Clevo Overclocker's Lounge

Doesn't the System Agent handle that?
I'm sure TS let's us modify voltage on that.

 

Yeah It might be SA voltage for skylake. Might need to bump VCCIO as well.

 

Yes, it is adjustable in the Mod BIOS.

 

Yes, we have that in the BIOS, XTU and ThrottleStop. I did not know that increased memory controller input voltage. In that case my answer is "no, that doesn't help" because I have elevated System Agent and so far I have not seen any improvement in RAM stability at times when the memory decides it want to be unstable. I had that set at 2.100V at one point and there was no improvement or change in the memory instability problems. I normally have it set to 2.00V (and have it set there now). @Papusan and I have been running 6700K with it set to 2001 for a very long time.

 

Just remember if you want to be on the bleeding edge there is usually some extra work and planning involved.

 

Try VCCIO.

Otherwise, try loosing your tRFC(Doesnt impact performance much but helps with stability considerably.

 

We don't have VCCIO unless Clevo calls it something else. I have IA voltage elevated as well as SA and neither one of those does anything to improve RAM instability.

Already ahead of you on the tRFC. That definitely helps more than anything else does, and I shared that some time ago. Unfortunately, there is on way to change that except manually using the Custom Profile in the BIOS. The SPD and XMP profiles have nothing for tRFC, so I cannot program that using Thaiphoon Burner. If you look at all of the manual 3000 and 3200 settings I have shared photos of, tRFC is always increased to improve stability and reduce the read/write errors.

We used to be able to set Ring/Cache independently, but that is no longer available. I have to use ThrottleStop to do that separately because neither the BIOS nor XTU allows it.

 

Hmmm... IO voltage? @Prema

I know it matters for SK-mem OCing.

 

THX @Meaker@Sager

Used your thermal pad guide as guideline to re-pad the entire system with Fujipoly (except for the two 3.5mm mosfets pads). One of your 1mm marked pads is actually 1.5mm (single power phase above the SLI connecter).

 

Can you point it out in the image for me so I know for sure the one you mean?

 

This one is 1.5mm (not same height as the upper part of the cooler where the 1mm pads are, but same power phase height on the mxm board side:



Also replaced all the other ones with 2mm, 2.5mm & 3mm except for the 3.5mm as Fujupoly doesn't offer them afaik. At least I don't have any...
Ran out of 0.5mm as well, so couldn't do the GRID connection to the CPU, but did the CPU pads themselves.

CPU is now delidded and using CLU, but left outside & GPU TIM as ICD for now until I man-up enough to go full metal...

 

Oh forum images, when will they work?

 

Sorry @Mr. Fox and @Johnksss@iBUYPOWER I'm just seeing these posts. I just downloaded CPU-Z and GPU-Z. Here's the picture if I can get it to work.

 

The .inf mod I posted should have worked with that just fine...but the beta driver does the same job.

 

I'm POSITIVE I did something Boneheaded. But I'm working on it. Thanks again for all the help

 

If the INF Prema provided did not have 1070 HWID, that's probably why it was not working. You have a 1070 and may not be G-Sync version (that also changes the HWID).

The @j95 mod should work fine for you.

Didn't see your post before I added the above.

 

The system is working now, It was something I did wrong. I'm still a newbie with computers but getting there.

 

You can see the settings I am using in this video...
SA Voltage Limit 2001 (2.001V) and same for Core/IA Voltage Limit... setting zero (auto) and as high as 2400 (2.400V) had no effect on RAM stability.

I do not find any VCCIO or CPU IO voltage setting available. I have that for the P750ZM, but not the P870DM1 or DM3.

 

Wait, what other ones?

And GRID connection to the CPU needs 0.5mm? (I'm looking for 145 x 145mm in 0.5, 1 and 1.5 thicknesses, will I need more sizes?)

 

Umm you might need to tweak it manually in ThrottleStop.
There is VCCIO as well, but not sure if it's tweakable.

 

Are you referring to Analog I/O in ThrottleStop?

 

Nope that's the uncore. There is a System Agent section to it's right.

 

Yep, Grid connection uses 0.5mm. You only need very little 2, 2.5, 3 & 3.5mm if you want to do them all. But 0.5-1.5mm will do the important ones just as good.

 

So the pads marked "original pads" I can just leave in place? Sounds good to me.

I want to make sure the machine will be the bee's knees, tip top, cool, powerful, perfect. I'm going to do it.

 

Yep. The only pads I did different, from the ones Meaker replaced as well, is the one marked orange, which are now 1.5mm pads. Everything else was spot-on and saved me a lot of time.

 

I thought you were talking about VCCIN.

Yeah, probably will have to use ThrottleStop for that for SA voltage, if it actually works. You can adjust the limits, but not the actual value, in the BIOS. There is no setting for SA voltage there or with XTU. I will play with that in ThrottleStop to see if it changes anything with the memory. If ThrottleStop does not change it, there may be no way to control that voltage value.

 

Great, good to know. I was going to use that picture as a default (I already got it bookmarked).

 

Anyone here able to find Fujipoly 60 x 50 x 0.5 - 17.0 W/mK in Canada?

 

To clear some confusions.
None of those settings work.
SA does not work
Vccio is greyed out and does not work.
Voltage in Throttle stop now works over a curve and not flat out on the voltage you set it too.
Throttle stop for changing cache and cpu voltage independently.
If you dont use Real XTU then you can't make on the fly bclk changes or voltage & multiplier changes when throttle stop quits working. Means you have to reboot and hope it works or may take a few reboots.
Uncore is used to help stabilize ram as well. That seems to have more impact on it than the SA settings which have zero impact.
CPU pll has impact on system, but too much and your system wont boot.
These are just some of the things i noticed...

 

About the CPU PLL, even the slightest change seems to cause no boot with KBL. With SKL it was working last time I used it.
Any idea why?

 

Here's one of my first.

 

No, not yet.

 

Well, here is something interesting on the temps. Turns out to be my ludicrous temps after changing CPUs is due to pathetic heat sink fit... typical of Clevo, unfortunately. Temps were decent with 6700K so I assumed the fit was good, but the gap was plugged with a butt-load of IC Diamond and that was masking the issue. @Donald@HIDevolution and @Ted@HIDevolution and @Eurocom Support and @Prostar Computer are probably going to want to see this. I previously tried using CLU with the 7700K and it was super hot, hit nearly 100°C with no overclock, which should have been my first clue the famous sloppy Clevo heat sink issue had struck again.

So, I decided I could not go any further with these truly asinine thermals. I pulled off the CPU heat sink and it looked like the spread of the IC Diamond was good. Cleaned it up and set the pressure sensitive film in place, clamped it down and guess what... a picture says 1,000 words. TL;DR - no pressure whatsoever. So I grabbed a 30mm x 30mm x 0.5mm copper shim from my stash of stuff, slapped CLU on all 4 mating surfaces and now the temps are just about appropriate.

Between Prostar, Eurocom and @Prema having key contacts in high places at Clevo, a loud and clear message needs to go out that correcting this problem once and for all needs to be job #1 starting today. This is now my third Clevo (out of five, counting the Sky X9 review unit) that had CPU heat sinks that were trash. My P570WM, P750ZM and now this new one had CPU heat sinks that are worthless. I paid for the first two out of my own pocket because the previous owners put up with it and I inherited their garbage. Only the Sky X9 and the P870DM-G that I bought from @tanzmeister and sold to @Rage Set had a CPU heat sink that fit worth a damn from the factory. And, there are a number of other customers that I know of who have faced the same problem. Sloppy, piss-poor heat sink QC has to stop if Clevo want people to continue buying their stuff. Thermal management excellence should be something they are burning a lot more calories on getting right than what they have been.



These temps are with the 0.5mm shim with CLU between the heat sink and IHS. So, between the delid and adding the shim to help offset the sloppy heat sink fit, we've got about a 25°C temperature decrease.

Without setting the laptop on my AC unit, before the shim was added this would have been a thermal shutdown running wPrime 1024M benchmark using nothing but fans. It was almost this hot sitting on the AC unit, LOL.


Edit: Just for good measure, I proved it to myself again with a second run of wPrime 1024M and temps were still good with the shim. It would have been +100°C and shutting down half-way through the benchmark without the shim.

 

A new set of screws would do the trick...
@Meaker@Sager what did you do when using CLU on the CPU heatsink to get the two closer together?

 

I'm all ears, Brother @Prema - hopefully Brother @Meaker@Sager can shed some light on the fix if he encountered the same issue. Even 5.0GHz after the shim, only using fans, is only 87°C in Cinebench. I'd like to ditch the shim. I looked into modding the screws by removing the c-clip but the screws have a shoulder on them and that would accomplish nothing as far as fit. The shoulder would still bottom out at the same point and the fit would not be improved like I initially hoped might be the case. We would need screws without a shoulder, or with a shorter space between the screw head and edge of the shoulder to create some contact pressure. That, or shim the CPU support plate to move it away from the motherboard a 0.5mm to make the stand-offs for the screws not as long on the heat sink side of the motherboard.



I also checked the heat plate with a machined steel ruler and it also was not flat. It is arched away from the IHS (opposite of what it should be) just like the P750ZM heat sink was. The copper plate is so thin and weak I was able to flatten it out with my bare hands, same as I did with the P750ZM heat sink. This should be a precision fit with a thicker copper heat plate strong enough not to lose its shape.

 

It's because of this and this alone that I feel if a GT73VR whitebook like the 16L13 comes out (desktop CPU, 1070 SLI or 1080 with the Chi Mei screen) and Eurocom & HID etc start selling it... well... P775DMx instantly becomes invalid. Prema mod fixed CCC being a general pain, but the heatsink thing is... yeah.

Actually, with @John@OBSIDIAN-PC doing so much work on each notebook before selling it... why don't you actually return every single bad heatsink to Clevo and demand new ones? I'm pretty sure if they get a couple hundred heatsinks back over the course of every few months that they might get the picture. Because improper heatsink fit should be something you can request a replacement for, correct?

 

No idea how much I have to wait for my Clevo laptop (things got really messed up), so I have bought a cheap Acer laptop for backup so I could play some games while I wait (to my surprise the GTX 950M can handle more games than I thought),
I ran FireStrike to check if the CPU frequency fluctuates there too like it does with Clevo laptop, the result:



Looks okay to me

 

Thanks for your insight once again, I ordered mine with CLU and this was my fear since my temps on the CLU didn't seem to be as good as yours with ICD. Between the GPU heatsink as having dimples in it and now this I really hope Clevo can get their act together and tighten up their tolerances with the machined components.

 

These are my exactly my sentiments too. Both my 870 DM1 and 870 DM3 each required a new Heatsink.
In both cases the CPU's would hit 100 degrees at stock clocks or below with an undervolt just like Brother Fox's unit did. That's a 100% strike rate of defect Heatsinks that should have never left the factory.
I was mortified to discover how little pressure there was and how much TIM was required to make contact - to the point it was also insulating.

These rubbish Heatsink's need to
stop getting sent out by Clevo!
It's unacceptable and it baffles as to how it's managed to go on for this long with complete disregard to step up and manage the problem.

 

Spoiler

 

Spoiler

 

He's now the Trojan entering the kingdom of MSI just to destroy it from the inside once their throttle kicks in and no-one comes to the rescue...he will grab his SSDs and start running as fast as he can to put them into something actually deserving their cost.

I am from the future btw...

 

now THAT explains a lot!

 

Hello, my name is:

Persistent
Recognitive
Engineering
Mod
Algorithm

thanks for participating in our studies!

One more tip, don't trust deep-learning / artificial intelligence / AI...

OK, gotta go back to 2108, we are gonna go diving to the remains of Disney World Florida today...it'll be wonderful.

 

I'm a bit confused on this. I don't think it is the surface of the heatsink itself, correct? If so, then the i7-6700 wouldn't have problems too.

Is the problem the i7-7700 has a lesser height (z-axis) than the i7-6700 that the mounts won't allow the heat sink to go lower and make better contact?

 

That's what I was wondering too. I didn't delid the 7700K I ordered because it's going back anyways, but the temps shoot up suspiciously quickly. I had fantastic cooling on my 6700K with liquid metal, so the heat sink is perfectly fine. Once I get my new 7700K and pop this one out to return, I'll get out my caliper and see what the z heights look like.

 

That is possible, I did not check. The heat sink plate is still cheap/weak/thin and loses its shape so that still needs to be made better. I do not own any calipers, but I could lay them down on the IHS, pin pad side up, and see if the 6700K is visibly taller than the 7700K.

 

that would surely be quite interesting to see!

 

Thanks for confirming that Prema. I will go back in when I install my 7700K and change out that one pad or more again if they tear. I am actually running low on alphacool pads and will probably use ARCTIC pads I bought a while back. Anyone else use ARCTIC 6.0 W/mK pads yet and how did they work out for you?

 

This is probably a foolish question, but can the IHS from the 6700k be used on the 7700k? I had them both sitting in front of me and didn't think to look or take a picture.