*** The Official MSI GT75 Owners and Discussions Lounge ***

There is no XMP profile in the BIOS bro, it's just Default Profile or Custom

But now it's all clear, from the MSI Spec sheet:

DDR4-2400 speed up to 4 slots
DDR4-2666 speed up to 2 slots, 4 Slots, Max 64GB

So it's running at 2400MHz. because I have 64GB of RAM

 

@Mr. Fox From CPU-Z:

 

Oh uh.... :/

Phoenix before you start playing deadlift lifting trying to lift the laptop up with 3 fingers, please at least fix that MSI CANCER BOOST MSI VID BOOST.

Remember let's do this by the book.

CPU VR Current Limit=800.
IA AC DC Loadline=1 (CPU VR Settings->Core IA Domain).

Run cinebench r15, post your voltages VID range and papusan mad at you for buying a castrated 8700k i9 8950hk

Then go back in your bios, set DC Loadline Enable and value to 0 uOhms in Overclocking Performance Menu->Platform voltage overrides.
Post another cinebench. If you are unstable after THIS last change, then this setting may be MSI's desktop version of LLC "Loadline calibration", in which case, change it back to 2100 uOhm and "Disabled".

I'm starting to think I was right.
@Vistar Shook
I suspected MSI was using internal loadline calibration ON TOP of IA AC DC Loadline=1.80 mOhms (180).

Because, for example, if you change IA AC loadline to 1, but change IA DC loadline to 0, you will notice a LOT of "vid droop" at full load, however the temps don't seem to correspond to the VID droop at all (its like 2C hotter than DC=1), so it seems the CPU is reading "Voltage droop" but the voltage droop is getting 'reversed internally' (this was my own experiments on the GT73VR).

Yeah.

I'm buying a Clevo. But I may not get to buy it right away because something else came up rather important.....

 

This is what you tested with. See if you can put as seen in Cpu-z

 

Phoenix.....the voltages, please.


And yeah.
You can run your RAM At 2666 mhz just like I ran my 8x4 GB at 2666 mhz
But you need to use timings of 19/19/43 instead of 17/17/39. But this isn't worth the latency decrease from looser timings
So the auto settings are unstable.

My own Kingston valueram is stable at 15/15/15/35, 1T, @ 2400 mhz.
But 2666 mhz requires 19/19/43 (actually 18/19/43 worked).
Much slower than 15/15//35.

 

Yikes! And, the default timings are as sloppy and loose as a 60-year old brothel owner, LOL. CL19 at 2666 is pretty horrible. Maybe you need to invest in some quality enthusiast RAM and sell that to someone with a Dell laptop. Even CL17 is sad for 2666.

http://www.memory4less.com/samsung-16gb-sodimm-pc21300-m471a2k43cb1-ctd

 

Skype ID is the same as this I'm on right now.

 

Yeah I'm not buying a Gt75 now, no matter what. Period. Not after seeing this......
And 1.280v required (once you remove MSI VID BOOST) to boot windows at 4.3 ghz on a "badly binned 8700K"???

 

See my post above. He need to test whats stable.

 

I'll add you shortly

@Vistar Shook suggested I change tCL and tRCD/tRP at 18 from the default 17 and I got 250 extra 3DMark points

 

I'm surprised you would even entertain the idea before seeing this. I would have thought the utter nonsense you have had put up with on your current turdbook would have been enough to cure you once and for all.

@Phoenix - Even CL17 is sad for 2666. You should not be running CAS that high until you get up into the 4000MHz range. 2666 should be stable around CL14 or CL15 at the most. My crappy Kingston HyperX sticks in Machete are 2400 CL14 and I have them running 3000 at CL16.

 

Yeah cas latency should be 18 for 2666 mhz probably. Should not BSOD.
I need cas 18, then subtimings 19 and 43, for 2666 mhz on Kingston Valueram. I only tested 2666 mhz at 18-19-43 timings briefly because while the WRITE speeds were better in AIDA64, the read speeds and latency were horrendous, compared to 2400 mhz with 15/15/35 1T.

Looks like the samsung ram you got is no better at all than the old Kingston stuff :/

 

The problem is MSI!!! I tried to help out last year. And who know how picky i9 BGA (scam) is. + MSI.

 

I never was intending to buy it. I just wanted a REVIEW Unit to see what it was capable of.
never, ever , ever, ever was i going to BUY it !

 

For comparison, the G.Skill Ripjaws that came on mine are CL18 2666MHz stock, but they run at CL14, at 3000MHz CL16 and at 3200MHz CL18. But I guess ajusting the timings won't help if the motherboard doesn't allow 2666MHz when using all 4 slots, bummer.

 

MSI and their competitors know exactly what they're doing, and things are not going to get better. They're only getting better at crafting lies and marketing hype... and, it's working for them. People keep buying garbage.

 

Set the bios back to factory first. Boot into windows, shut down. Boot windows and do all os repairs then restart again.
Go into the bios and set everything in red to 0 and set 18 to auto if it will not go to zero. Set Memory ratio to 10. Don't worry about command Rate


And let it attempt to boot on it's own. What this does is it goes through all the settings that have been programmed in in or to get it to boot. It usually finds one easier this way.
If nothing works say after 10 to 20 minutes, then up voltage to 1.25V. usually you need to add uncore voltage with more ram so hopefully that will raise automatically as well. (provided you don't have access - and if you do, that it actually works)

This is only to see if you can get the ram to boot and be fully stable on it's own. Get past this part, then you would record the numbers and start the tinkering from there.

 

@Phoenix bro I hope you and @Falkentyne and @Papusan find some solution to fix the ram.
Atm I feel really disapointed for you about this...
Other question to understand it better did you get a repast for your GT75 and it hits 95 degrees with oc and cooler boost activated?

 

@Papusan @Mr. Fox

This is with the Falkentyne magic:

 

I'm interested in temps. And several rounds with stock clocks.

 

Nice improvement. What are the maximum core temps?

 

@Mr. Fox @JeanLegi @Vistar Shook

The temps are superb now with max fans on

picture uploader

 

Temps were 67C with coolerboost at max @100% fan speed AND with IA AC DC Loadline=1, I think.

But with the default IA AC DC=auto, it gets 1.4v through the chip instead of 1.28v through the chip.(voltage override is set to 1280mv).

 

Yeah, that is a big improvement. The 8700K in my Clevo uses around 1.130V to 1.160V to pass Cinebench at 47*6 with AC DC load line set to 100 (100 = 1/100th or 0.01 in Clevo BIOS).

It seems that 1.400V is stock voltage for BIOS defaults with 8700K as well, which is ludicrous.

One would think the "experts" at MSI would have this figured out by now and ship products with BIOS defaults that actually work half decent, but nobody does the right thing anymore. Their jobs are to sell stuff, not make things that actually function correctly off the shelf.

 

This laptop should NOT, I repeat, should NOT be shipped with the MSI stock BIOS settings otherwise it would go into flames in no time. @Falkentyne magic decreased the temps by 15C. This guy is a genius.

 

Oh man...I will definitely need your guy's help when I get mine!

 

Excellent advice, as usual.

 

Wow much better....I suppose Phoenix 's OS tweaking sessions will now have to include Falkentyne's BIOS tweaks as well.

 

Great work guys, happy to see the 8950HK tamed and working so well. Looking forward to seeing the settings and further tuning, with non-100% fans and a custom fan curve to make it more liveable as a daily driver. Amazing cooling on that GT75 8950HK @ 106w!!

You outta send those updated tuning settings in to MSI and suggest they ship it that way out of the box. With those tunings even the GT83 8950HK might be importable now.

 

The trouble is not so much the need for tweaking as it is the target market is not the kind of customer that actually knows how to go about correcting the problems. The fact that it ships with such screwed up BIOS defaults is going to be a very frustrating situation for the average buyer.

 

No kidding, some of the settings, i.e. IA DC loadline are only available to the user if the Bios is unlocked.

Enviado de meu Pixel 2 usando o Tapatalk

 

The chip tamed is actually working like a badly binned 8700K now.
temps are reasonable. But there is no such thing as as an 8700K that requires 1.280v to load windows at 4.3 ghz. Most can load windows at 1.1v at this setting.
Binning is real, folks.

The i9 is nothing more than an 8700K that failed Intel's binning QA for LGA.
but tbh I have never seen a MSIbook ever have a max VID of 1.4v at load at fully automatic settings. 6820HK never had that. 7820HK never had that.
The highest default VID i ever saw I think was 1.3v.

But of course with Intels terminal velocity boost, they probably created more VID steps. This can make sense....if a chip required 1.25v VID For 4.3 ghz (minimum voltage required). But MSI should not be boosting this to 1.4v.

Basically here is is what MSI is doing:

1) The default value of "IA AC DC loadline" is 179 (1.79 mohms).
this boosts the VID (Voltage) signal by 1.79 mOhms of resistance at full load--so the higher the current draw, the higher the voltage is boosted.
All MSI laptops with skylake, kabylake and onward are doing this exactly. All of them.
This is why hmscott's undervolt suggestion is required.
But it isn't Intel doing this. It's the Bios. It's MSI.

TBH it's technically SUPPOSED to do this by design with adaptive voltages. BUT THIS SETTING IS SUPPOSED TO BE IGNORED when using manual voltages. On desktop 8700K motherboards,
If the IA AC DC loadline setting is left at "Auto", the 'VID" on an 8700K (with 1.30v static voltage) will show up as 1.45v at full load, but the *VCORE* sensor will read 1.30v.
So in this case, the IA AC DC is only making the VID get over-reported, while the vcore itself is unaffected.

( @Vistar Shook 's evGA laptop functions like this, which is proper behavior).

However, on the MSI books (and at least some alienwares), if you use a static voltage, the IA AC DC loadline AUTO setting STILL boosts the STATIC voltage by 1.79 (1.80) mOhms of resistance at full load! When it should NOT affect STATIC voltages. if you set 1.25v, you should get 1.25v, not 1.4v. This is a *PROBLEM*. And the only way to fix this is by unlocking bios menus and setting IA AC DC loadline to 1.

I mean think about this.
Does it make sense to 1)
use static voltage of 1.25v.
Then reduce the voltage by -100mv "undervolt" so now you're using a static voltage of 1.15v, to COMPENSATE for the VID BOOST that was boosting 1.25v to 1.4v?

This defies logic.


The SECOND PROBLEM is that MSI is using an INTERNAL setting for "Loadline Calibration" (LLC)--that desktop setting that all of you are familiar with for reducing 'voltage droop'.
We have NO access to this setting whatsoever. But it's used by MSI's voltage regulators/Bios and is hardwired

I found this out by doing something bizarre:

I set IA AC loadline to 1
then I set IA DC loadline to 0 (Auto), which is (what I thought before is or should have been 2.1 mOhms, but is 1.8 mOhms).
Then set my static voltage of 1.178v.

first, temps were about 2C higher than before (no big deal. slightly more power draw). I verified this by looking in the Embedded Controller RAM And saw the value in EC RAM register C9 was about +3 higher than before. (This register shows combined CPU+GPU amps draw).

The second thing I noticed was that at FULL LOAD, the "VID" was DROPPING from 1.178v to 1.10v !!!
However the temps and power draw did not correspond to this drop. So the "real" voltage was still 1.178v (maybe slightly higher like 1.182v). But the VID was showing substantial vdroop.
The fact that the VID is dropping like this at full load shows that there is or "should" be voltage droop present, but in fact there isn't. That can only mean that MSI is applying some sort of loadline calibration internally.

 

To the topic with the fan curve. Gigabyte has a nice curve with 10 or more points to set a real nice curve for your own use. And most Importen they show degree in numbers.

 

That's how *all* stock BIOS's go out, with *hot* CPU voltage settings, undervolting -100mV is enough for most to be tamed and within the skill level of most enthusiast owners, but not all.

There really needs to be an auto-tune done in the BIOS for best voltage / performance, there's really no excuse for not having this in 2018.

With such a simple tuning as undervolt, 99% of well designed laptops are AOK out of the box, with no thermal throttling.

The rest of the tuning to extend resources for better performance should be a matter of pride of design and delivery, shipping with throttled power has no excuse with such excellent cooling with a fully unlocked HK CPU.

But, then again, I see Clevo's getting tuning out of the box, even with Prema BIOS there are additional tunings needed.

Nobody ships with the tuning done out of the box, yet.

 

Just imagine a n00b spends all this money on this taptop because of all the nice advertisements only to see smoke coming out of it once he pushes it. Mind you, it was horrible despite my having Thermal Grizzly Conductonaut, Fujipoly pads, and the bottom panel mod. Crazy

 

It's not just the Noob's, HID relayed that MSI USA themselves thought it ran hot, which is the excuse MSI USA gave for not importing the GT83 with the 8950HK.

MSI Global / MSI USA BIOS / tuning Engineers are letting down their companies badly.

Maybe @Donald@HIDevolution can relay this out of the box vs tuned results and tunings to MSI USA so they can "fix" their out of the box problem and ship an amazing laptop with the amazing performance it deserves to deliver.

 

All MSI has to do to fix this BS is simple.

FORCE IA AC DC LOADLINE TO 1 INSTEAD OF AUTO.
that's all!!!!!!!!

then if the user wants to overclock he can apply a positive voltage offset. PROBLEM SOLVED.
The only problem with that idea is.....it makes sense?

The reason why this setting is .....overworking, is because MSI is also using internal loadline calibration to stop voltage droop. The problem is there is no vcore sensor on these laptops. Only the newer Clevos have vcore sensors. VID is not voltage. It's rather what the CPU *THINKS* it's getting. That's why changing IA AC loadline to 1 and IA DC loadline to 0 makes the VID start reporting vdroop (and a lot of it). The IA AC DC loadline setting was ORIGINALLY DESIGNED FOR DESKTOPS WITH "Auto" loadline calibration (meaning: Vdroop). So the VID gets boosted enough to COUNTER the effects of voltage droop. And that's it.

If MSI were doing this, Phoenix would have a VID of 1.4v, but cool temps with cooler boost (70C) and expected temps with auto fans (84C). The VID would show 1.41v (true) but vdroop would DROP this down to about 1.3v at full load.

But MSI is using internal loadline calibration to remove vdroop IN COMBINATION WITH VID BOOST at the same freaking time. So now you have extreme Vrise of true voltage, not just VID.....

this is bad.

Having a medium unchangeable default loadline calibration is a good thing (most people on desktops use medium levels of Loadline Calibration (LLC)). Using LLC in combination with IA AC DC loadline causing VID boost boosting up the already "boosted" voltages is bad.

 

bottom panel ventilation mod.

What's that?

 

@Mr. Fox @Papusan @Falkentyne @Vistar Shook @MBN @Dr. AMK

3x 960 Pro's in RAID 0

 

@Falkentyne

How can the screen of the GT75 be 3ms but the GT73 is 5ms when they are the exact same model? my screen was manufactured in 2016! Go figure! The stuff the MSI smokes seems to be really good!

 

So I did a test with memory timings 2400 vs 2666 Mhz and here are the results:

2400 Mhz:



2666 MHz:

 

Specifications show a range, or at least 2 values for that model, and the model numbers have sub numbers sometimes not showing up if the tool doesn't have the details on a specific version. It could also be down to tuneables in the firmware, choosing which spec to use. It may have some out with the goal of 3ms, but didn't have time to validate before shipping, so now same panel pushed will do better in current shipping units:

N173HHE-G32
Response 1.5/3.5 (Typ.)(Tr/Td) ms
http://www.panelook.com/N173HHE-G32_Innolux_17.3_LCM_overview_28924.html

 

Guys im really new here and i wanted to know if i shud go with the GT75 TITAN 8RG with i9 and a gtx 1080. I am willing to spend that much for it as it's for my work purpose i do content creation and lots of heavy load works.
Just wanted to know if there are any problems with it as i don't really wanna give my laptop for servicing for small problems as my work would stay pending
Please help

 

go on a pure white background with black text and scroll the screen up and down vertically.
if you see green shift in the text, its definitely the older panel.

But yeah.
and nice bench score improvements!!!

 

No green shift bro

 

Ugh i meant, green shift on blue/light blue text, sorry, not on black.
basically blue turns light blue and light blue turns greenish?

(I'm with stupid).

 

See, the problem is MSI, Clevo, ASUS, HP, Acer, Dell and all the rest just do what Intel says they should as a reference for stock. That means a starting point where every machine, regardless of CPU model or bin quality, should POST without problems using identical BIOS defaults, but it was never meant to be optimal. When it doesn't work correctly they shrug it off and say it functions as intended even when malfunction is the end result, because they don't care. Caring is not worked into the budget. It takes time to care, and time is money. None of the ODMs have the intelligence, desire or decency to correct mistakes and do what is right for their products and best for consumers. Consumers want to spend as little as possible and expect everything to be peachy.

Then the ODMs lock down the firmware, make it signature-protected cancer, hide the menus and cut the gonads off so that nobody can fix all of the garbage they screwed up. Then, for the ultimate insult, they slap the illegal "warranty void if removed" stickers on them to scare the noobs into using their Über-Gamer toy in a perpetually broken state.

Hmmm... where have I seen that before? Oh, yeah. It was that crusty old guy with the Tornado F5 that did it first.

Some things are just too good to not keep doing them.

Or, it could be really bad stuff, depending on perspective. If they believe their own lies and don't know they are selling broken crap, then it's really bad stuff. If they know what they're doing, then they're evil rather than stupid and the joke is on the consumer. But, I'm not really sure when "good" actually enters the picture with either scenario.

 

Hi, can some1 explain where i can find "Power and Performance, ->CPU VR Settings, Core IA AC DC Loadline" setting in bios...

 

Right now you'll have to contact SVET on the MSI official forums to have him unlock the Bios to access those menus, and that requires a $20 donation (it takes a VERY long time to unlock menus and he unlocks everything and writes his own automatic flasher program to flash it with a few instructions and prompts).

You can also unlock the bios menus yourself by reading the backup with FPTW64.exe, making a backup then unlocking menus, setting bios lock to disabled with RU.exe, then reflashing the capsule. Svet basically automates this but you CAN do this yourself (MSI has bios guard disabled (BIOS GUARD IS NOT BIOS LOCK! Bios lock prevents windows changes to writing to the bios, Bios guard is basically DRM meaning GG WP UR SCREWED).

http://forum.notebookreview.com/thr...-clear-cmos-and-prevent-common-issues.812372/

There's another way that involves no flashing, but I need to see if EVERY user who buys these laptops winds up overheating in basic apps or not, and I'll have to talk to a few people for permission first.