To De-lid or Not to......

Has anyone gotten their hands on some indium solder that Intel used to use, and tried to solder a water block directly to the die after de-lidding? This seems like it would be the most effective method of heat removal (i.e. water in a water cooled loop), as someone pointed out less steps to get the heat from the die to the source carrying it off would tend to work more efficiently (or so I would think). All you would need to do is to take block apart (so as to not destroy O-rings or plastics on block) heat and apply pressure. I'm sure with some calculations you could figure out how much solder to use to get the achieved coverage at the desired thickness so as to not push solder over the edge and make contact with the rest of the cpu, then it would just be how much pressure to apply to get an even spread. Sound simple, surely someone here has done this and wants to share the results?

Edit:
On post 57 above, what are those cracks, are they just were the old solder is still bonded to the die?

There are no cracks or solder. Just a little discoloration, I assume from the heat used to de-lid.

Do you have any idea how expensive Indium is?

Example: 3 feet of .030" Indium solder is $100.

Indium solder ribbon 1" x .002" x 12" = $150

http://buy.solder.com/

Mr.Scott said:

There are no cracks or solder. Just a little discoloration, I assume from the heat used to de-lid.

Do you have any idea how expensive Indium is?

Example: 3 feet of .030" Indium solder is $100.

Indium solder ribbon 1" x .002" x 12" = $150

http://buy.solder.com/

Click to expand...

Cool, just from the picture it looked like they were cracks on the die, or it was parts of the Solder left from

I did the lazy ting and asked instead of looking up price. It definatly doesn't seem cost effective for a single use, but not really out of the realm of possibilities that someone here hasn't tried. not something I will try this weekend though I can tell you that.


Edit: after some looking at that website, this really seems like the best method for thermal solutions, Although I was looking at some of the documentation and it is really pressure dependent on its thermal resistance since its not an actual solder. But if your looking for those last few 1/10th's this seems like it would do the trick.

Soldering a water block to a cpu die wont work because of the retention bracket and pin gate. You couldnt swing the lever up or down to remove or install the cpu. Thats why youll never see this done with an AMD processor.

So soldering isnt the way to go with a water block. Just gotta use the best thermal paste you can get your hands on.

t1nm4n said:

Edit: after some looking at that website, this really seems like the best method for thermal solutions, Although I was looking at some of the documentation and it is really pressure dependent on its thermal resistance since its not an actual solder. But if your looking for those last few 1/10th's this seems like it would do the trick.

Click to expand...

That's why I put the link up.
I know a few people that have tried sheet Indium. There was a reason other than cost that this didn't take off, but I can't remember what it is.
I think it was un-removable or destructive in some way. Can't remember.

dyckah said:

i used pcpartpicker on USA to get those prices...
in canada, 9590 is $280, and cheap 8 core fx is around $160

Click to expand...

I guess you don't have Microcenters in Canada.

chrisjames61 said:

I guess you don't have Microcenters in Canada.

Click to expand...

lol, nope, no microcenter.
and $1 USD is equal to about $1.25-1.3 Canadian.
heck, other than newegg, im pretty sure that most of the bigger canadian retailers are either owned by, or get their stuff from the one store, which i think is also the canadian distributor for a number of products.
not entirely sure on that last bit, just something i've heard rumor of.

ShrimpBrime said:

Crap its not the right one....

Sry lemme try again....

Click to expand...

Thats certainly one way to do it xD

bob4933 said:

Thats certainly one way to do it xD

Click to expand...

It's good reference to see competition under the hoods also. I'm not really new to this. Intel = easy de-lid. Go from 5 m/wk to 8.3 m/wk thermal paste and see improvements. And decent ones too. So you'll see countless threads. Each with good results. A must do. If it where mine, I'd use my Silver plate rather than the copper one and make a backing plate if need be.

Why use copper ihs nickle plated when you can upgrade? I lapped a Morgan silver dollar both sides. works great. food for thought.

EarthDog said:

zOMG its an Intel in a AMD thread!!!!

(please find that humerous... )

Click to expand...

Was meant to be

- - - Updated - - -

JR88 said:

That's awful..I hate Intel's all Overpriced, show boating/unfair and just Bull**** Marketing...say otherwise...I will stand by AMD...You want to fight about it?

Click to expand...

This GUY!! ^^

lol. no fight needed. Just some misunderstanding here and there. It's all good. He proved solder is a better conductor of heat. which is fine. But not in the application that was intended for AMD de-lidding. that's all it was. I got it.

But yea, 8 cores runs fine. 4.7ghz daily. doesn't actually "need" de-lid. Just fun to do and good for science.

Intel/Nvidia Fanyboy's make want to Throw up.

Click to expand...

Nothing wrong with NVidia bro. Gotta get yourself a 750ti or something and use it for physx. Back in the day, it used to be Ageia Physx. NV bought em and only improved this hardware physx. I game NV. I like AMD cards, they are great, but I also enjoy NVidia's Physx to the max. I still have agiea physx cards laying around. I sure am glad NV can run physx cause those cards had like 500mhz clock and like 50 transistors. made physx pretty choppy even back then. But it was new and awesome. I have CellFactor installed. It's super intense on physx. It really make a GTX 760 work it's butt off.

JR88 said:

No I'm serious it Pisses me off with Intel/Nvidia it's Bull****.

Click to expand...

Well fyi....

Most games are made on and optimized on Intel and NVidia cards while in the process of the make. AMD/AMD just so happens to be able to run the instruction sets. And if you look at games that are made and developed on AMD systems, Intel/NVidia doesn't shine so well.... even though there are few of them.

- - - Updated - - -

Tell you what.... wanna know why I like AMD?

Was the first chip I ever had.
Fairly priced
Zipping
8 cores no hyper threading needed
extinguished cold bugs
now hits wr clocks
still has unlocked reference clock
A challenge to de-lid

probably more, that's just at the top....

Well I'm glad and happy to see people keep a faith with AMD.

BUT, you gotta look at the future. US hardcore FX guys have nothing really to look forward to in the near future except APUs or expensive many core server processors.

So really what's today is already history, we just keep on trucking with our FX chips and game our butts off and enjoy. I figure mines good for a few more years at least.

Here's hopen'

JR88 said:

Know as well as I do AMD got screwed over bad by Intel/Nvidia back in the 2000s....horse **** paying off other manufacturers/catering/Anti-competitive garbage...

Click to expand...

I have no new Intel. Last ones were socket 775. I am assembling a fairly stout FX-8350 rig with new parts and spent monies. So I do AMD. But all this about AMD got a raw deal...well they supposedly received good monies for it.

What is so easily overlooked or misunderstood is that AMD was already being called out for the FX Bulldozer processor before they even were in REtail. Many with heavy computer engineering savvy knew that Bulldozer was going to be bulldozed in performance scoring before the cpus were in REtail. That scenario had nothing to do with somebody eles's doings but fell squarely on AMD's own staff from the top down.

And currently AMD has left us without a thing new in the performance market for far too long and really there is nothing concrete to all the rumor mongering going on. The same sort of BS went on before BullDozer finally made it to market. Hype and hoopla and then a feeling of a slap in the face from AMD for many of us 'the faithful' if you will.

I don't need more oomph than I can obtain with an FX-8350 at 4.8Ghz for days on end, but it sure requires a lot more substantial hardware than an Intel rig at 4.3GHz. I am willing to shoulder the necessities needed to run that fast with the HEAT from such daily overclock. Foolish of me maybe but I enjoy being able to go in the bios and change some settings other than Vcore and multiplier. Just me I suppose.

But I am not at all fooled by any poor me AMD leanings. Whatever has happened in the long run to AMD is nobodies fault but their own. We have entire cities and communities and associated companies that are GONE because they could not or did not compete. Plenty of whining with it all, but for years too long there has been no real stand up and work to do better.

If you go and read the history of AMD you will see how it began. Who began it and the level playing field that both Intel and AMD shared at the beginning. It is just that one did the overall job to a greater success to date.

Now I will go and continue my FX-8350 build for video editting and contemplate if I want to try one of the less than 16 month old FX 8 cores for use in a daily rig since they clock up to 5.0GHz but with seemingly less Vcore than the older than 16 month FX 8 core processors.

RGone...

JR88 said:

Know as well as I do AMD got screwed over bad by Intel/Nvidia back in the 2000s....horse **** paying off other manufacturers/catering/Anti-competitive garbage...

Click to expand...

In my eyes, Intel isn't exactly competitive. Just in a world of their own. See they have all quad cores. K series has hyper threading. It takes e-peen from four cores to run this. So lets say a non k model is within price range for 8 core FX. obviously, you'd buy an FX for 8 cores. Keep e-peed and have all the features I listed for the reasons I like AMD included. Gaming aside, AMD would be a fine choice.

Then you take into account the new kids pre-teens and young teens getting into PC gaming. Well most parents arent going to buy their kid an entire gaming system, however they CAN get AMD APU and no need for discrete graphics and be gaming. THIS is where AMD has Intel by the nuts so to speak. People don't want to pay high dollar for high end CPU computing then pay another high dollar for a video card cause Intel's onboard while decent, can't touch AMD APUs.

Then you have hybrid Crossfire on these APUs. So you can buy a lower end GPU, but get some decent graphics power with the crossfire and game at better resolutions and graphics details even over XBOX for probably around the same price.

I haven't gotten into the New APUs for de-lidding though. I should hunt one down (higher end) and try it. I know the Llano A8 where not soldered.

Ah, here should make for good reading.

So far I know of 5 people successful de-lids including chew also. I think he got a working cpu also.

IronBalls. Had great results, I shared the FX-5000 in this thread.
http://www.xtremesystems.org/forums...-newly-bought-FX-8320-PICS&highlight=FX+delid
http://www.xtremesystems.org/forums...delidded-my-old-1090T-PICS&highlight=FX+delid

This young German fellow with an interesting approach to de-lidding. Use and Iron for clothing. Looked like it worked rather well!

This kid, but he wasn't exactly praised for it seemingly. Never the less, he did it!
http://www.overclock.net/t/1540447/i-delidded-my-fx-8350-and

There's probably a couple more guys in the world successful. Just all I've looked for now.

The 4th link is antoniskots fx-8350 solder delid on oc.net, he got worse temps after delid, unfortunately though he only used paste time, did not try liquid metal.. he posted:

I use a custom watercooling system. Ek supremacy for the cpu. The temp makes spikes to hihger temps. Bofore delidd the cpu was never passing 35°C. Now it spikes to 45°C.

Click to expand...

In that same thread, last post, spacebug delidded an 8370E, said he got no better temps with paste, but then used liquid metal and got better temps. But he did not say how much better on temps with liquid metal.

ironballs on xtreme delidded an older lower power density soldered amd cpu, was making same point as I did, no point in delidding solder unless using liquid metal (and he was using lower power density)...he got 2-3C better temps with gelid, but 8-9C better temps with liquid metal. First quote:

It works like a charm. But I was expecting better temps. Doing this for 2-3 degrees is just not worth the trouble.

Click to expand...

his second quote from links:

With Gelid Extreme temperature maxed out @ 46 C after 30 mins OCCT.
With phobya liquid temperature maxed out @ 40 C after 30 mins OCCT.
In both cases room temperature was 26 C.
That is an 6 C improvement compared to Gelid Extreme. If I compare with the temps I had before delidding and with coollaboratory ultra I would estimate the decrease to be 8-9 degrees.
So it is only worth if you are willing to use liquid paste. I would not advice anyone to do this for only 2-3 C. But 8-9 degrees is a nice decrease in temps.

Click to expand...

Basically same as my experience, and same as others I have seen.....delidding paste cpus is definitely worth it where your replacing 5 w/mk paste with 40 w/mk liquid metal for tim1 seeing 20C temp drops with high watts. But delidding solder 87 w/mk and replacing tim1 with 5 to 8 w/mk paste (benefit of removing a layer is overshadowed by using crappy paste 5-8 w/mk for tim1 low surface area) doesnt have great results....if you go to trouble to delid solder better to use liquid metal 40 w/mk.

Again curious to see how 9590 does because of its power density... looking forward to the results, if you can, post some before and after load temp pics.

Oh they re_used the IHS plate. Thats a no no.

Never use IHS plate over again. It defeats the purpose of removing it.....

Other than that rge, yed tim is no good while re_using IHS plate. Full agreement, you wont see my results using the ihs plate after removal.

All of them ran bare die, antoniskots got 10C worse temps, he posted pics of the bare die mount. I have never seen anyone remove 87 w/mk solder attach ihs, and then put ihs back on with 5 w/mk paste, that would be purposely trying for 20+C worse temps.

The ones that reuse the ihs are starting with paste 5 w/mk die attach, and trying to replicate the solder mount...by delidding, removing 5 w/mk thick applied paste, put thin 40 w/mk liquid metal and replacing ihs.....and getting 20C temp drop ...or run bare die and drop temps another 5C.

btw...isnt that you on xtreme that posted you delidded the FX5000 and had no temp improvement, post 54 and 55 from one of your links above.
http://www.xtremesystems.org/forums...FX-8320-PICS&p=5236939&viewfull=1#post5236939

rge said:

btw...isnt that you on xtreme that posted you delidded the FX5000 and had no temp improvement, post 54 and 55 from one of your links above.
http://www.xtremesystems.org/forums...FX-8320-PICS&p=5236939&viewfull=1#post5236939

Click to expand...

Yea it was an quad core on my very stout liquid cooling loop. It was merely a pratice chip? Low end 2.2ghz quad.... yea it ran really cool from the get go. C2 revieion core Phenom based chip. Not an 8 core FX chip. Heck I think that was a 65w processor. Didn't need it, again was a practice de-lid.

All chips vary, but I've never had negative results unless I reused the IHS plate. swear to god if he was real.

Look rge, I've done some ... what?... 14 processors with success. You read other threads, guy did one... this happened. Some other guy did this other chip and this or that happened.

I give you 14 successful de-lidded processors on my plate and give an average temp drop of up to 10c.

So you believe one guys bad results over my many in the teens good results? Hows the comparison working for you? Not to be a mean guy, Ive just been there and done quite a few. How does one or two bad or equal results put you on this path of bashing a de-lid from a very experienced person? Since 09' brother.... years before any one else has tried this. I've perfected the art of an AMD de-lid to the best of my own practices.

It's like taking your car to a mechanic that's worked on one car before yours. Then you take it to that mechanic that's been around, generations of cars, was there before OBD1 and done it already. The nuts and bolts are all the same. Just one old guy knows how to take them apart properly and effeciently.....