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

But if i have any problems will i get services in india through MSI

 

To see all of the MSI Service Centers, please go to https://in.msi.com/page/service-location-new and select Asia.

If you get any upgrades, you should look to HIDevolution for warranty service. HIDevolution.com Global warranties cover 2 instances per year of 2 way international shipping (by FedEx, DHL, or USPS/EMS - at HIDevolution's discretion) should the system and/or parts need to be sent back to the US for warranty related repair or replacement.

Please email me any other questions you have about HIDevolution and our policies...this thread is mostly to talk about the laptop, and we don't want to

 

Sigh... Niche markets never get what they want...

 

Greetings,

I got my new one yesterday. Haven't had much time with it but I did get some stuff. I have another data point in the crazy voltage/temp unpleasantness w/these things. Attached two pics; idle and another just after the cinebench run.

Also, I was following this -> http://forum.notebookreview.com/thr...-clear-cmos-and-prevent-common-issues.812372/

And all is good up to the point where I need to dump my bios. I grabbed that linked ver of fptw. When I run it I get (admin shell):

Code:

C:\tmp\FPT11\WINDOWS64>.\FPTW64.exe -D unlocked.rom -BIOS

Intel (R) Flash Programming Tool. Version:  11.05.15.1000
Copyright (c) 2007 - 2016, Intel Corporation. All rights reserved.


Error 367: [FPTw64.exe] cannot be run on the current platform.
Please contact your vendor.

I have googled around and only found unresolved instances of Error 201 and that message. Anyone know what 367 is and/or how to get beyond this? Assuming it's windows getting in the way, is there a tool I can use from dos that would work better?

Thanks!

PS On a side note, linux is kind of a fail atm on this thing. Quick and dirty attempt w/Ubuntu 17.10, and 18.04, don't do squat w/out disabling acpi. And w/that the trackpad doesn't work/isn't seen, bt is seen but not wifi, wired aquantia nic does show up, the 1080 is seen, I didn't see the igpu, but I'm stuck at 800x600, can't see nvme drives due to the controller being in raid mode. I thought or sure this was a resolved issue by now. <sadness>

 

Try version 12 and see if that works.

P.s. all known versions are here:
https://www.win-raid.com/t596f39-Intel-Management-Engine-Drivers-Firmware-amp-System-Tools.html

Let me know if it does.
I know for a fact that v12 doesn't work on a GT73VR. It starts whining that it needs cannon lake...

http://www.mediafire.com/file/b7g58727c58osdp/

 

Confirmed to work! I now have a dumped image. Thank you very much!

Using AMIBCP 5.02.0023 (also from that forum OP), I'm a bit confused as _everything_ seems to show up as "YES" under the "Show" column. There are options there that I know aren't visible in the BIOS. What am I missing there?

I found the ac/dc loadline entries and have modded them to 1. Just removed Dragon Center and about to reboot and then will flash the modded bios.

 

Don't change the Yes/No things.
Just change all options you want to "Supervisor".

BTW did you change bios lock?
you won't be to flash anything without that.
you need to use RU.exe as instructed in the GT73VR mod thread to get the hex ID for Bios Lock. You may be able to 'search' for it in RU after you find the SETUP field (there are two, the proper one is the larger size one), but it's far better if you just make a 'backup' of your backup (for safety), run UEFITOOL on it, search "text" for Bios Lock, then at the bottom double click the field it finds. e.g. doubleclick where it says "UNICODE BIOS LOCK FOUND AT...."

Then above, click the PE32 thing that gets highlighted, then right click and choose "Extract as is..."

Then run UEFITOOL on that file, and save it as a txt file. then you can search that text file for the Bios Lock text and get the HEX ID key for it.

You'll need that for RU.exe.

 

BTW you know you can unlock other menus to appear in the bios settings by setting the menus, option and SUBMENUS Under it to "Supervisor" right?

 

I did all of that and I'm still denied wrt being able to flash the thing. See screenshots.

Note that I do have an unmodded backup of unlocked.rom.

Code:

C:\tmp\Intel CSME System Tools v12 r2\Flash Programming Tool\WIN64>.\FPTW64.exe -F unlocked.rom -BIOS

Intel (R) Flash Programming Tool.  Version:   12.0.0.1062
Copyright (c) 2007 - 2018, Intel Corporation. All rights reserved.

Reading HSFSTS register... Flash Descriptor: Valid

    --- Flash Devices Found ---
    MX25L12875F    ID:0xC22018    Size: 16384KB (131072Kb)

Error 301: Protected Range Registers are currently set by BIOS, preventing flash access.
Please contact the target system BIOS vendor for an option to disable Protected Range Registers.
FPT Operation Failed.

Please advise.

 

I did not. I guess that's what I was conflating w/the 'Show' column. Thank you for the clarity. If I can get beyond the current issue I'll certainly give that a whirl as well.
Thanks!

 

As I said, you need to disable Bios Lock :/
read.

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

 

However there is a MUCH MUCH MUCH faster way to get the Bios Lock key without doing ANY of that "GUID" stuff that Paloesco mentions.
The only thing you need from paloesco's post is UEFItool, IFR extractor and RU and the rufus key to make a usb flash drive with RU.exe/RU.EFI installed.

Here is the easy way to get bios lock.

1) open your bios backup you made with fptw64.exe (make an extra copy for safety) with UEFITOOL.
Search text for bios lock Double click the result at the bottom.
highlight the part above that appears. double click it and look for PE32 whatever as I said.
right click that part and choose extract as is. Name it whatever you want.

2) open what you just extracted with universal IFR Extractor.
Then just save it as a txt file.

3) open the text file, look for BIOS LOCK. Note the hex key to the left of the field.

Example:

0x40A37 Setting: BIOS Lock, Variable: 0x8AA {05 91 01 08 02 08 84 07 01 00 AA 08 10 10 00 01 00}

The 0x40A37 is the important field.

Then you can follow paloesco's instructions in RU and use that field or open the proper SETUP field and look for bios lock that way (tbh you don't even "need" the hex field but then u have to do it the long way through the list )

Once you set Bios lock to 0 in RU, choose save and exit and it should reboot to windows.
Then you can flash your modded capsule.

P.S. i take no responsibility if you bricked something. You DO have a SPI programmer available, don't you? (the bios chip is easily accessible with a Pomona 5250 clip, SKypro programmer and jumper cables, no soldering required).

 

I never used the hex grid method.
Is there a way to find the Bios Lock VAR through the list and access it that way?
You have the hex offset to the very left.
I can't help with what field is what though. I never had to bother with that.
Try to follow what is in Paloesco's thread please.

 

That's what I did/what that is. I uploaded 2 images in that post. One showing the setup.ffs IFR extraction text file at the point where I found the BIOS Lock text showing 0xA12 as my offset and the other image of in RU at that offset showing that it is now 00, or supposedly unlocked.

 

@PrimalNaCl
Any updates?


Try to search for "BIOS LOCK" in RU to see if you can actually find the text itself.

I looked for my boot disk with RU but i cant find the flash drive!
so i cant help you right now.

I'm sorry.

Did you try changing both 01's to 00?

 

But when you boot to windows, FPTW64 will NOT FLASH?
I just verified that it DID flash for another user in a PM. And it was successful.
I can't do anything. I do not have my flash drive. I can't find it. And if I can't find it, i can't boot RU....but it SHOULD be working ! There is ZERO reason for it NOT to work.

I know you have SECURE BOOT DISABLED.
did you try rebooting the computer back to RU and make sure that it is still set to 00?
Im sorry i cant be of more help to you. But I'm not about to format my other USB drive for this. Sorry. But disabling bios lock worked for another user.

You HAVE to be at the wrong variable or something~_~ Try changing both 01's there.

are you running an ELEVATED command prompt as administrator?

 

With the HIDevolution bottom panel mod, the target goal is to lower the temps of the CPU + GPU but it also lowers overall system temps, it can vary anywhere from 3c-5c.

I have yet to try it on my own with smaller holes... then bigger holes. Then test to see if any difference in temps. I'll try it and let you know.

 

hey guys..here's a new review of the GT75 8RG

 

I know, right? And, the smart folks in the niche market will never accept or embrace something they don't want. Compromise is for compromisers, and that always turns out sucky in the end. Besides demonstrating weakness and lack of resolve, compromise is almost always a guarantee that you'll continue never getting what you want. Best to reject the filth, wash your hands, shake the dirt off your feet, close the door on your way out and never look back. It's not hard to do that when you pause and consider that they have done nothing to be worthy of your patronage. Giving them your money just reinforces that, at the end of the day, you are willing to tolerate their crap and don't actually mean what you say or abide by principles.

 

Correct.

No argument here. I have followed the instructions pretty religiously. My only deviations are for obvious deltas due to model and tool changes due to that.

Several times.

I appreciate the help you have given. Even though I have had no success, at least it has served to confirm I'm not doing some bonehead mistake in the process.

Wrt the other user was it for the same machine and same var/offset? (0xA12) Based on the extraction from both the downloaded BIOS from MSI and from a dumped image of my machine showing the same offset it apparently isn't _machine_ specific. Model-specific but not machine-specific. So if that's the case, if that user's model is == to mine, the offset should be ==. If that wasn't the offset that user used, then there _IS_ some issue w/how I have arrived at the same offset 3x now.

So it would seem. But I have gone through the process several times. 2x for the original (shipped) BIOS, and once just a few minutes ago for the updated version off of MSI's page. I flashed it, did the EC clearing and whatnot, set time/date, disabled secure boot, and booted into RU to check that address and it was back to 01. I exited (changed nothing), booted into windows, dumped the BIOS, and followed the procedure for finding the setup ffs, extracted it, and searched that text file. Once again it showed up as 0xA12.

I booted back into RU went to that address, changed the 01 to 00, wrote it (CTRL-W), shutdown the machine, powered it back on, went back into RU, and verified that that address was still 00. Then booted back into windows, admin shell and the attached screenshot is the result of the flash attempt. (Same error 301 issue.)

Indeed. Attached screenshot for confirmation.

How much of a pita is it going to be to get to the BIOS chip so I can flash it manually? Any idea where it is located on the mainboard in this model? Dare I even _hope_ for it to be on the bottom side? :/

 

@PrimalNaCl Can you PM me, please?

 

Just a heads up.
You're at the wrong location.

You should be at 0xA12 (0A12).
You're at 00A1

That's completely wrong.
You need to go down much farther.





^^ wrong location!

this is 0x00A0, not 0xA12 (the first bytes are not important).
0xA12 should also show as 0x0A12

 

Hello
What samsung ssd 970 evo/pro with phoenix controller can run 3500 r/d on 7820hk ? Or must have 8th gen ?

Sent from my CPH1723 using Tapatalk

 

Indeed. Thank you very much for helping track down my issue via PM. Thank you _very_ much!

The newer RU works much better.

No more stupid high voltage through the CPU.

Sent from my Pixel 2 XL using Tapatalk

 

On another note, I picked up a Eurocom 780W brick to replace that dual 230W Frankenstein abomination MSI puts in the box.

From the Eurocom indicator wrt output I haven't seen it go above 267W. That's at full brightness, 4k screen, Firestrike Ultimate. And it was only for a brief period of time. Seems a single 330W would have been fine.

In any case, I do like the Eurocom very much. I saw Mr.Fox's yt review. When I ordered I mentioned him. Hopefully that will help w/further consideration for seeding of review units.

Thanks!

Sent from my Pixel 2 XL using Tapatalk

 

Glad to hear you made it through patching your BIOS

Would you mind testing that firmware change temperature drop at full load vs. a simple undervolt of -100mV? Or whatever your CPU will allow for undervolt.

Since you have an unlocked BIOS you can set the 'minus sign' in the BIOS for CPU Voltage offset and enter -100mV directly. Otherwise normal users (non-modded BIOS) will use XTU or TS for that same undervolt.

Reports of 10c less temperature reading at load for the BIOS change, is the same as a -100mV undervolt reports give, 10c drop in 100% load temperature.

Dropping from thermal throttling 10c, either with BIOS mod or undervolt removes the need to re-paste.

You'll want to undo the BIOS change to test the undervolt, you can't do both at the same time.

It's for science! And, to save others from thinking they need to mod their BIOS to get the same 10c temperature drop that a simple -100mV undervolt will also provide.

Thank you!

 

Can you please run PRIME95 blend (disable AVX and FMA3 first, by editing local.txt and add these lines:
CPUSupportsAVX=0
CPUSupportsFMA3=0

to the top.
Then while prime95 is running, run VALLEY benchmark (or Heaven benchmark).

Have Throttlestop 8.50 or 8.60 open at the same time. Open the limit reasons checkbox.

Please tell me if you are getting any power limit throttling.

The MSI can detect whether the "dual" combo PSU adapter is being used or only a single 230W PSU.
Sirgeorge tested this on his GT73VR SLI, by unplugging one of the two 230W PSU's.

The embedded controller then throttled the CPU once 230W total AC power draw was exceeded.

People have not been successful bypassing MSI's cancer power restrictions. I do not know if the EC is smart enough to detect that the Eurocom PSU is infact able to pull more than 230W, but this was an expensive gamble to take. Clevos with Prema bios can pull all the power required. But MSI.......

Please try the test.

 

Ok. With prime95 and Valley running it was only pulling ~262W...briefly. Then it would settle back down a bit.

So I ran an instance of Furmark at the same time. That kicked up the draw to 371W briefly and then it settled down to a fairly steady state of 333W.

I used the boost fan button widget whilst doing these runs if that's of interest/important.

NOTE: This is also _not_ w/hmscott's undervolt request (saw this post first).

 

Good idea, better to not mix the tests together.

Is this power test on the 2 x 230w? Do you have a 330w to swap in?

That's about right then, as MSI usually trys to keep the load at 80% of rated power maximum, which would be 375w @ 460w rated.

 

Oh Sorry. This was w/the Eurocom 780W supply. It has the nice IVPT display on what its outputing.

My WattMans (WattMen(??)) are hooked up to my NAS which is resilvering atm so it's not in a good state to bounce it. So I can't give 'from the wall' data. But here's a grab of the same test w/the dual 230W bricks. In single and dual config.

 

Ok excellent.
So MSI isn't limiting you to 230W power by not detecting the adapter "dongle."
So that's good news for other users who want to buy the Eurocom 780W PSU.

 

Hmmm, I actually got a totally different read on that...

It looks like even though there is 780w available, power is still limited to what the laptop expects from the 2 x 230w PSU's, so nicely less cluttered, but no more power delivery than stock.

 

/nod

I guess while it's not the super blasto-jammo best case outcome, _my_ aim was merely to lose the dual supply abomination. And in that regard it's a raging success. Insofar as getting all that MSI intended one to get out of the thing given the supplied power 'solution', this won't be any worse. And given the overhead/active cooling of the Eurocom unit, it _should_ last longer.

 

I'm trying to comply w/your request but I am missing something. I have tried to enter the 'minus sign' in the offset field w/no success. It doesn't take freeform text entry. +/- but minus has a floor of 0. I even went into AMIBCP to see about adding it there in case that's what you were intimating. It won't let me add a - value there either.

I looked around, further and the only spot I can find a 'sign', although the text calls it 'Prefix', immediately under IMON Offset in the Core IA VR Domain tree.

Where is it that this should be set?

 

Check out this thread post, and previous / subsequent, he had to uninstall DGC before he could set the minus CPU voltage offset:

http://forum.notebookreview.com/thr...cussions-lounge.794968/page-798#post-10577695

"By the way, I have unistalled dragon center and installed xtu, afterburner and silent option. Now the negative voltage offset that I set on the BIOS works fine...." Last edited: Aug 5, 2017 - jaime360, Aug 5, 2017

I use XTU only for setting the undervolt. It's easier to resolve the lowest stable offset voltage in Windows, and let it load there at boot. For me XTU and DGC worked ok together, but I recommend uninstalling DGC so there is no conflict with XTU, and now it looks like DGC interferes with the BIOS CPU Offset Voltage setting.

Some have reported setting the final undervolt value, negative voltage offset in the unlocked (Svet) BIOS, so they can run undervolted on Linux.

I would assume you would select over to the minus sign, or use the arrows / tab from the text entry box, then +- or Enter then select + or - from a list... play around.

You could use XTU to zero in on the best undervolt using XTU stress / benchmark, then when you find the lowest undervolt you can uninstall DGC and set that undervolt in the unlocked BIOS - don't forget to reverse the other BIOS voltage tweaks.

 

It looks like the 8950HK CPU is a great undervolt'er even at high OC:

If we can do the same on the GT75 8950HK, around -140mV @ 4.5ghz - given CPU variances he might have a "golden" sample - undervolting at that level might run cooler than the fixed BIOS hack.

We need some people to try the undervolt method on the GT75 8950HK...

 

Give me the money and i will try it

 

what about the gt75 titan with the 8850h ? anyone got one of those ? would it bottleneck a 1080 ? im also interessed in benchmarks and temps. Also wanted to know what they mean by partially unlocked x)
thx in advance

 

It wouldn't bottleneck a GTX 1080

email [email protected] and ask him about the temps they have recorded for that laptop and ask him about your NBR Discount

 

Doesn't work. I took a video and converted to an animated gif but it's too large to upload. Regardless, it doesn't allow for a negative voltage.

XTU does some really opaque BIOS setting changes. I entered a -140mV offset, applied it, rebooted, and entered the BIOS and navigated to the OC menu. In the offset field was 140. So I thought perhaps since it's a range from -500 - 500 that values < 500 were interpreted as negative. I loaded the default values, saved the settings, and went back to the field and entered 140, saved, and got back into windows. After bringing XTU back up, the offset was a _POSITIVE_ 140mV. So _SOMEWHERE_ it's setting some sign. And that's only the beginning of the XTU unpleasantness.

I was able to get down to -150mV undervolt, through XTU, stable. I apparently didn't take a screenshot of that so that's why -140mV is the lowest attached. In any event, it ended up getting into a state where regardless of the undervolt, it almost immediately throttles due to power limit. Note the -100mV and the throttling. Then I backed out the offset (back to 0), went back into the BIOS, set the AC/DC loadline values back to 1 and the current limit to 800 (200A max). Once back in windows I was immediately throttling. I could barely break 1k in Cinebench.

Back to BIOS, reload defaults, reset loadlines to 1 and current to 800. And _that_ put things back into a state of sane baseline perf.

W/those BIOS settings, I can run a multiplier of 45 across all cores stable @ full tilt for about 10min. Then it'll start to get too hot and throttle.

 

What's the reliability effect of undervolting?

Can it cause mid to long term damage to the CPU ICs?

 

XTU lets you increase the power limits, you'll find them if you look, it's right there in the basic list of settings listed set / ready to apply to the right.

Don't fight it, thousands have done it successfully and I am sure you can too.

IDK about setting the negative sign in the BIOS, I do know many have done this, so it is possible. Like that post said having DGC installed messes with the settings in the BIOS, so keep that in mind too.

DGC can interfere with XTU setttings done through the XTU interface, but I can get them to behave if I set the DGC settings first (4.1ghz), then increase the multipliers with XTU (4.5ghz), but if I make a performance change in DGC I have to redo the XTU settings again - power limit increases too, so it's best to run with out DGC.

Running XTU + MSI Afterburner + RTSS (limit FPS to display refresh) + Silent Option gives you finer grain control, and there's no need to set anything in the BIOS.

IDK if going back to the stock BIOS would help with your negative voltage setting in the BIOS, use the unlock sequence to access the ability to change the sign - that's all I've heard about making the negative undervolt in the BIOS.

Typically before the 8950HK, undervolt at full OC was small - -15mV for me at 4.5ghz on the 7820HK, and no undervolt on the 5950HQ at 4ghz with +100mV needed at 4.3ghz.

So the 8950HK is the first one that gives a substantial undervolt at full OC, as you can see by my other post about the Aorus.

To me this means it's worth pursuing. If you hadn't had the BIOS hack to try first, you'd already be there, and not worrying about trying to set the undervolt in the BIOS... sigh. Too bad, undervolting is quick and easy, without complicating things with BIOS settings too.

Anyway, others will be coming along soon and may be able to show the way for you later.

I wouldn't waste time if you aren't having fun, but would be happy to see your results when you get there.

 

CPU IC's?

How would you imagine that tuning the voltage to the CPU to not be excessive would have negative consequence?

There are 10k's that have been doing this for a very long time and there is no negative consequence.

So, no, no problems so far after many years of tuning the CPU voltage to be correct for each specific CPU, on laptops or desktops, or servers - either through CPU voltage offset (undervolting), or using whatever tools the BIOS offers for tuning the CPU voltage specific to that CPU.

Tuning the CPU voltage to be precise with a voltage offset is a good thing, not a bad thing.

Spoiler: More info...

No problems OC'ing with the stock voltage hasn't been a problem either. But, overvolting too much to get higher clock rates has been known to "go too far" and kill CPU's, so make sure to do that with care.

Fortunately laptops have far more limited cooling options than desktops, so only a few extreme OC'er's overvolt laptop LGA CPU's, and those that do don't worry about blowing their LGA CPU's because they are socketed - extreme OC'ing like that isn't of interest to me.

BGA laptops like the MSI GT75 are a far better and easy situation to work with. Undervolt, OC with undervolt or stock voltage, tune the fans, limit the FPS to display refresh, and you are done. Also much safer.

 

Did you not look at/see the screenshots I attached? A few of them contain XTU shots of the very settings you're telling about. I know of the settings.

I'm not fighting it. I find it to be a **** tool because it makes opaque alterations to the BIOS settings that it doesn't cleanly back out when one asks it to do so.

It's not a DGC issue as that's not installed. It's 100% not the VR offset voltage field that can be negative. At this point unless or until I see some screenshot or direct path to the setting in question, I believe it not.

DGC was the first thing I uninstalled. From the time of my initial post after having gotten the 2nd instance of this machine, has DGC been uninstalled.

It's a gsync panel so limiting refresh to the panel doesn't make sense.

It may well be in some other location. I mentioned earlier the only location I found that had any potential to be sign-like but was unrelated to the VR voltage offset field. The stock BIOS doesn't alter this.

I'm only 'worried' about it because I think it's ********. As in it's a myth. An untruth. An impossibility. _That_ field absolutely can not contain a negative sign. Not whilst in the BIOS, custom nor stock nor through trying to 'force' it with AMIBCP. There aren't a great many permutations on +, -, or, ENTER. The only freeform entry allowed are the numeric keys. + increments by 1 and - decrements by one until it hits 0, then it does _nothing_!

BIOS OC and voltages and whatnot tied to that are what I am interested in. That's the single source of truth/authority for the things. Not in some after-the-fact app of questionable construction operated by an OS of equally suspect pedigree.

I have zero desire in predicating the health and safety of any longterm OC I might wish to emplace on the success or failure of Windows starting it at the correct time. Window's lease on life on this machine lasts until I can get Linux as a first tier entity going. After that, it gets relegated to VM status only. And if that can't occur, I unload it onto my son and wait for System76 to refresh w/an 8950HK variant.

Sent from my Pixel 2 XL using Tapatalk

 

G-sync uses V-sync or other frame limiter - like RTSS - to stop over rendering frames not used past refresh. It's an effective method of reducing load on the GPU / CPU and generating too much waste heat.

Do some searching on the site, there have been posts over the last couple of years with people confirming that they have set the negative CPU voltage offset in the MSI BIOS.

Whether it's a patched BIOS alone, or a bit of Svet BIOS tool magic, IDK. Perhaps you can DM (direct message = PM) those posting about and connect for some additional insight on how they accomplished this.

You could also contact Svet yourself and ask nicely, he may just answer you without requesting a toll payment...

Private Unofficial Modified BIOS'ses & EC-FW
By entering this section, you know these are UNSUPPORTED AND UNOFFICIAL BIOS files, and donation is required.
https://forum-en.msi.com/index.php?board=51.0

Whether you personally like it or not, many thousands use XTU and TS in Windows to undervolt and OC successfully.

If you get it working, please let us know, but please don't upset yourself further, it sounds like you've gone as far as possible with it and I wouldn't want to suggest you waste any more time on this.

Hopefully someone interested in undervolt / OC via Windows will show up with a GT75 8950HK so we can explore the OC undervolt, as on the Aorus, I'm sure the GT75 cooling can take things even further.

Good luck, and have fun

 

@hmscott

I am not sure if there are any long term side-effects, but damage by undervolting is not unheard in power electronics.

https://electronics.stackexchange.com/questions/65824/can-electronics-be-damaged-by-undervolting-it

 

DON'T USE XTU TO UNDERVOLT !!!!!!!!!!!
you already undervolted by setting the IA AC DC's to 1. XTU undervolting is not needed. XTU is only needed for users who were unable to change the AC DC, because *THAT* setting acts as an overvolt basically.

OC menu only shows the offset. Entering any value in the OC menu will become positive. It is NOT possible to enter a negative value in the OC menu. People have complained to MSI about that for years now.

The negative value you entered originally in XTU is still negative. Its shown in Overclocking Performance Menu->Processor. The "OC" menu only shows the offset. But entering anything here makes the offset turn positive instead of negative. So make all your changes in overclocking performance menu->Processor.

And why are you using XTU? Uninstall it. You don't need it anymore.

 

Those are all voltage control voltage starvation situations, completely different from the undervolting of a CPU.

Undervolting as a word can be applied in different circumstances that are completely different, and I would think that would be easily differentiable.

Tuning a CPU voltage to match supply vs need isn't starving the circuit of needed voltage and current, it is fine-tuning the voltage to the need. Very different than starvation - not supplying enough for the need.

You're barking up the wrong meaning of undervolting. )

 

He knows that they aren't compatible, as do I, and he disabled the IA tuning, but it's possible that isn't enough, and I suggested to reflash the original stock BIOS and default settings to test undervolting as a solution vs BIOS IA modding.

There is a post I quoted here from an Aorus laptop owner with a 8950HK that is undervolting at -140mV at 4.5ghz, with excellent temperature results without re-pasting.

I was looking for volunteers to test it on the G75 8950HK to see if there is more headroom for cooling with undervolt on the new 14nm process the 8950HK uses - as it appears to be - previous gen couldn't undervolt that high at full OC.

The IA BIOS tweak gives about 10c @ 100% CPU load, the same as a -100mV undervolt gives, but that is at stock freqencies, with less undervolt available previously at full OC, but now if the 8950HK - and other 8th generation CPU's - can undervolt higher at full OC, there might be even more cooling available.

There is no reason to re-paste or punch holes in the beautiful G75 when undervolting or IA tweak drops temperatures by 10c at load, and if there is even more thermal drop available in the 8th gen CPU by even higher undervolting at OC, this is even more so a good reason to not apply more extreme cooling measures.

I think he was looking to lock in a higher undervolt in the BIOS for his eventual Linux use, a perfectly good goal - using XTU to quickly find the best undervolt setting before fixing it in the BIOS is also a good method toward that goal.

He is having problems mixing the methods and I suggested he find out from Svet how to get the minus undervolt set in the BIOS directly, as there have been reports from NBR members that is how they did it.

Hopefully someone with less need for a BIOS change to tune the voltage for non-Windows use will take delivery of their GT75 8th gen CPU and we can see how far the undervolt will take the thermal savings at 100% load, soon.

 

@hmscott
You're confusing the poor guy and making things FAR FAR WORSE here. And it's really selfish and elitist for you to persist in this manner.
I saw his screenshots.
He was TDP throttling because XTU decided to put a 45W TDP on his CPU for some bizarre reason. I don't know why.
Also, XTU can cause strange things to happen to the MSI Bios. One time i tried putting a power limit of 511W into XTU (back when I still had it installed). I BSOD'd almost instantly. That was the last time I ever used XTU.

And he was thermal throttling later because him just touching the offset in the 'OC' menu made it turn positive. He was probably throwing 1.45v into that poor laptop.

XTU is NOT NEEDED at all. He unlocked his Bios menus. I do NOT know if he unlocked "Overclocking Performance Menu"--I doubt he did (to unlock that, he needs to set that menu and all SUBMENUS in overclocking performance menu to "Supervisor" in AMIBCP, and problem solved.

The issue is that he doesn't know about the "problem" with the default "OC" menu that MSI allows everyone to use, that does NOT allow negative offsets for adaptive voltage, nor did he know that 1) negative offsets CAN be entered in overclocking performance menu (no one told him this before, not me, not you, not anyone), and 2) XTU's negative offset when entered in XTU, DOES appear properly in overclocking performance menu (which is a hidden menu), but does *NOT* appear in the normal OC menu.

If you, for example, enter -100 mv adaptive offset in XTU and reboot to Bios, the "OC" menu will show 100.
If you then go to XTU and enter a POSITIVE 100mv offset and reboot to Bios, the OC menu will show 100 again. Because MSI for some dumb reason didn't allow the offset to be SHOWN or CHANGED here.

However the offset that was entered in XTU originally **WILL** be shown in "overclocking performance menu-->processor", whether you entered negative or positive. It will be shown correctly here.
HOWEVER IF ANY OFFSET IS ENTERED IN THE "OC" menu (the default one), it will **ALWAYS** ALWAYS be positive !! ALWAYS !!!! So NEVER EVER EVER enter an offset here. Period.
You are free to enter direct static voltages, cpu multipliers, RAM timings, whatever, in the OC menu (although personally i would avoid doing ANY CPU settings here and do them ONLY in overclocking performance menu to avoid any "gotchas). Example: If you are using static voltages (override voltages), in "overclocking performance menu", and you previously had an offset programmed for an "adaptive" voltage, that offset will still appear in the OC section (but will be ignored). Likewise, if you previously had a static voltage programmed (example: 1175mv), and then you switch to adaptive voltage in (overclocking performance menu), with a -50mv offset (negative flag in overclocking performance menu), the STATIC voltage will properly disappear and be replaced with adaptive voltage (offset) in (overclocking performance menu), BUT IN THE OC SECTION, the STATIC VOLTAGE will STILL BE SHOWN.

Complete utter spahgetti code.

Now I'm going to say this again and then it's up to him if he wants to take my advice or wind up being confused more:

1) UNINSTALL XTU. it is not needed !!!! everything xtu can do can be done through the manually unlocked menus in the msi bios !!!!! (except the stress testing). EVERYTHING!!!.
2) even the default locked bios is powerful enough to set the package power limits to 200w and the ICCMAX (VR Current Limit) to 800 (200 amps) manually.
3) Set the IA AC DC loadlines to 1, in CPU Power and Performance ->CPU VR Settings->Core I/A Domain and leave them at 1. If VR Current Limit isn't also set to 800 in (Core IA Domain), set it to 800.
If you're trying to permanently unhide that menu via AMIBCP 5.0.2 (NOT 5.0.1), dump your bios capsule with fptw64 -d mybios.bin- -bios, set that and all SUBMENUS (required) for accessing this, to SUPERVISOR, then reflash.

And finally:
4) use overclocking performance menu to set your adaptive or static voltages and undervolts or overvolts. I WOULD NOT UNDERVOLT if IA AC DC loadline is set to 1 unless you are not overclocking very far, and do it very carefully.

That's all.