Clevo 2020

Obsidian Fan Control works better than CCC. Fn+1 sets fans to max.

 

Obsidian-PC Fan Control ist ghodly.

 

I reckon the "x170" to be LGA1200 whenever it launches in 2020.

 

there are rumors of cometlake getting pushed back but that is expected when it launch. 10 cores on 14nm+++ and rocketlake maybe 10 cores or 12 cores on 14nm+++ with tigerlake design unsure.

10 cores under some heavy overclocking load will easily use 300w this machine is gonna have one hell of a problem keeping it cool.

 

10 cores sure does sound packed on 14nm+++. Overclocked 8 core is already at the limit for P870TM1 both thermally and electrically as that is one hot and power hungry chip. But it's still impressive to run 9900k at 5ghz in this chassis, albeit for benchmarks and games.

 

Anything is going to start having issues as those cores are densely packed.

 

14nm+++++ lesser that density all gud to go, just give us double the heatsink size with vapor chamber

 

That would be nice, but from looking at the original x170 cooler leak, I kinda doubt it. And we wont see anything for the P870 sadly, as it's EOL.

 

Double the size so no one buys it at all tends to be a bad plan lol.

 

Double the heatsink size doesn't mean double the laptop.

 

if they're smart about it. reuse the chassis/mobo with minor mobo adjustment such as socket and additional m.2 slots. re-route x8 PCIe lanes for additional storage directly to the CPU. then expand CPU heatsink towards where 2nd GPU fan is located, give that middle fan to cool for primary GPU AND CPU, add the small 4th fan call it a day.

now 10 cores should run just fine at like 4.7-4.8ghz and they still reduce lots of cost instead of redesign and remaking new parts.

@Papusan they saved all those money reusing the chassis for 3-4 yrs, this is the time to reuse it but they choose not to, bad idea.

 

I'd buy it.

 

That would be great, as the 870 chassis has room to expand. Seems like it would be logical to not have to reinvent the wheel and work on existing chassis. But as Meaker pointed out, maybe it's just not cost effective for them as people tend to gravitate towards thinner more rgb'ish laptops.

 

Even on here the p570wm owners thread was not exactly hugely populated.

 

Im waiting for a laptop that breaks from the Fans in the rear methodology, give more length versus thickness of the heat exchangers, as Ill be doing exactly that to my Ranger.

 

those were old times people aren't too into big laptops. since MSI/ASUS and acer has being pumping these out, same with AW, its worth a try again.

 

I'd say the p775 is large enough to get the job done with some tweaks and you are not risking the company to do that.

 

Perhaps for the p775tm1, you could pack more heatsink and cooling real estate if you were to discard the space used for the 2.5" inch drives. But then again wishful thinking.

 

Depending on how much you can minimize the board size and battery footprint it might not have much of an effect at all.

 

870 wasnt that much bigger than 770 though yet it can fit so much more. extra 2.5" and 1 more GPU, heatsink, fan, clearly a much better thought out design.

definitely need a lot more copper for capacity reason. lots of copper big heatsink = capacity = quiet system and cool system. takes a long time before the fan to start spinning and just overall more surface area.

 

Extra M.2, you mean. Both have the same number of 2.5” bays (2).

 

hard to give up on 2.5" cause they can store so much but extra m.2 is fine too. eventually it'll be moved there and m.2 capacity drives capacity will increase too. 8TB should be soon within a yr or 2.

 

I'd consider 25% extra by weight and volume on an already large machine to be very significant. That's not including the second brick. It has a marked impact on portability.

 

According to Prema, the X170 will have LGA 1200 Comet Lake CPU and MXM graphics in Q2 2020:

Spoiler

 

Interesting

 

I think we all expected the x170 with the new socket, but the layout of the cooler didn't quite look like it was for the usual MXM layout. That had us worried that it perhaps was BGA. Glad that it's not.

 

Well there goes my money in 2020
If isn't a BGA crap lol

 

Hopefully it'll still be possible to use the 2020 MXM cards in p77x/p870x models.

 

Hope so. Would be an awesome upgrade for P775DM3-G with a 9900k and a TM1 heatsink. Hope so but I don't have my hopes up seeing Clevo changing the sizes of the mxm gpu's over the years. Marketing tactics

 

They will probably retain their current MXM design just perhaps move things a bit like they did from 1080 to 2080.

Also means I'm SOL with my Area-51m for CPU upgrades (Thanks Intel) but hoping for GPU still.

 

Yeah, you can hardly say that it's as upgradable and future proof, when the only post launch CPU upgrade you got was basically a better binned chip.

 

To be honest I think that's mostly Intel's fault. I mean we can run from 8700K to 9900K/KS/KF so the CPU is upgradable for people who bought the lower tier ones LOL

I'm pretty sure we'll see an R2 with the LGA 1200 socket. I was kinda surprised Intel needed a new socket. If I remember correctly, the new 10 core is still running on Coffee Lake hardware?

 

I was thinking more in the lines of people who got the "top tier" 9900k/2080 combos.
But the new socket does bring more cores. But we're still stuck with pci-e 3.0

 

We saw Z170 being able to go 4-core>6-core>8-core

Z390 was made to fully support those 8-cores

If they're using the same architecture from the previous gen for 10-core, I am sure they could have done the same.

I'm betting they made this new socket 1200 to kind of get rid of people upgrading from 6/8-core systems to the 10-core. And in the future we may see even 12-cores if they're upgrading cores now and a new socket.

Even der8auer showed the pins were capable of handling higher core counts with Z170, it was just VRMs at that point.

I think I saw a post somewhere saying Intel is skipping PCIe 4.0 and going straight to 5.0

 

The initial socket was designed for 4 cores, we have gone up to 8 with some modifications for power. Another 25% increased potential power draw again and it's going to need a bit of work on the socket.

 

Oh that makes Donald look like he has a mohawk and that would be awesome lol.

 

the puny heatsink just isn't going to cut it. gambling on taobao making some custom heatsink expandable to several fans looks like thats a dead end too, theres already a waterblock made of it which is external.

the 10 cores to 8 core won't be much of an issue because 10 cores likely binned better than 8. intel will decrease density yet again to accommodate more cores like they did from 4 to 6 to 8.

new sl binned 5.2ghz 9900ks goes for 1.325v. my 8700k was 5.2 at 1.425 i believe. not to mention the default value changed from 210 to 140 mohms? (w/e the name of that value is @Falkentyne might know) so it's a pretty big improvement.

 

1.6 mOhms (8 core)
2.1 mOhms (4/6 core)

 

fml 1.6. well still better than 2.1. when i ran my 8700k difference amount to peak v reduction, so to speak a 1.4v chip using 2.1 would have a higher peak v in respect to a chip 1.4v chip using 1.6. might make zero sense as both chips using 1.4 but if ran with ac/dc at value of 0.01, it increases v required to have these chips stable due to removal of vdroop.

less mOhms means less upping the voltage when running them at 0.01. too bad it cost upward $1250 for a 5.2 chip.

 

Actually, DC 0.01 doesn't remove vdroop!

Unless @Prema somehow was able to 'route' DC Loadline to VRM Loadline somehow, which would be near impossible (well maybe not for him) as I've never seen a motherboard which links DC Loadline to VRM loadline.

In order for DC Loadline to affect VRM Loadline, DC Loadline must be able to communicate directly with the VRM and issue a loadline command, but the problem is, Intel specifically says that DC Loadline only affects "Power measurements" (CPU Package Power=VID * Amps).

So....
It simply removes the droop on the VID but the vcore isn't linked to VID after AC Loadline affects it! So the VRM gets the voltage target from Thermal Velocity Boost (this raises or lowers the base multiplier VID based on temps) then AC Loadline affects it, then sends it to the VRM as a voltage signal (by sending it off the CPU +12v line--that's why it's called AC Loadline--it controls operating voltages). Then the VRM droops it by whatever the loadline is set for for that CPU SKU line.

Only then does DC Loadline change the VID downwards (for the purpose of power measurements via CPU Package Power).

So all you're doing is removing the AC Loadline boost from the "default" VID and relying on 'thermal velocity boost' voltage optimizations instead (-1.5mv every 1C temp drop starting at 100C) and you're getting the full 1.6 mOhms of vdroop, but AC loadline isn't boosting the VID target (signal to the VRM) to 'help' you out. DC Loadline only takes effect AFTER the VRM has the voltage target.

So setting AC Loadline to 0.01 mOhms and DC Loadline to 0.01 mOhms gives a vcore equal to the default CPU VID + whatever Thermal Velocity boost boosted it up to.
Then DC Loadline would show that VID to the operating system!

If you want to see what your CPU is actually using, you need to set DC Loadline to 1.6 mOhms (160). Then you will see what AC loadline+TVB+Vdroop is setting for your actual vcore.

I didn't know this until I messed around with the Aorus Master, which has 'VR VOUT' access to show the true CPU Voltage reading from the VRM directly (on-die sense). Then I set Vcore to "Auto" to function like the laptops do, and set Loadline Calibration to Intel Defaults (Standard or Normal in the bios), and compared the "CPU VID" to the "VR VOUT".

When I changed DC Loadline, the VID changed based on that value, but VR VOUT did not change at all. It was only affected by Loadline Calibration and AC Loadline (limited to 1.520v maximum cap before vdroop)..

The APTIO bios capsule has a setting called "DC Loadline Override" on some bioses, in the "CPU Overclocking Menu" section, which states clearly that this changes the 'VRM Loadline', and it has an actual "mOhms" selection value in "uOhms", but the people who unlocked it on MSI BGAbook Bioses said it did absolutely nothing, like it isn't wired to anything.

------

When you guys set a "manual" (override) Vcore on these laptops, you are NOT setting an actual vcore like you are doing on desktops.

You're overriding the CPU's default VID with a "Custom" VID, which is called "override voltage." That's why the VID changes when you enter in a 'manual' vcore.

That's why AC Loadline affects you when you set an override voltage.
On desktops, setting a manual "Vcore" programs the VRM directly with a target voltage, bypassing the AC Loadline completely (thus VID is affected by the ACLL value on desktops, but Vcore is not affected).

 

It nets you 10% better power consumption for top flight consumer binning on the 9900ks.

That still leaves a hefty change and would kill yields.

 

Damn man, that's some write up !
fair play !!

 

The WC heatblocks from Taobao are actually VERY good, i had one and reported full stats before and after against the p870 vapour chamber...
But yes it was external, between 20-30% better across GPU's and CPU to be exact - 30% when only the CPU or GPU's were being used but lowest was 20% gain when both being stressed same time.

 

so thats what happening internally. basically what im seeing is just the result of all of those coming together hitting hwinfo. where does the older system fit in this? maybe the ac/dc thing existed way back in the days just not in any of the laptop bios as an option.

The effect im trying to achieve with this, adjusting the ac/dc values is to get similar behavior of older generation cpus. back in the days with ivy, you'd set a manual voltage for stability then let everything else does it's own thing, issue with 870 (not sure if this applies to other machine) is that if a static v was set i could not get cstate/core parking to work properly and TS can't seem to override what is in my window power setting profile unlike when i had my ivy laptop.

to work around this and get similar behavior, only adaptive v would work. the strange thing is as you'd require higher voltage to run these cpu stable going from 210 mOhms down to 1, the CPU runs cooler which made little sense. it benefited me so i sticked with it just worked out that way.

 

It's certainly designed for adaptive as a laptop rather than static.

 

Wait, does that need to be changed for the 9900k in the bios for the Clevo machines?

Spoiler

@Falkentyne @ole!!!

 

according to @Falkentyne the 9th gen uses 160 on default. did you get yours starting with 8th gen? 210 is default for 8th gen.

lowering those value lowers peak v reading and up the under load voltage. i find it lowering those value lowers the temp as well, by noticeable amount 3-4c

 

Interesting. No, initially HIDevolution had sat it up for my 9900k, but I've recently been memory overclocking (lots of cmos resets), so this is just what it defaulted to.

Should it default to "Core VR loadline override to 1600 mOhms" and "AC/DC Loadline to 160" for the 9900k?

 

Yes, AC and DC should be 1.6 mOhms.

But does "DC Loadline Override" (uOhm) setting even work?
On every bios I've looked at, DC loadline override (even on Gigabyte Z390) defaults to 2100 uOhm, but the setting doesn't seem to be linked to anything that works, since "Loadline Calibration" is linked to the VRM instead.

Well, it doesn't work on the MSI laptops.

Since VID is not linked to VRM loadline calibration at all (DC Loadline is not VRM loadline), the only way to test this is to set DC Loadline Override to something like 100 uOhm or 400 uOhm or 800 uOhm (try to avoid 0 uOhm) and see if your temps increase. Your temps should skyrocket if Loadline Override is set to 100 uOhm.

Without a vcore sensor, you wouldn't be able to tell what's going on without looking at temps.

 

I'd go with the 9900k default, it's likely matched to the larger die.