Clevo Overclocker's Lounge

i got them, pretty happy thus far good performance, nice and squishy and sticky as well, perfect for a good component fit

 

Nice, thanks Jaybee83. Have you taken them off yet to re paste and how did they hold up if you did?

 

I would not recommend that. If there is a difference under the IHS you could crack the die or PCB if you swap them. That would be very tragic.

Just measure them or set them down on the table/desk with the IHS down and confirm the 7700K is not "shorter" than the 6700K. The PCB on each CPU should line up perfectly when placed side-by-side.

If you can take a picture showing if they are or are not the same that would be awesome. I wish I would have thought to do that.

 

I disagree, but will explain. What is in question is the depth of metal of the IHS, meaning overall height is meaningless. When we delid, if not resealed, often, there is still a gap between the edges of the ihs and pcb. The ihs was instituted because desktop air heatsinks got heavier and people were over tightening, thereby injuring the die or pcb. The ihs spread that weight to a more even distribution to protect the die. But, as said, if you delid, you transfer the pressure to focus on the die so be aware of what you are doing).

Now, if the ihs makes no contact with the pcb, it is only measuring the inside to the top, not including the lip that used to connect to the pcb, that matters. To determine this, use a caliper to measure a washer. Then, put the washer on the bottom of the ihs where the die connects. Measure top of ihs to the washer. Now subtract the washer height from the measurement and you have found the thickness that matters. Compare the value of both IHSs...

Edit: if you do have ihs to pcb contact after delidding, you must make sure of contact between the ihs and die. Otherwise, you will still have heat issues. At that point, you must sand down the lip of the ihs until good contact is made between the ihs and die.

Edit 2: think of the ihs as a shim and it makes more sense.

Sent from my SM-G900P using Tapatalk

 

they are exact same size.

 

How mine is setup:
CPU VRM pads re-done, lid removed and placed back on without glue, liquid metal on both sides of lid.

 

EXCELLENT! Thank you Brother John. I figured they would be identical, but it is always nice to have confirmation.

Thanks, Brother Meaker. Exactly what I have done, including no glue. The absence of physical contact between the IHS and heat sink cold plate prevented me from using CLU without a shim to fill the gap. I have die>IHS>shim>heat sink with a layer of CLU between each. Trying to plug the gap with IC Diamond failed miserably.

So, from that I gather I probably need to ask for a new CPU heat sink and hope the replacement is better than the one I have. Does that sound about right to you? The shim is working, but that's not the ideal solution for long-term.

 

But the screws used by factory don't allow for the heatsink actually touching the CPU LID, there is a safety gab by design. Let me check part numbers if anything has been updated since release...

 

That would be great to know. Thank you. I will wait to find out before I ask for a replacement heat sink.

 

So... is this to say that I need a shim? Just trying to make sure of everything I need to acquire before I get it and start working and things are missing (things are hard to come by here)

 

I think you misunderstood what I was saying. If one IHS is thicker on the die side, not the top, because the die underneath is shorter on one model of CPU compared to the other model and you lock it down with the retention mechanism that could crush the die/PCB and/or damage the CPU socket if the IHS is too thick and does not have enough room under it. You would need to measure carefully before trying to play "musical IHSes" and start swapping those around like spare parts.

In my case sanding off the edges of my IHS would only exacerbate the no-contact issue by making the gap even bigger. The IHS is basically not even touching the CPU heat sink. A narrow strip of thin paper inserted between the IHS and heat sink during a dry assembly was able to be slid out without tearing, LOL.

Hopefully not. Using a shim should only be done if a proper fit is not possible by replacing the heat sink with one that fits correctly. Obviously, some of them do..

If you have to resort to using a shim, eBay is your friend. Look for the 30mm*30mm size and that fits the IHS perfectly. They come on a slow boat from China, so order them before you are in a hurry.

 

OK, there are two revisions of DM3 CPU (GRID attachment type) heatsink 101 & 102.
I don't have an exact changelog but it looks like a CNC production process update...

@Meaker@Sager @Mr. Fox you may wanna check if you have the same model.

 

Ok, I am not sure, but I think I've narrowed down the issue. I don't think either are wrong, but we haven't come to a common conception.

If the die is shorter on one, then the edges of the ihs that form the lip on the underside attached previously with the pcb would contact the PCB on the other chip (which is what I was suggesting sanding, the lip on the other side to allow die and ihs contact).

My measurement was to remove the underside lip from the equation to find how thick the ihs is for the latter between the die and the heatsink. If the thickness without the underside lip is thicker on one over the other, then that one would be best to use with the heatsink as it acts as a thicker shim.

In all delidding, by removing contact between the ihs lip and pcb, we increase downward pressure on the die, which increases risks. This must be assessed when using the thicker ihs, but is Ann assumption taken when undertaking the process...

Sent from my SM-G900P using Tapatalk

 

If true, then removing the motherboard and placing shims (non-conductive) between the baseplate and the motherboard will shorten the mounting arms thereby closing the gap.

Sent from my SM-G900P using Tapatalk

 

How can we determine this? Is there a label? And, we are speaking of the CPU heat sink with a fit issue, as my grid seems to fit fine. Is the 101 and 102 a kit that includes both parts?

 

It's the CPU heatsink which has the GRID attachment and is a separate part than GPU/GRID itself.

It may have a sticker or embossed part number idk...if you both have the same than you know that you got a bad part, otherwise it may just be the production change...

 

Having a "safety gap" seems like a way to get away with building heat sinks that don't work correctly, LOL. If there is a gap, the end result is going to be bad, like what we already see.

That was something I mentioned in another post. That might be the best approach if trying to sort out the part number thing and determining if I have the wrong one, or the correct one that is merely defective, turns out to be a futile exercise.

 

It is a solution I devised on the ZM, but can be used on most machines. It just amazes me that rather than figure out proper tolerances, the solution developed is to leave a gap as a "feature."

Sent from my SM-G900P using Tapatalk

 

Spoiler

 

It would be really nice not to have to do that, but sometimes a creative solution is the best one. That might have worked on the P570WM as well. What did you use on the ZM to shim it? Fiber washers or something similar? Got any pix?

The pressure film on the left shows the fit of the CPU heat sink on the machine I sold to @Rage Set (which was sufficient for CLU to work excellent) and the film on the right is my current machine. Part of the red is from me handling it. When I first removed it there was no real evidence of physical contact.

 

I actually opted for the simpler removal of the spring screws, replaced by straight screws, sanded the ihs to remove the concave center, and pressure tested using different amount of screw turns per screw (using 1/8th turns with the screws marked, reversing until you hear the click, then screwing down from there). This worked whereas using the spring screws left a gap on a corner when fully screwed. If the switch to straight screws did not close the gap, I intended using a compressible pad or even using stacked ring enforcers (office supplies) to be able to use different amounts to assure even heights and compression across the plane of the ihs/heatsink connection (in the event that a mounting arm was slightly higher than the others, etc.)...

Edit: another option is using a low cost thermal pad...

Sent from my SM-G900P using Tapatalk

 

If the stand-off posts (mounting arms) that the screws thread into are too long, even replacing the screws with straight screws having no shoulder on them won't be sufficient. If the heat plate makes contact with the top of the arms/stand-off posts before the heat plate comes into direct contact with the IHS swapping out the screws would not serve any purpose and the only solution would be to shim the support plate or remove it and precision machine down all four stand-off posts by equal amounts to make them shorter.

 

Another option could be to enlarge the screw holes enough to allow the posts to pass through the heat sink plate (if they need to) and use spring tension to hold things down. The pressure would be less, but the risk of damage from crushing would be all but eliminated. You could stack flat washers between the screw head and top of the coil spring to increase pressure.

 

I have a hypothesis on this. Either the washer that was on that corner was just enough to prevent contact, or there is space, when fully screwed down, between that and the plate. Remember, the corners flare upwards from the heat plate. If it is slightly too little, it could prevent contact. By bending it further up (or away from the mounting arm), you increase the pressure on the plate in that corner. The other solution, to make that arm shorter with extra ring enforcers, shortens the arm, which then also pulls that corner down more.

I was lucky that the switch, whether because of that washer or removing a built in gap for the spring screw, which means with spring screws, the heatplate (edit autocorrect) never contacts the arms, was enough. If it wasn't, rather than adding more metal through which heat must travel, I would have used materials I have around to adjust the base plate, which is held to the MB by the three screws for the hinge to hold the CPU in the socket.

Edit: if there is a built in gap where the spring screws stop before the corners of the heat plate do not contact the mounting arms, extra cautiion must be taken going to straight screws because you can cause enough downward pressure to destroy the socket or crush the die.

Sent from my SM-G900P using Tapatalk

 

Funny how great minds think alike, LOL. That was exactly how I planned to address it at first, but after taking the screw/spring assemblies apart I made the sad discovery that these screws are a different design than what I am used to seeing. The c-clip/washer fits in a groove above the shoulder where the threads stop, so the c-clip is only there to hold the screws/springs captive. The shoulder sticking down past them is what would hit or bottom out on the tops of the posts, so using straight screws would be a requirement. You would not be able to use these screws in the manner I had planned (which is what you are also suggesting). I believe the way those screws are designed might be a contributor to the absence of contact. Using straight screws will work as long as the heat plate itself has no interference with the tops of the posts. I cannot get my eyes in that area to see well enough. An arthroscopic surgery "snake" camera would be handy to have in a situation like this.

 

I'm not sure if it's the modded drivers or what but I can't get SLI to scale properly in titles/benchmarks I know it should, single card and clocks are fine but the utilisation is not there.

 

Have you tried one of the @j95 desktop driver mods? That is what I was using before and after 7700K and nothing changed. SLI still works correctly and no tweaking for 7700K was necessary... totally plug and play.

Here is a link to the thread... all recent versions available. My favorite is 375.70. https://www.techinferno.com/index.p...er-support-modded-inf/&page=63#comment-153562

Be sure to follow all instructions, including deleting the garbage folders that are not necessary and using DDU to erase all of the problem-causing registry keys installed by notebook drivers.

 

I used #1 Phillips drive M3-0.5 thread size 6mm length fully threaded screws for my mod. Don't know if they changed the screw used between the ZM and 870DM...

Sent from my SM-G900P using Tapatalk

 

I'll give that a try.

 

Thanks for that info. I'll have to check on the size.

I think you will be pleased. Be sure to follow all instructions, including deleting the garbage folders that are not necessary and using DDU to erase all of the problem-causing registry keys installed by notebook drivers.

 

oh yes, many times already while fiddling around trying to get rid of my gpu-based black screens and im still using the same pads on it, they have great resuseability, one of the reasons i got them in the first place (the other one being the absolute AWESOME price/performance ratio )

on another note, nice to see my good ol` 6700K being able to reach 5 ghz, after all (48x on the cache, but this is only a 47x cache screenie), albeit at ridiculous VCore of 1.46V :

 

I noticed that issue since 376.30 , 376.19 works as it should. The 376.66 Dev is much better though.

 

Yes, 376.66 dev was pretty decent. It lowered my Fire Strike score by a small amount and had some weird flickering things going on, so I went back to 375.70. Everything up to 379.95 was actually pretty decent for me.

 

Also, if anyone has a pressure plate, I'd love for them to do a torque reading for the screws as torque doesn't convert to PSI, and that is what I previously calculated for the socket tolerance (no information for force on fire is available). The psi on the socket is what determined the pressure film I used. Meanwhile, the four screws contribute 1/4th of the downward force, so if someone has a way to test the downward force (like a pressure plate) and determine a torque for each screw, we'd just have to get a bicycle torque screw (a driver that works in in/lbs or very low ft/lbs) and could recreate that exact pressure by way of using a torque screwdriver...

Sent from my SM-G900P using Tapatalk

 

Weird. For my 376.66Dev works better than 376.19, which was better than all other releases.

All thanks to @j95 though.

 

What size pads did you use for the CPU VRMs?

 

The 376.66 Vulkan driver is actually using the older 375 branch.
376.60 is still the latest until they finally get their WQHL certification from MS...

Edit:

Driver 376.66 is VK375_17-21
Driver 376.62 is r375_00-229
Driver 376.60 is r376_33-10
Driver 376.56 is r375_00-217
Driver 376.48 is r376_33-9
Driver 376.39 is r375_00-197
Driver 376.33 is r375_00-190
Driver 376.19 is r376_06-6
Driver 376.09 is r376_06-3
Driver 375.95 is r375_86-5
Driver 375.86 is r375_00-134
Driver 375.76 is r375_70-3
Driver 375.70 is r375_00-108
Driver 375.63 is r375_45-10
Driver 375.57 is r375_45-7

 

0.5mm and either 1mm or 1.5mm, it has been a long time.

 

That is a lot less than stock size, maybe that was enough to get the extra pressure from the springs for the CLU.

@Mr. Fox

Edit: Stock CPU thermal pad size is 1mm and 2.5mm

 

Actually that is the same case with the DM1 CPU heatsink as well.
I reduced the thermal pad thickness to .5 and 1.5 from the stock ones and my contact was perfect after that.

 

Thanks for the suggestion. When I take it apart next time I will use the pressure film to see if there is any contact between the IHS and heat sink with the thermal pads totally removed for the testing. I have a feeling that might not be enough of a change to help, but it certainly bears further investigation. The pads are severely dented from the MOSFETs and VRMs, so that alone is an indication they are way too thick.

We're going to find out. I hope so. That would be too easy.

So, you are no longer using a shim on the CPU? I thought you were still using a shim.

 

Nope, no shim. Only Kryonaut on the IHS/HEATSINK and CLU on die.

 

I love you guys. Ya'll like testing as much as I do.

 

I've tried a few drivers now and all are refusing to scale properly :s

 

Something you guys might find useful.

http://overclocking.guide/optimizing-ram-performance-time-spy/

 

Checks in line with what I say. Speed is a great benefit, but timings are also key.

 

I would say that keeping the laptop off the table can decrease the max temps with 8C or so if it is 3 cm or more (?)

Was testing this to give a hand to those who have higher temps.

 

Saw this months ago... But thank you!

Sent from my SM-G900P using Tapatalk

 

Got the Sager Bios installed and my 7700K installed. Here were just some quick benches. Have not tested much but was able to hit 5.0 in the bios at adaptive 0mV. 5.1 when I ran cinebench crashed but have not tried adjusting the voltage yet.

 

I've been a multi-year believer in the folded toilet paper mod.