*** Official Sager NP9175 / Clevo P775TM Owner's Lounge! ***

Won't happen(ODMs have locked out for dangerous voltage). But some play with etc AC/DC Loadline and will see their chips run into hardware heaven before than later

 

You wont believe what this systems smoke effects are like, so realistic.

 

The magic blue smoke?

 

Smellovision is included for free too.

 

Well it better be, considering it's a 500$ CPU..

 

AC Loadline is still capped at 1.520v target to the VRM, unless "Serial VID Offset" command is set (VRM command 33h This is NOT the same as 'offset voltage' for setting vcore/VID). Raising it higher than 1.6 or 2.1 mOhms will raise the idle voltage only, if the load boost from the resistance (AC Circuit) hits 1.520v. And that's before vdroop. Vdroop can't be removed, and is either 1.6 mOhms or 2.1 mOhms of VRM Loadline (Intel spec for 8 and 6 cores)

The IR 35201 can accept up to 63 mOhms, but VID is still limited to 1.520v without SVID offset command 33h or 1.720v with SVID offset command.

No one seems to know if the "Platform Voltage Overrides" option "DC Loadline Override (uOhm) does anything at all or not. (This is NOT the same as "DC Loadline" in the VR domain section, which controls power measurements (reported VID and CPU Package Power MSR)). Probably depends if the VRM supports changing the loadline, and if that option is wired to the bit that controls it. The APTIO says this is for VRM loadline, and defaults to 2100 uOhm (the value for 6 core processors). A MSI throttlebook user tried changing it and nothing happened.

 

Prema did say to leave it default as it only causes instability on a Clevo. And I kinda have a feeling that if I mess with it, I'll end up ordering a new board for xmas

 

It's more than about the Voltage... Going beyond specs can have unexpected consequences. I talk about best longevity and troubleless experience for the users. Not that it will go into fire once you change some settings for test purposes.

http://forum.notebookreview.com/thr...cussions-lounge.806253/page-105#post-10721289

And we have seen the trickery from AW's Dell's firmware engineers and last years AW models. Power consumption vs, voltage etc is totally wacko and out of specs. I expect we will see more of the AW models pop up with broken/damaged MB etc due the engineers fiddling with something they don't know even how works...

 

I'm 200% sure that Prema said leave the "AC Loadline" value alone.
No one ever mentioned the "DC Loadline Override (uOhm)" value before.

 

Have anyone found an unlocked version of bios for p775tm1? I have flashed xmg version, its not bad, however would like to have options to disable some features such as wake up on mouse/keyboard/usb from bios... Can anyone share their experience?

 

I don't think that's something to worry about in modern transistor design in terms of undervolting.

We are not using it as an analogue transistor.

 

In the end a transistor is an Analogue electronic device (MOS-FET). What makes to call it Digital is just how you work with Gate voltage levels.
What stated @Papusan is true, the Source outputs are applied to other transistors Gates, and this means Gate input voltage has to deal with main power rails (cores).

Minimal voltage is somewhere ~0.55V, and each milivolt counts a lot here. Pinning it with less, let's say 0.5V then the internal circuitry loose their state, resulting in a hazard, huge power consumption plus transistors are not in their optimal limits which means more power consumption. I am not sure what are the real values or parameters of these transistors like open time, saturation speed...
I believe the SA keeps track of what's happening on the other modules and has some failsafe, when cores do not report with some kind of whatchdog signal it tries to somehow protect the CPU from going crazy, like locking all gates to a low level... this is just an idea.

 

0.5v is below the threshold voltage where you stop treating it as a digital transistor and where the chip will cease to function.

 

Probably the only thing to worry about

 

Basically worry about anything else.

 

Only thing that worries me is, when summer comes, I'll have to downclock.

 

Maybe me too. My home is usually 70F in the winter. But during summer time in the dead of heat my A/C at home struggles to maintain temps under 78F.

we shall see but 5Ghz with a (0) AVX offset may no longer be acceptable 24/7.

 

Have to drop down to a measly 4.9

 

When I first got it back together I was testing 4.9Ghz. It runs at 4.9 like its stock. 4.9 allows massive undervolting. A silly 100Mhz extra adds like 25 watts and 12-15C higher temps just to get that legendary 5Ghz. It’s almost as if the cpu wants you to fail.

 

Benching laptops in the summer is no fun unless... AC!

 

Or walk in freezer

 

That's what the chips do, every 100mhz costs a lot past the knee of the efficiency curve.

 

And here I am, struggling to make i5-8600k stable under 4.5 with a mixed workload. Guess, I picked the "best" in the lottery.

 

struggling with 4.5Ghz on a 8600K? Man that should be easy.

Delid, liquid metal, replace thermal pads around cpu socket, and polish the heatsink and cpu IHS. And then start dropping voltage off at 4.5Ghz. You should be able to
Run crazy low temps as these speeds. Maybe 55-60C load in R15.

 

Yea if you had a walk in freezer you could just leave the laptop inside and run an external monitor and peripherals outside on a desk lol. I’d be gaming at 5.5Ghz if I had one of those freezers haha.

 

Just a single thunderbolt cable would do the trick.

 

The i5s did tend to be a step down in silicon quality on average.

 

I feel like if I turned off HT I could run much higher frequencies. Never tried it though. The gains with HT are incredible. So I always thought a i5 8600K would really run 5Ghz like it’s nothing. I guess not.

 

Bin statistics from SL

https://siliconlottery.com/pages/statistics

 

You could probably do 5.1GHz on your 8086K if you turned off HT, assuming you don’t become voltage limited. 5GHz is pretty easy for 8th gen Coffee Lake in desktops since you can easily give them more than 1.3V to stabilize those higher speeds. These laptops are limited to about 1.3V under load before they start throttling, so 5GHz is usually out of reach unless you have a good sample or a binned chip like the 8086K that does 2-3 bins higher than a typical 8600K/8700K at the same voltage.

 

I know, right? I thought so too, the CPU is already delidded, although, I haven't lapped my IHS. The voltage is already low enough, and I could even get 4.8 from time to time, but it's just for usual, non-AVX workload. For some reason, I have one, and only one app, that always crashes on 4.5 no matter the reason, even without undervolt, which is weird. Guess, I gotta check more on that, but I'm not sure if I dig up the cause of the issue.

 

Yeah, I saw that, but I would never reach these voltages on my laptop, unless I do a professional lapping and delidding of my CPU with maybe even different heat spreader.

 

How long ago was it delidded? May be time to check this again. Also your app could be crashing due to heat.

I’ll say this 4.7Ghz-4.8Ghz with AVX 100% load on a 8086K with a -200-215 undervolt is the easiest overclock in the world. I mean it doesn’t even break a sweat doing this and mine. “Wasn’t even delidded”. And you do not even question stability with AVX at these speeds. Oh and, 4.9 to 5.1Ghz non AVX loads or 99% of gaming are really a pretty easy for a 8086K. I was after daily stable at 5Ghz with a 0 AVX offset 24/7 stable. It took some work, but I finally got it there.

Running 1080P in BFV really benefits from a 8 core CPU or maybe even a 10 core. So squeezing these little 6 cores to death really helps them perform like a 8 core.

 

Also did you relid with sealant or was it left loose?

 

I set turbo at 4.4 GHz for 8 cores, and 5.0 GHz for 7 cores or less in my i9 9900KS.
I have the fans in low speed and only the PS3 emulator use 8 cores with AVX2, and at 4.4 GHz the temps stays in the 80'sºc.
In all the other apps and games that i use, turbo is at 5.0 GHz but the temps are in the 60's-70'sºc.

The fans at max in my EUROCOM Sky X7C are very loud, without a doubt the worst thing of this laptop.

 

Wow, didn't think it's possible to put these CPUs into laptops without sacrifices. Maybe I should look into 9700k, since it's already soldered (sort of). My main concern is that, it's not stable under any load if i make a ~-50 undervolt under, let's say 4.5, basically any app will eventually crash. Not even talking about 4.8, where without undervolt (or otherwise crash), my limit is temperature - it hits 90s and PROCHOT.

Yup, sealant is there, used the one for heatsinks and such, so it's pretty tight. But it would be hard for me to make another delid, cause I don't have my tool anymore.

 

Problem with the soldered 9th gen is mainly the thicker die, twice the thickness of the 8th gen.

 

The sealant raises up the lid again defeating some of the gain from the delid.

 

True, although, I did a thin layer so it should be fine. Well, I was lucky with the thing anyway, cause I did it for the first time. Maybe next time? Maybe 9900 if it goes cheap?

 

You can always delid again

 

This is undoubtedly the most underrated mistake many do after delidding.

 

And liquid metal won't build as high as the solder. The gap between the lid and die increasing if you remove the STIM. Aka you run into double mistake

 

It's like brother @Mr. Fox said, measure twice, cut once. That goes a really long way, saving you lots of headache later.

 

It is best to omit any kind of sealant for the reason Brother @Meaker@Sager mentioned. It is totally unnecessary and any amount (no matter how thin) has the potential to interfere with contact between the IHS and die. The CPU retention bracket will hold everything securely without sealant, you simply need to hold the IHS in place with your thumb to keep it from sliding when you lock it down. (On a desktop it is best to lay the system down so the motherboard is horizontal, same as a laptop. This will avoid the possibility of the IHS accidentally coming loose from the CPU and dropping onto the GPU, potentially causing damage from the fall.)

Unless you used something too strong (never use Super Glue as it can de-laminate the CPU PCB) and you truly used very little, you can most likely delid with minimal effort. Maybe even a sheet of paper to cut through it or your fingernails will be enough to separate the IHS from the CPU. If you used RTV silicon, that is generally very easy to separate. In this case you could even use a razor blade with ease. Just keep the sharp edge parallel to the PCB to avoid cutting into it, and only slide the blade a few millimeters under the edge of the IHS on all four sides. (I did my first two delids with a razer blade--before purchasing the tool--and it was quite easy, with appropriate caution.)

The best type of razor blade to use is the old fashioned double-edged retractable type. They are super cheap, extra thin and crazy sharp. How thin they are makes them especially suited for slipping between the IHS and PCB. Use some tape on one side of the blade to make a handle and avoid cutting your fingers to shreds. If you fold the tape over the cutting edge and use two layers you should be good to go. I like using the cloth type of electric tape since it is thicker and not slippery like the black plastic type.

 

You can use Kryonaut or any thick thermal paste by applying four dabs each of the four corners of the IHS underside edges to help secure it from moving so easily, if you don't want to reseal the IHS with RTV (any RTV reseals should be four very tiny dabs in each corner). The Kryonaut will apply just enough tension to stop the IHS from moving by itself. It works amazingly well.

The RTV method is ok to use if you sanded the IHS Underside by 0.2mm to eliminate the Z-height issue. I would not use RTV at all (even tiny dabs) if the IHS edges were not sanded down first.

I don't think anyone thought of using regular thermal paste to secure a floating IHS before.

 

Yeah that would work fine if a person were concerned about it moving. Great idea. But, it's still not necessary to use anything if you just hold it in place firmly while latching the retention bracket. It doesn't have to be perfectly placed either. It could be off center a millimeter or so with no adverse implications.

Some people think they are sealing it with the intent to avoid liquid metal leaking out, but the viscosity of the liquid metal is such that the air space (if any) normally occupied by the sealant is so small (like the width a human hair or less) that there is no way any of it can escape through that gap.

 

Wow, @Mr. Fox , you gave me a motivation to do another delid after all. This thing is lying around anyway, so I'll try after the New Year.

Spoiler: METAL!

Yeah, razor blade should do a fine job, I already tried it before on this sealant.

 

I'll see how loose it is, I have some thermal paste left.

 

BTW, how are guys removing the liquid metal, that started merging with copper plate? It's been there for a year or so, and I can't quite get it with the isopropyl spirit. I have some on the heat spreader that goes to the GPU.

 

Use the scuff pad that comes with Liquid Ultra (if you have one) or something like 1500 grit wet sanding paper. The liquid metal will stick to that surface better as well. If it is too slick it is more difficult to apply because the liquid metal doesn't want to stick to it. You don't need to remove any of the stain and shouldn't try to do that. You only need to remove any dried clods so the surface is flat. The stain from the liquid metal leeching into the copper is actually beneficial. It fills pores and increases the surface contact area at a microscopic level.