Delidding Gone Bad?

[GMart84]

Okay so I am a sucker apparently and I hope someone can help me understand what happened.

 

This is a bit of a story so bear with me.

 

Four weeks ago I really wanted to delid my 8700k to get temps down and possibly get a performance increase out of it.  I bit the bullet and sent it to Silicon Lottery for their services, while it was there I got the opportunity to get an 8086k chip at lower than the 8700k price so I thought why not, I will likely get better performance out of it already and I am not shy to say I like limited edition items.

 

So I get it home and pop it in my rig and off we go.  5.0 Ghz no problem at 1.29 vcore stable in P95 non-AVX for 1 hour, okay lets bump it up. 5.1 Ghz @ 1.33 vcore stable in P95 non-AVX for 1 hour, okay up we go. 5.2 Ghz @ 1.36 vcore stable in P95 non-AVX.  Sweet, this is awesome, I love it.  Especially because my 8700k could only do 5.0 Ghz.

 

Fast forward a few days, my 8700K comes back which i was going to use in a different build with the message from Silicon Lottery, 5.1 Ghz Stable at a 1.412 vcore delidded.  I was amped up, sweet lets get this going, I plug it in and bingo as indicated it was stable and idles at 25° C and at load is at 48° C.  This experience made me think, okay if this happens with this lowly 8700k what could I get from the 8086k, which yes I know is a binned 8700k.  So clean it off and pack it up and ship it off to Silicon Lottery for delidding.

 

To say I was nervous with anticipation would be an understatement, I wanted my chip back hopefully hearing it had jumped 0.1Ghz as well.  Today is that day, I get it back, rip the package open and "Your CPU should be stable when using the settings below and blah, blah, blah" 1.410 vcore, -2 AVX offset, and.....wait for it 5.1Ghz!  Wait what?!?!?  Sure enough I plug it in and get it going, and nope, now I have a CPU that requires more voltage to run at a lower speed.  Okay, I admit, 1 hour may not be long enough to detect instability, but I am sorry this is a little unbelievable to me.  I will complete the exact testing as described to me by Silicon Lottery, but there is no way to go back now and from the 30 minutes of installed time, I cannot even get into windows at 1.36 vcore and 5.2 Ghz anymore.

 

Please tell me I did my testing wrong so that I can stop the conspiracy theories going through my head right now.  What tests would you run to test stability and how long would you run them for?  Thank you for your time.

[geo3]

Did you get the same chip back?

[AngryBeaver]

I would confirm you got the same chip back. I would explain to them you were getting much better results before you sent it in and want to know why AFTER a delid it is performing worse across the board.

Chances are they just sent you back a chip they had already delidded.

[nick name]

Ooooooh I would raise quite the stink if they sent you a different CPU back. But if their testing differed from yours then you're gonna have to be very meticulous in how you go about making certain they sent you a different chip.  I am assuming their testing was more thorough and rigorous than your own and may have shown, to be as stable as they require, they needed higher voltages. 

 

Your experience, before and after, at 5.2GHz and 1.36V does seem to indicate it may be a different CPU though. Or that there is some degradation. 

[aezakmi]

you wasted your money for nothing

 

if you knew the 8086k could get to 5.2 WHY did you sent it to those guys?

 

and are you sure it's the exact same chip? maybe they scammed you when they saw the chip was actually pretty good so they decided to keep it for themselves and sent you another random CPU, what? you think they wouldn't do it? that CPU at that speed is worth at least $800 and I'm being gentle with the price, this might be a bold statement but that CPU, yours, could easily hit 5.4 or even 5.5 on non-AVX loads, if those guys decided to keep it you're done, there's no way to get it back as you can be sure they'll try to erase all of the serial numbers both inside and in the IHS and might even alter the ID if they have the proper tools

[nick name]

5 minutes ago, aezakmi said:

you wasted your money for nothing

 

if you knew the 8086k could get to 5.2 WHY did you sent it to those guys?

 

and are you sure it's the exact same chip? maybe they scammed you when they saw the chip was actually pretty good so they decided to keep it for themselves and sent you another random CPU, what? you think they wouldn't do it? that CPU at that speed is worth at least $800 and I'm being gentle with the price, this might be a bold statement but that CPU, yours, could easily hit 5.4 or even 5.5 on non-AVX loads, if those guys decided to keep it you're done, there's no way to get it back as you can be sure they'll try to erase all of the serial numbers both inside and in the IHS and might even alter the ID if they have the proper tools

You think scolding the OP is helpful?  And then making him feel even worse on top of that?  What is wrong with you?

[Razor Blade]

On their website they do have this...

 

Quote

We guarantee a successful delidding process, and will offer a replacement of the same processor model if any physical damage occurs during delidding. Please note that we do not service previously modified or delidded processors.

 

Might it be possible the CPU sent in was damaged? I would contact them and ask about it.

[GMart84]

1 hour ago, geo3 said:

Did you get the same chip back?

It seems like it is the same chip back as the batch numbers and everything match.

[GMart84]

1 hour ago, AngryBeaver said:

I would confirm you got the same chip back. I would explain to them you were getting much better results before you sent it in and want to know why AFTER a delid it is performing worse across the board.

Chances are they just sent you back a chip they had already delidded.

I am working on it, the replies from SL indicate that they are sure it is the same chip.  I am beginning to think this was a mistake.

[GMart84]

49 minutes ago, aezakmi said:

you wasted your money for nothing

 

if you knew the 8086k could get to 5.2 WHY did you sent it to those guys?

 

and are you sure it's the exact same chip? maybe they scammed you when they saw the chip was actually pretty good so they decided to keep it for themselves and sent you another random CPU, what? you think they wouldn't do it? that CPU at that speed is worth at least $800 and I'm being gentle with the price, this might be a bold statement but that CPU, yours, could easily hit 5.4 or even 5.5 on non-AVX loads, if those guys decided to keep it you're done, there's no way to get it back as you can be sure they'll try to erase all of the serial numbers both inside and in the IHS and might even alter the ID if they have the proper tools

So this is the exact question I have been asking myself, I do not think they purposely did it if they in fact sent me a different chip.  I did know it could do it as I had been using it 24/7 for a couple of weeks before sending it to them.  I decided to do it based on the results I got with my 8700K and because I like the huge temp drop it enables.

[GMart84]

The person I am dealing with at SL has been very helpful thus far and is trying to help me identify what happened.  They (because I do not know if it is a he or she) are actively helping me run the CPU through some testing to see if I can get back to the results I had before.  They also indicated that the P95 for 1 hour non-AVX does not guarantee stability and that I need to let it run overnight.  In addition I am thinking that I may do an exchange with Micro Center on the CPU as I bought the policy on it and that way I can try again.

[AngryBeaver]

14 minutes ago, GMart84 said:

The person I am dealing with at SL has been very helpful thus far and is trying to help me identify what happened.  They (because I do not know if it is a he or she) are actively helping me run the CPU through some testing to see if I can get back to the results I had before.  They also indicated that the P95 for 1 hour non-AVX does not guarantee stability and that I need to let it run overnight.  In addition I am thinking that I may do an exchange with Micro Center on the CPU as I bought the policy on it and that way I can try again.

The problem is you said it wouldn't even POST at 1.36 at 5.2. When you claim you were using it for a few weeks at that frequency and voltage before. That is a pretty noticeable different right there. I would expect performance to change for the better with a delid... i have never seen result decrease. Actually from my experience the 20-25c average I have seen in temp drops on a delidded chip has also let me reduce the voltage needed for previous overclocks.

 

Something is fishy here and I really wish you had took a picture of the chips s/n before you sent it in.

[GMart84]

10 minutes ago, AngryBeaver said:

The problem is you said it wouldn't even POST at 1.36 at 5.2. When you claim you were using it for a few weeks at that frequency and voltage before. That is a pretty noticeable different right there. I would expect performance to change for the better with a delid... i have never seen result decrease. Actually from my experience the 20-25c average I have seen in temp drops on a delidded chip has also let me reduce the voltage needed for previous overclocks.

 

Something is fishy here and I really wish you had took a picture of the chips s/n before you sent it in.

Me too, I am kicking myself for this.  I did a CMOS reset and got the machine to post at the 5.2 now @ 1.40 vcore, but it bluescreens going into windows.  The latest from SL is that my chip may have been stable using the workloads I was placing on it, and they have a rigorous process for testing and binning their chips and therefore may have tested it with different workloads.

 

When asked who they used they gave me two tools, P95 and linpack which I will try.  The also told me that it has been long hard work to develop their binning process and that giving me too much information would potentially lead to revealing an industry secret.

 

Because I obviously can't go back now I am convinced that if I had let the stress tests run longer that I would have realized some instability, in everyday gaming.  I am beginning to wonder if i damaged the chip removing it, or reinstalling it and that is part of what is leading to this.

 

I have had a couple of these times before where it is something very dumb that I forgot to do that is causing the problem.

[aezakmi]

1 hour ago, nick name said:

You think scolding the OP is helpful?  And then making him feel even worse on top of that?  What is wrong with you?

OP and everyone should be aware of this kind of stuff, everyone always say "nah I won't get scammed" until they get scammed and then don't know what to do and end up with no money and a faulty component.

 

sending something of value to some random dudes you don't know is always risky, this is not OP's fault because the site seemed legit but if they indeed sent him a different chip on purpose then he's been scammed and the chances of getting the original chip backs are really small because no one except for whoever took it knows where it is or what happened to it, be realistic

you know how easy it'd be for those guys to just use a different IHS on it? very, same with the ID, expert programmers could edit it easily to mask it, sounds elaborate, but possible, more if the CPU was that good (over 5.2).

 

if OP can prove it then he should start actions against the company to at least get a compensation for both the cpu and the trouble they caused him, at least, that's what I'd do

[GMart84]

1 minute ago, aezakmi said:

OP and everyone should be aware of this kind of stuff, everyone always say "nah I won't get scammed" until they get scammed and then don't know what to do and end up with no money and a faulty component.

 

sending something of value to some random dudes you don't know is always risky, this is not OP's fault because the site seemed legit but if they indeed sent him a different chip on purpose then he's been scammed and the chances of getting the original chip backs are really small because no one except for whoever took it knows where it is or what happened to it, be realistic

you know how easy it'd be for those guys to just use a different IHS on it? very, same with the ID, expert programmers could edit it easily to mask it, sounds elaborate, but possible, more if the CPU was that good (over 5.2).

 

if OP can prove it then he should start actions against the company to at least get a compensation for both the cpu and the trouble they caused him, at least, that's what I'd do

This is exactly what I do not want to think.  They have great reviews and most everyone spoken too about them shows them as trustworthy and legit.  I am sure the chip was that good, just hot 75-85C under load and wanted to cool it off.

 

I will do more testing.

[AngryBeaver]

24 minutes ago, GMart84 said:

Me too, I am kicking myself for this.  I did a CMOS reset and got the machine to post at the 5.2 now @ 1.40 vcore, but it bluescreens going into windows.  The latest from SL is that my chip may have been stable using the workloads I was placing on it, and they have a rigorous process for testing and binning their chips and therefore may have tested it with different workloads.

 

When asked who they used they gave me two tools, P95 and linpack which I will try.  The also told me that it has been long hard work to develop their binning process and that giving me too much information would potentially lead to revealing an industry secret.

 

Because I obviously can't go back now I am convinced that if I had let the stress tests run longer that I would have realized some instability, in everyday gaming.  I am beginning to wonder if i damaged the chip removing it, or reinstalling it and that is part of what is leading to this.

 

I have had a couple of these times before where it is something very dumb that I forgot to do that is causing the problem.

This isn't about the testing though.

 

You could boot in to windows for several weeks at 5.2ghz at 1.36v. So when you get it back unless they managed to degrade the hell out of the silicon (unlikely) then no matter what you should still be able to boot windows at 5.2 and 1.36v.

Running a benchmark for X amount of time longer will give you an idea of the stability, but if you were bootable without issues before at that frequency and voltage... then their whole argument about their testing is invalid. If you could do it before, but cannot do it after... then something has changed there.

 

Once again all they did was remove the IHS, apply LM, and reapply the IHS. The only thing that should have changed in that regard should have been your temps. So unless they damaged the chip, gave you a different chip, or seriously degraded the silicon... something has happened.

Honestly, did you even ask them to provide you with a stable overclock and voltage? So I would point out to them that regardless on whether your testing proved 100% stability, you COULD boot in to windows at 5.2ghz at 1.36v and did not notice any stability issues for several weeks. Now you have received the chip back and can't even boot windows at 5.2 at that voltage and that there is no way that a delid should give you WORSE overclocking results. So they can either choose to make things right or you can blast them on the different avenues available for what happened. Them eating the cost of giving you a confirmed golden chip in that scenario is often times much cheaper than the customers they stand to lose. Most people considering this are on the fence anyways.

[aezakmi]

18 hours ago, AngryBeaver said:

unless they managed to degrade the hell out of the silicon (unlikely)

you need years of running the CPU overvolted and under stress 24/7

[AngryBeaver]

1 hour ago, aezakmi said:

you need years of running the CPU overvolted and under stress 24/7

In general you are correct, but there have been cases where voltage and temps were too high for a short period of time and it also degraded the silicon.

[GMart84]

So I have to say I am impressed with how Silicon Lottery has responded to this whole thing.  They provided me assurances and the ability to test that the processor is indeed the one I sent through this link https://www.intel.com/content/www/us/en/support/articles/000021613/processors.html.

 

They also noted this thread which makes me know they care about their work.  I have now used their process to check it is indeed the same chip that I sent.  I have also managed to get it back to performing where I was at through lots of trial and error.  I also want to make note that it does fail at around 2 hours in P95 non-AVX I was clearly not running it long enough.  Due to that instability I have taken it down from 5.2 to 5.1 and it is solid at that point.

 

Basically this whole experience has proven to me that despite everything I think I know there is more to be learned.  Silicon Lottery helped me better understand what it means to test an overclock and provided me with an additional benchmark to run that I was not doing before.  It has also taught me that if something is stable under certain loads it may not be stable in all loads, so while I was getting great performance for basic work and for an hour of non-AVX beyond that and in other benchmarks it was not.  A complete battery of tests is necessary to achieve a truly stable OC.

 

I also want to again give a shout out to whomever has been emailing me from Silicon Lottery as they have been extremely helpful and spent time explaining everything to me which is good to see from a company.

 

I also want to thank everyone here who replied, it is good to know that a community will jump in when someone has a question.

 

Conclusion:

 

My previous overclock of 5.2 Ghz @ 1.36 vcore on non-AVX workloads was stable for 1 hour or so.  After the delidding and binning service at Silicon Lottery came back with their results, I was forced to realize that I had not truly established a stable overclock, due to it being only on the limited loads I was providing it.  Since then I have gotten the chip to do what I had it doing before, at a 20-25° C lower temperature thanks to the delidding.

 

Thanks again for all the help.