*** Official Sager NP9877 / Clevo P870TM-G Owner's Lounge! - Phoenix 4 ***

Again. I only listed a bench from my everyday settings. Letting card loose at it's defaults does an even better job in leaving the 1080 behind lol. Like @m4gg0t said. If it weren't for being expensive I'd recommend it to everyone. But again. I didn't pay no where near that price for that card. Lol

 

I think a 7nm version could be quite tasty.

 

Next generation

 

Well yes, that's kind of my point.

 

Hello there,

I want the backlight of my laptop's keyboard to automatically turn off after a certain amount of idle time, and then when I press a keystroke or move the touchpad cursor to turn on again. Is there a way to get such behaviour? My CCC has an option called "set sleep timer" but once the lights turn off no matter what I do the lights never go back

 

Even with the software and keyboard shortcut?

 

When "sleep timer" reach it's idle time the only way I get the lights back on is by unchecking the "Enter sleep timer" option. the keyboard shortcut does nothing. I'm using Control Center v5.0001.1.73 by the way, downloaded from Clevo website.

 

that is absolutely beautiful. thanks for sharing

 

Hello brother. I come back to this post. You say in short the 9th gen original lid is 2.33 mm thick. The 9900K PCB is 1,15mm. This is more like 3.48mm without the gasket glue Intel use. I have measure my own 8700K and the full thickness is 4.37mm.


You see the difference? This is a lot and hence I wonder if this can be correct. The gasket glue will add thickness but not so much as near 1mm. I can understand Intel determine what thickness they prefer on two different gen chips, but around 20% lower

 

Intel do change the z height quite a bit.

 

@Papusan
I measured just the IHS dimensions itself. But that does make me wonder how thick the entire thing is. I'll measure some of mine like you did

 

They did make the substrate a bit thicker to help avoid warping at some point iirc.

 

They could still used same total Cpu package thickness.

Probably think incresed thickness would add higher teps

 

The chip itslef is a bit thicker for the solder I think and then the heat spreader needs to cope with larger desktop heatsinks.

 

Yeah, 9900k is a way thicker and hotter chip. To get decent thermals you need to go out of your way with this one.

Besides delidding the chip, lapping the die looks to drastically improve temps. However, the downside is there is no long term testing for how this impacts the chips lifespan, and whether LM will eventually kill it if you remove too much of the disfusion barrier.

Something I've been considering to try out and see how it compares to the 9900KS.

 

If it did the existing chips with thinner barriers would also be having issues.

 

Hello everyone, I understand correctly that there is no video input in the clevo p870? only through video capture? if so, what advise?

 

Guys,
is there any chance to put a 2080 into a 870tm1 (km1 w/ stock tm1 bios)? I can mod the heatsink, however as far as I know new bios is needed... thx

 

A thunderbolt based device is your best bet for high performance capture.

 

you are talking about external devices thunderbolt ?

 

Is there such a thing as internal? Either way yes, a thunderbolt capture device.

 

Good afternoon gentlemen. I have a conundrum. ive installed a Samsung 970 evo plus. I’ve format it fine and cloned the boot drive onto it fine and it boots normal, speed test is fine however I have no temps and it doesn’t show in the bios as I would expect. I also can’t update the driver with obsidian updater. In the bios, the sata mode is intel rst rather than ahci. Is this where the problem lies? Or is there something else I’m missing?

please and thank you...

 

Yes, it will hide the interface as the chipset is controlling it so firmware wont read it directly.

 

Cool thank you. I will try this when I get a mo:

I assume this is the solution...?

 

Yes that would take you back to a single drive config. Most drives don't need any manual intervention however.

 

Hello guys,

It's me again, trying to get the XMP profile of the Ballistix Sport RAM to work. Searching on the web most sites say that I have to enable XMP on the system BIOS, but I wasn't able to find out that option on the system BIOS (I'm running the Prema Mod version 1.05.EVOC2). So maybe XMP isn't supported or is automatically supported? I don't know. I tryed to select the profile on my BIOS called XMP 1 but then the system won't boot.

Reading on the service manual I found out this sentence:



I have only two modules installed... Reding the above information that means I have to disable dual channel to get XMP to work? But again I wasn't able to find the option to disable dual channel on the system BIOS. Does that means that I have to physically place one module on the RAM1 slot and the other module on the RAM3 module (or RAM 2 and RAM 4 modules) to force to disable dual channel? Currently I'm using the slots 1&2 and that enables dual channel. In the service manual also says:

Is that because if you install two RAM modules on lets say slots 1&3 you don't get dual channel?

PS: I know that 2 modules @ 2400 MHz in dual channel beats 2 modules @ 2666 MHz running in single channel, I just want to know how to correctly enable XMP on my system.

 

To find the XMP:
Advanced > OverClocking > Memory > Memory Profile

Usually the system will boot loop a few times while it's ram training.
You don't need to disable dual channel for XMP to work.

Dual channel works with this setup:
Dimm 1+2 (primary slots under the bottom cover)
Dimm 3+4 (secondary slots under the keyboard)
Dimm 1+4
Dimm 2+3

Single channel:
Dimm 1+3
Dimm 2+4

 

Thanks for the answer!

How many loops are we talking about? .. when I tryed the XMP profile, my system turned on on black screen (no image), then shutdown itself, then turn on again with no image, and so on... It did it like five or six times then I did a hard shutdown and took off the bios battery.

Also, my XMP profile seems to be missing a few parameters

is that normal?.

​

 

That doesn't look right, XMP profile should have all rows pre populated with predetermined values.
Can you try rebooting and checking if it still persist?
Btw, What ram are those?

You can also try this, I'm assuming you are trying for 3000Mhz.
Set the profile to manual, then set primary timings to CL16, tRCD to 18, tRAS=38, tWTR=7 zero out all the other timings set multiplier to 100x15 with voltage at 1.25v. Let it train for 10 minutes, if it doesn't boot, reset the bios and try with 1.3v

 

Actually the XMP profiles has been always been that way, missing most of the parameters. I'm not sure if my modules are corrupted or my system BIOS is failing on reading the little memory inside the modules that contain the XMP parameters table.

My modules are the 8GB Crucial Ballistix Sport LT DDR4



I'm trying actually to achieve 2666 MHz as the modules are rated at that speed, but they are running just at 2400 MHz. Do you think I need to raise the voltage to get 2666 MHz stable? CPU-Z seems to read the table for the XMP profile at some extend and it says the voltage is 1.200



By the way where is the memory multiplier located? I couldn't find it. Maybe I have to manually set the "XMP" parameters, but I don't know what parameters to use

 

You should be able to run 2666 at 1.2v

Advanced > OverClocking > Memory > Memory Profile > Custom Profile > Memory reference clock to "133" and Memory ratio to "10"

Then set primary timings to:
tCL "16"
tRCD/tRP "18"
tRAS "38"
tWTR "7"
Voltage "1.2v"

Zero out all the other timings or set them to "Auto".

 

Well I did what you said. Used those values on the system BIOS, the laptop boots just fine zero issues. How can I check stability? Running memtest86 would do the job?

 

Glad that it worked. You might want to tighten up the auto values up if you care about performance, or even push the frequency higher. As for stability, really depends on how stable you need it to be. You can run something like MemTest or memtest86.

 

Given the small benefits in performance I would tune towards stability

 

Well I've been using the laptop the whole day and haven't had any BSOD or hangs. I haven't played though, just ran a few tests of Batman AK. I just don't want to have any BSOD I hate those. I guess as the service manual says the laptop supports 2667 MHz RAM that speed shouldn't be as hard to keep it 24/7. Do you see any real world difference going from 2667 MHz to 3000 MHz on RAM?

 

Maybe it's the RAM instability at 3000MHz?
I am running 2 dimms in p750dm2 with 8086k at 3000 (XMP) and no issues at all, even temps aren't getting high on RAMs.

edit: check the chips installed on dimms, if these really have correct timings set to whats in the memory spec sheets, if you can find specs for them..

 

The short answer is it depends.

Longer answer, if you mostly use the laptop for gaming, I'd lookup some benchmarks for the games you play comparing 2667Mhz ram vs higher clocked ones. Some games will see greater benefits than others, and you can judge whether it would be worth it to spend time on pushing your ram further or not.

 

It's the highly threaded games that load all the cores that tend to benefit.

 

Wow I watched a couple of videos on YT on Batman Arkham Knight and Assassin's Creed Origin @1080p and you get a solid 10 fps+ with 3000 MHz over 2133 MHz, that's a lot IMO. How far you think I can push my 2666 MHz modules for a 24/7 stable use????

 

It depends, my old 2666cl18 gskill modules from 11/2015 do 2800cl17 @ 1.3vdimm, but I hate ram OC and have never ever found a good flowchart for tweaking secondaries so there may be a limiter there

 

Every kit can be so different, but the primary timings are the ones to focus on more.

 

My approach to ram overclocking,

1. Set very lose primary timings, ie. 21-25-25 and AUTO or 0 for the rest.
2. Find max frequency with reasonable voltage. ie. 1.35v
3. Begin to tighten up primary timings one by one, ie. cl 19 to cl 18cl etc. then tRCD and tRP.
4. Slowly bring down the secondary timings.
5. Slowly improve tertiary timings.

I use Buildzoid's approach to quickly test all the ram up to 5% in MemTest, whilst tightening up the timings.
Majority of the time, If the timings are too tight, your memory will begin puking errors before 5%.
Then you proceed to either loosen up or tighten the timing further, testing it and redoing the previous step to find the lowest stable setting.

It's this rinse and repeat process until you have gone through all the timings you wanted to improve. Last step would be running MemTest to 400% or more for stability.

 

I’ve gotta get my CPU heat in check. I’m having a hard time getting voltage to stick at a fixed amount. So I’ve been running default adaptive voltage with a -0.175 undervolt at 5Ghz on all 6 cores” seems extremely stable”. But, Battlefield V at 1080P 144hz has proven to be extremely demanding. And pushes temps to 95C at times lol. Jeez what a game.

I may need to check this CPU’s TIM out. Maybe reapply some LM and do the toothpick or paperclip mod as well.

If I run LM on my GPU with the paper clip mod I should be fine right? Or would I still need a shim?

I’m going to order Krapton tape, Conductonaut, and standard thermal grizzly, and run some foam like other users have in the past.

I seriously need to get these temps in check.

 

BF is really cpu intensive and gpu also heat up to the top. I do not think the 750tm heatsink can handle 5ghz 8086k and full power 1080. Have you tried to cap the 1080 at 150W (0,9V) and also come down with the cpu a little bit? To be honest 5ghz looks great in the signature, but does not give the extra few hundred mhz to the fps ingame.

 

I popped my keyboard off 3 times today lol. I’m sure you guys already know what that means. Memory overclocking didn’t go so well. I think I’ll buy a better kit instead.

my memory runs at 2666mhz cas 19 1.2v 2t default

I couldn’t even get 1T with 1.35v.

2800mhz was crashing after boot up.

 

This laptop has been running at 5Ghz ever since I first turned it on directly out of the box it was mailed in. BFV has been the only game to really push cpu temps like this.

 

Also 5Ghz is not that unrealistic with a 8086K it is already delidded. Plenty of 8086K’s run 5Ghz with 1.3v. Most 8700K’s required about 1.45v for this kind of frequency. And plenty of users are running 4.7Ghz on a 8700K in a laptop with more voltage than I am running now.

these chips run 5Ghz like it is nothing. After some fine tuning with thermal material and adjustments I’m 100% solid.

Also, even in BFV at 5Ghz in my current state it still doesn’t even throttle.

The GTX1080 is another story. I need to use LM for sure. And paper clip mod. I will easily see a 20C reduction in temps.

We can easily run these components to there fullest. It just requires some work. I’m going the foam dam route with LM for my 1080 for sure either way though.

 

For Bf5, up the graphics detail to shift the load onto the gpu or, do a -2 AVX offset, as it uses that instruction set which hammers the cpu.
And I am relatively certain that 1T command rate does not work on these laptops.

Lastly, talking about the 8700k/8086k voltages, it all depends on the bin. On average the 8086k has a better vol/freq curve than the run of the mill 8700k. However, there are 8700k's on here that do 5.3Ghz all core due to superior volt/freq curve. Those chips can run at ridiculously low voltages for 5Ghz and below, effectively eclipsing the majority of 8086k's.

So it comes down to how well binned your 8086k is, as it's still a binned 8700k rated for 1-core 5Ghz boost.

 

I’m running 1080P ultra everything with AA on BFV. My frame rates are pretty good at around 95-120 give or take. I may increase the in game resolution slider a little to push more GPU activity while still having some room to enjoy the high refresh rate, instead of the -2 AVX offset. For some reason when I apply the -2 AVX offset which I tried earlier today around lunch time it immediately set my CPU to 4.8Ghz all the time in any application?

I was considering delidding again but, I think my CPU’s delid is actually pretty good. I looked through HW info, and all 6 cores are within 2C of each other. Testing thermals with the CPU 100% stock speeds results in temps in the high 50’s at 4.3Ghz/4.3Ghz Cache.

but yea it is unbelievable how 5Ghz is daily doable on a laptop. These 8086K’s really do go far on very little voltage.

So I think a better thermal paste, and LM on GPU, with more heatsink pressure will help a lot.

 

-2 AVX should make it run 4,8Ghz for application using that instruction set.

If you do end up taking it apart for inspection, make sure to check how the contact between ihs and heatsink is. I always have the ihs on the heatsink and check if there are any gaps by shinning a light on the opposite side. This is where lapping can make a difference if you use LM for all the interfaces.