: Agaetis Byrjun Endalaust
you don't have to read the maximum values. Those are just the records of the peaks reached during the test, but the 4 cores never ran at the max at the same time, they run as the big black bold numbers say

The sensors are the same who were showing 80-85° C before the mod, but that could be just a lie afterall. Right?

Wrong! Ask people haivng a 3570k to run a prime test for 20 minutes at 4.3 GHz (not an extreme overclock indeed but good for a daily use) and post a screenshot of their temp, with an air cooler with only 1 fan in extraction, then we can talk about evidences and make some comparison

__________________________ just because you're paranoid it doesn't mean they're not after you
Rain6639
GoonWaffe
Goonswarm Federation
#14 - 2013-01-21 00:58:25 UTC
capping temps is a big claim, and the premise was misleading. I understand you're proud of your CPU modding capabilities and what you've done, but you should have made it clear this is only necessary for blemished/flawed CPUs. I think it's fair to say this topic was meant to make me think I should do the same thing to my CPU.

I'm not singling you out. I become very scrupulous over claims that I have some "need" that I was not aware of, and I prefer they are wrong.
Agaetis Byrjun Endalaust
#15 - 2013-01-21 11:51:00 UTC
Momma told me "don't feed the troll" but you're really a clueless troll Lol

search the dictionary for "blamished" "flawed" and most of all for "RMA". You could end up being wiser than now, not that it's so hard but meh..... my 2 cents

Also reading the first sentence of this thread should answer this

Quote:
I think it's fair to say this topic was meant to make me think I should do the same thing to my CPU.


oh damn, I prolly fed the troll again Lol

__________________________ just because you're paranoid it doesn't mean they're not after you
Akita T
Caldari Navy Volunteer Task Force
#16 - 2013-01-21 14:19:40 UTC  |  Edited by: Akita T
Agaetis Byrjun Endalaust wrote:
The sensors are the same who were showing 80-85° C before the mod


The guys in the linked article say they screwed up a memory channel with their CPU mod on one of the tries. It is unlikely but conceivable that the temperature sensors might have gotten a bit decalibrated or mildly distorted, leading to less than accurate temperature readings after the mod (or, it might even be possible they were over-reporting before but under-reporting after the mod).
Barring an infrared camera measurement both before and after, you can not be completely certain of what those sensors are actually reporting, taking it more ballparky than exact.
Also, I can remember the older chats about IvyBridge overclocking, and the "doing it wrong" initial versions which resulted in overheating (something with a needed counter-intuitive voltage reduction to get the best overall results). Did you also keep the exact same overclocking settings ?

The point remains, the INITIAL temperatures you were getting were NOT ok, since you most likely had the misfortune of getting a chip with the TIM applied badly. If the TIM (as cheap as it was) would have been applied correctly to begin with, you would have noticed far less of a temperature drop after the change you made.
Could also be you're one of the lucky few where the exact shape of the IHS in combination with fresh thermal paste (and heck knows what other factors) got you well above-average results compared to the other such upgrades, basically, a lot of good luck after the initial bad luck.
So it could be that you went from an unusually bad to an unusually good situation, an extreme in both directions. Not certain, but possible.

Again, the point was that, given the non-negligible risk of damage to the CPU (and the voiding of the warranty), most people in your situation would have been better off RMAing the chip instead of trying to fix it themselves, even if the potential gains might end up being pretty huge.

Akita T evewiki profile | | | Build your own EVE-ready PC
Eternal Error
Doomheim
#17 - 2013-01-21 20:43:47 UTC
Ignoring the rest of the back and forth ITT, I think you guys are really off base if you think Intel would accept an RMA for a "high" temperature that's still sub-90C under full load WITH overclocking.
Rain6639
GoonWaffe
Goonswarm Federation
#18 - 2013-01-22 05:39:54 UTC
Eternal Error wrote:
Ignoring the rest of the back and forth ITT, I think you guys are really off base if you think Intel would accept an RMA for a "high" temperature that's still sub-90C under full load WITH overclocking.


totally feel you. though they might not like it, chances are they will authorize the exchange through the distributor before testing the item themselves.