Delidding 1366 CPUs [Soldered]

[snorlaxgangs]

That temp @4.41GHz is quite the same as mine. Really good. BTW do you pain the area around the socket? A couple weeks ago, i found a little bit leaking near the mosfet. Idk it's because i put too much or the magnet from my screwdriver draw some out, kinda accidentally put in near the socket when i installed the card. Anyway, i pain the whole board with nail polish to make sure.

Speaking of the waterblock for SR-2, did u order from nateman_doo from evga?

 

[Dinnercore]

That temp @4.41GHz is quite the same as mine. Really good. BTW do you pain the area around the socket? A couple weeks ago, i found a little bit leaking near the mosfet. Idk it's because i put too much or the magnet from my screwdriver draw some out, kinda accidentally put in near the socket when i installed the card. Anyway, i pain the whole board with nail polish to make sure.

Speaking of the waterblock for SR-2, did u order from nateman_doo from evga?

I did not paint anything with nail polish. If you look closely at this picture:


You can just barely see the Kapton-tape I put on the SMDs of the CPU to protect them. I had no spill and mounted the cooler multiple times. Thats another nice thing about LM, you can remount with the same application and don´t make a mess.
I was very careful to use just the right amount, very little LM is needed. I spread it around without any 'puddles' forming and used even less on the cooler surface. On the cooler I just used what I had left on the Q-tip to make a shiny silver surface but no visible drops.

This way I had nothing spilling or squeezing out. I had the board vertical with a big piece of Kapton underneath the socket and gave it a shake, nothing came out.

The waterblock is from Liquid Extasy http://www.liquidextasy.de/ . They still make those waterblocks new, but production may take a few months.

 

[Dinnercore]

Just a short update:



So far the system is still alive and no signs of instability or degradation from sanding down the die and using LM. I´ve run a few bench-sessions with it, turned it on every day to check on it.
I think now the gain from going direct die is becoming obvious, 60°C on the cores @1.56V.

It has been nearly a month and the CPU is still alive, I pushed it shortly to 1.62V on two cores for a 5,2GHz validation: https://valid.x86.fr/7i74mt

Tough little chips. So for the short term, there is no worry about lapping a die directly. I´ll put it into storage now and keep the liquid metal on there. I´ll test it again when I´m done with some other projects

 

[delshay]

Just a short update:



So far the system is still alive and no signs of instability or degradation from sanding down the die and using LM. I´ve run a few bench-sessions with it, turned it on every day to check on it.
I think now the gain from going direct die is becoming obvious, 60°C on the cores @1.56V.

It has been nearly a month and the CPU is still alive, I pushed it shortly to 1.62V on two cores for a 5,2GHz validation: https://valid.x86.fr/7i74mt

Tough little chips. So for the short term, there is no worry about lapping a die directly. I´ll put it into storage now and keep the liquid metal on there. I´ll test it again when I´m done with some other projects

5.2GHz does not get you into the top 15 from the year 2015 -2016. Were in 2019 now with better tools. The top overclock is 9.2GHz.

 

[Dinnercore]

5.2GHz does not get you into the top 15 from the year 2015 -2016. Were in 2019 now with better tools. The top overclock is 9.2GHz.

Actually I´m top 15, exactly place 15th: https://hwbot.org/benchmark/cpu_frequency/rankings?hardwareTypeId=processor_2105&cores=6#start=0#interval=20

Anything above 6GHz is a faked validation. There is no way to unlock the multi on those and you will not get the BCLK any higher. Not even LN2 can help you there and please remember I´m still on ambient water.

 

[Tomgang]

Just a short update:



So far the system is still alive and no signs of instability or degradation from sanding down the die and using LM. I´ve run a few bench-sessions with it, turned it on every day to check on it.
I think now the gain from going direct die is becoming obvious, 60°C on the cores @1.56V.

It has been nearly a month and the CPU is still alive, I pushed it shortly to 1.62V on two cores for a 5,2GHz validation: https://valid.x86.fr/7i74mt

Tough little chips. So for the short term, there is no worry about lapping a die directly. I´ll put it into storage now and keep the liquid metal on there. I´ll test it again when I´m done with some other projects

So close to my best score with my I7 980X. Personal best score is 1103 in CB R15. Dit it on plain air cooling, but at ideal temperatures out side. So if you are up for a challenge, beat my best score so far.

 

[delshay]

Actually I´m top 15, exactly place 15th: https://hwbot.org/benchmark/cpu_frequency/rankings?hardwareTypeId=processor_2105&cores=6#start=0#interval=20

Anything above 6GHz is a faked validation. There is no way to unlock the multi on those and you will not get the BCLK any higher. Not even LN2 can help you there and please remember I´m still on ambient water.

It looks like you could be right making a lot of users cheats. I don't know anything about this processor, but c'mon, one user has a multiplier of x66.

 

[Dinnercore]

So close to my best score with my I7 980X. Personal best score is 1103 in CB R15. Dit it on plain air cooling, but at ideal temperatures out side. So if you are up for a challenge, beat my best score so far.

Heh already did that: https://hwbot.org/submission/4290997_

Currently the fastest CB15 on ambient cooling for the X5650.

 

[Tomgang]

Heh already did that: https://hwbot.org/submission/4290997_

Currently the fastest CB15 on ambient cooling for the X5650.

Oh i see. Well done then. Now crank it up to eleven and hit the 1200 score.
But your X58 is a fun platform to mess around with. All throw i am at the start to move on to a AMD Ryzen system.

 

[Dinnercore]

As a follow-up to the one delidded CPU with a chipped substrate, I tested it today and it is alive.

But the physical damage goes even further:



Not only did the substrate split open on one edge, an SMD cap got knocked off too.

But like I said, it is alive and well despite its scars.



It is not the best overclocker I have, but it can still manage to hold 4,7 GHz and at default settings its running perfect.

 

[ShrimpBrime]

That's awesome!! rare to see that extent of damage and the chip still be in working order. I never ever get that lucky!!!

 

[basco]

no offense but why ya all run a command rate of 2T ?
on newer platforms i think its not as beneficial as on x58 and older

 

[ShrimpBrime]

no offense but why ya all run a command rate of 2T ?
on newer platforms i think its not as beneficial as on x58 and older

Can you be more specific to whom you are talking to and what system/s you're talking about?

I ran DDR3 @ 2200mhz (1100 effective) gave better results and stability at that frequency.

 

[basco]

i saw now that if benched for hwbot.org they go for 1T

in this thread nearly all talk is about socket 1366
post 10+ 21+25+31

i just wanted to know because some are comparing their scores and thats why i asked why not command rate at 1T which for me is not gaining any mhz or stability more on these older platforms with 2T. except for 48gb+ of ram maybe.

 

[Dinnercore]

i saw now that if benched for hwbot.org they go for 1T

in this thread nearly all talk is about socket 1366
post 10+ 21+25+31

i just wanted to know because some are comparing their scores and thats why i asked why not command rate at 1T which for me is not gaining any mhz or stability more on these older platforms with 2T. except for 48gb+ of ram maybe.

When I post scores here I usually dont do it to compete, more a thing of testing stuff out. And for this I don´t care to optimize RAM timings, I just go with the auto settings. On my hwbot score you can see that I switched to 1T for exactly the reason you said.

There are limitations tho, I had to use 2T on my SR-2 with 48GB to get some stability, so I agree with you there too. For this platform.

 

[basco]

i did not want to be mean just a question why and now we got it or i got it

forgot to say : i love this socket- still going strong these days

 

[ShrimpBrime]

i did not want to be mean just a question why and now we got it or i got it

forgot to say : i love this socket- still going strong these days

Right on, just looked you up at the Bot. You have some pretty impressive DDR3 memory clocks there!!

Keep up the good work mate!!

 

[snorlaxgangs]

Hi guys, it's been almost 10 months since i delid/relid my cpu and applied LM on both sides of IHS,the DIE, and the bare copper EK block. I use this pc for folding 24/7 since November, still haven't experience any problem so far, temperature is still the same around 45-50*C under max load. TBH i haven't take the block off yet so i'm not sure if LM is still "wet". Anyway, wow are you guys' cpu doing? Any degradation? Did LM eat thru the DIE no?

 

[Dinnercore]

Hi guys, it's been almost 10 months since i delid/relid my cpu and applied LM on both sides of IHS,the DIE, and the bare copper EK block. I use this pc for folding 24/7 since November, still haven't experience any problem so far, temperature is still the same around 45-50*C under max load. TBH i haven't take the block off yet so i'm not sure if LM is still "wet". Anyway, wow are you guys' cpu doing? Any degradation? Did LM eat thru the DIE no?

I`ve been running my SR-2 as daily system with delidded and sanded (removed diffusion barrier) X5690s in direct die cooling with LM. So far its still stable and no issues yet.

I only encountered one issue with the LM getting 'dry' as I was super careful and applied to little. Over time the copper block from my cooler absorbed the LM and there were parts of the die no longer covered. I noticed that as a few cores had suddenly 10°C higher temps then others. I just added some more LM and it has been fine now.

 

[snorlaxgangs]

I`ve been running my SR-2 as daily system with delidded and sanded (removed diffusion barrier) X5690s in direct die cooling with LM. So far its still stable and no issues yet.

I only encountered one issue with the LM getting 'dry' as I was super careful and applied to little. Over time the copper block from my cooler absorbed the LM and there were parts of the die no longer covered. I noticed that as a few cores had suddenly 10°C higher temps then others. I just added some more LM and it has been fine now.

My issue from beginning were the washers and standoffs weren't even so it caused poor contact. I also applied a bit more than enough LM so once awhile i found a little escape and hanging around on the board, but it's been a couple months without "leaking".

How far you can push your Uncore on SR-2? Some people said it's pretty hot when overclocking.

 

[Dinnercore]

How far you can push your Uncore on SR-2? Some people said it's pretty hot when overclocking.

I could not go much further then 3,8GHz for Uncore:



Anything close to 4GHz did not make it into Windows for me. Temps are fine, but you can see in this screenshot already that I had too little LM, as on the second X5690 one core hit 60°C and the others were low 50s.

 

[snorlaxgangs]

I could not go much further then 3,8GHz for Uncore:


Anything close to 4GHz did not make it into Windows for me. Temps are fine, but you can see in this screenshot already that I had too little LM, as on the second X5690 one core hit 60°C and the others were low 50s.

Same for me. If i want to boot w 4GHz or 4.2GHz UCLK i have to set QPi Voltage above 1.4V for booting, and ~1.42-1.45v for stress testing. Good score in AIDA64 tho.
About the 2nd cpu temp, mine also has 1 core that is 10C different than the lowest one. No matter how many times i tried different setup, LM or normal paste, with or without IHS it's just like that. Maybe i had to sand the die down like 0.1mm to make a the different?

Anyway, do u have problem with SB report temp sometimes a lot higher than NB? Mine shot up to 51*C while NB is at 34C, both are under wc. Idk it's bc i use 3 graphics cards or the bclk 200 is not a sweet spot like i think it was. I switch back to 133 and HWinfo still shows the SB temp sometimes increased up to 51C

I also found that G.Skill sniper 2400 C11 is a lot more compatible for 1366 than any others ddr3 for generation after 1366. I tried ribjaws, tridentx 4GB 2400, all needed to tweak timing. Kingstons hyper beast black 2400 can't even boot above 2133. The G.SKill ripjaws(blue) 2400 C11 sometimes failed booting when the temperature is too hot. I wonder if it's has anything to do with auto self-refresh( depending on temp).
The old tripe channel ram kit are too expensive for some reason, and so hard to find them. So i settle with 3x4GB G.Skill snipers 2400 c11 for now till i can get 3 more, it's around $10-12 in my country.

 

[Dinnercore]

Maybe i had to sand the die down like 0.1mm to make a the different?

I don't think that would make any difference. Most likely the sensors have a big tolerance. But mine went from 10°C difference to 20°C difference after 2 weeks which told me there was something going on.

Anyway, do u have problem with SB report temp sometimes a lot higher than NB?

I don't have many boards to test with, the SB on my Rampage III Extreme was fine. It was similar to the NB at around 35-40°C. On my SR-2 its all under water and I don't have clear software readouts, it only labels temp1, temp2, temp3 etc.. Still all sensors on my SR-2 show 25-30°C max..