[Rediculous] De-Liding of CPU's. Anyone ever try it on here?

All about them.

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

[Rediculous] De-Liding of CPU's. Anyone ever try it on here?

Post by ~El~Jefe~ » Mon Jun 27, 2016 10:10 pm

I know that SPCR tests cpu's. I know they never took a IHS/Lid off of one and re-applied the TIM and re-tested it.

Seems like a dangerous idea, but the temperatures of the chips are 3-6 C less or more in some instances. INtel supposedly uses really garbage thermal paste and also badly applied.

Any thoughts for the modern chips?

[Mod: you edited the title back to misspelling ridiculous - that is your choice, I guess.]
Last edited by ~El~Jefe~ on Tue Jun 28, 2016 10:08 am, edited 1 time in total.


CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by CA_Steve » Tue Jun 28, 2016 6:35 am

Luca: Heh - many of the figures are taken directly from a 2006 Intel paper without attribution.

"The Material Optimization and Reliability Characterization of an Indium-Solder Thermal Interface Material for CPU Packaging", Journal of Materials, June 2006. (PDF here)

El Jefe: Short version: Intel moved away from solder based interface for smaller CPU die to the current TIM due to thermal cycle microcracking reducing the CPU lifetime. Yeah, they could have used a better TIM. Is it worth de-lidding and replacing it with a better TIM? For stock speed or OC with very little to no overvolt - I don't think so. If you plan to massively overvolt and have money to spend on a new CPU if you goof up - sure. I think there are some tools/jigs out there to help reduce the chance of killing the CPU during disassembly.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Tue Jun 28, 2016 7:16 am

CA_Steve wrote:Luca: Heh - many of the figures are taken directly from a 2006 Intel paper without attribution.

"The Material Optimization and Reliability Characterization of an Indium-Solder Thermal Interface Material for CPU Packaging", Journal of Materials, June 2006. (PDF here)

Thanks a lot for the heads up!

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by ~El~Jefe~ » Tue Jun 28, 2016 9:53 am

I am considering delidding as I intend my cpu to number crunch for like 12-16 hours a day for several years.

I would imagine the lower temperatures would show up at stock voltages as the chip will be moderately stressed. Hm? Making this machine a sci-fi number cruncher. I only build one main machine every like five years or so.

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by CA_Steve » Tue Jun 28, 2016 10:30 am

<shrugs> Might be worth building the new system and seeing what your temps are before taking this step.

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Tue Jun 28, 2016 10:32 am

~El~Jefe~ wrote:I would imagine the lower temperatures would show up at stock voltages as the chip will be moderately stressed.

Delidding + liquid metal TIM is expected to bring in -10/-15°C at stock, depending of your chip: do you need that for crunching? I can't help, but in case pair the mod with a proper shim like this one.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by ~El~Jefe~ » Tue Jun 28, 2016 11:21 am

quest_for_silence wrote:
~El~Jefe~ wrote:I would imagine the lower temperatures would show up at stock voltages as the chip will be moderately stressed.

Delidding + liquid metal TIM is expected to bring in -10/-15°C at stock, depending of your chip: do you need that for crunching? I can't help, but in case pair the mod with a proper shim like this one.
That shim is really neat :D

And yes, it might not be necessary to de-lid once I test things. I wonder that longevity would increase if I de-lid though, lower temps = longer life? Maybe not!

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Ridiculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Tue Jun 28, 2016 9:22 pm

~El~Jefe~ wrote:I wonder that longevity would increase if I de-lid though, lower temps = longer life? Maybe not!

Maybe: you'd remove a point of failure and reduce the life-long thermal stress but an exposed die is always more fragile than an IHS (and require some maintenance).

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by ~El~Jefe~ » Tue Jun 28, 2016 9:45 pm

I was thinking about delidding and then sticking it back on but with good TIM and no silicone goo

shrug?

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Wed Jun 29, 2016 1:22 am

~El~Jefe~ wrote:I was thinking about delidding and then sticking it back on but with good TIM and no silicone goo

shrug?

I can't help: I did so with a 3570, I reapply the Coollaboratory liquid once a year and it's still alive and kicking.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by ~El~Jefe~ » Wed Jun 29, 2016 6:40 pm

you use the word "can't" in a special way my Italian friend :)

Do you remove the old IHS and clean it off and reapply the TIM once a year?

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Thu Jun 30, 2016 12:45 am

~El~Jefe~ wrote:you use the word "can't" in a special way my Italian friend :)

I meant, albeit in about two years I got no problems, I can't grant about long term reliability, particularly with Skylake given its thinner PCB.
Probably with those LGA1151s I'd use the shim approach.

~El~Jefe~ wrote:Do you remove the old IHS and clean it off and reapply the TIM once a year?

Yes.

~El~Jefe~
Friend of SPCR
Posts: 2887
Joined: Mon Feb 28, 2005 4:21 pm
Location: New York City zzzz
Contact:

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by ~El~Jefe~ » Thu Jun 30, 2016 8:57 pm

wow. You are vigilant

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: [Rediculous] De-Liding of CPU's. Anyone ever try it on h

Post by quest_for_silence » Fri Jul 01, 2016 1:31 am

~El~Jefe~ wrote:wow. You are vigilant

I was mostly a bit afraid of the combined factors of the heatsink's cantilever effect and the liquid TIM (and I like tinkering with parts, of course): OTOH after two manteinance rounds I must say that probably there's no need to do like I did.

Post Reply