The Official MSI GT76 Titan Owners and Discussions Lounge

Even though I don't have much experience here. But I still disagree with unclewebb. I think if there were any problems with the power supply, the laptop might not turn on at all. But I do not exclude the hardware problem.
I would advise you to check the laptop's behavior in other games (for a complete picture) For example, Battelfield, Tomb Raider. You should look at how stable your FPS will be. Also, during games, the Open Hardware Monitor must be turned on to track the frequency and power consumption indicators. Then you need to run the Aida64 test with indicators (cpu, fpu, cache, gpu) 5 min. This will show the maximum load on the cpu and gpu, as well as their maximum power consumption, and whether the laptop really lacks power.
Also show your power settings on Dragon Center.
You also need to try removing the Dragon Center and resetting the BIOS settings, because many users here on the forum said that the Dragon Center will not work properly and interfere with the power settings. We need to check how much this will change the situation

 

I removed the Dragon Center from my PC and the BIOS and Firmware are E17H1IMS.10D and 17H1EMS1.103. As far as I can tell, Open Hardware Monitor is showing both Frequencies and Power.

When I tried running Aida64 (Extreme), I just get a very repeatable BSOD (" BAD_POOL_CALLER"):


I do remember an issue w/ the old BIOS not charging the battery to 100%, but after upgrading to the E17H1IMS.10D version, then the battery started charging again.
The Dragon Center (2) is gone now, and like I said, I only installed it originally because they mentioned it had the battery calibration tool in it.

 

@jpjones69 - Your new log file shows a large negative number in the BAT_mW column. This and the Power Status Change notifications both confirm that your laptop is constantly switching back and forth between battery power and AC power. Every time this happens, your CPU cycles between 798 MHz and full speed.

It is quite possible that less demanding games will not cause your power adapter to overload like this. Other games or less demanding tasks might run perfectly fine. All I know is that power delivery is not adequate for the game you are trying to play. A properly functioning laptop will not show any Power Status Change notices in the log file while gaming. There messages are not being created by ThrottleStop. They are messages that Windows is sending out. Windows is designed to notify software when a device switches from AC to battery power.

https://docs.microsoft.com/en-us/windows/win32/power/pbt-apmpowerstatuschange

 

have you tried other demanding games?

The battery did not charge to 100%, possibly due to a life-saving function. This is how the battery is programmed, otherwise, with a constant charge up to 100%, it will quickly lose capacity. In addition, the Dragon Center has a function for switching states of charge (50 - 70 -100%), this can be switched.
If you get blue screens, then it is likely that your Windows 10 has defects, there may be unsuccessful updates. Recently, many users have been complaining about the latest versions and updates of Windows 10

 

Reimaging Windows 10 is such a painful process for me, but that's all I have left. I'll try that and see if it resolves the AIDA64 BSOD. If not, I guess I'll be forced to downgrade both the BIOS and Firmware back to 2019 (shipped from factory) versions and try again.

The only games I've tried is either Warzone, Overwatch, and Internet Chess.
The issue I'm having occurs easily with Warzone and Overwatch (Ichess is fine).

Edit: I saw some other MSI computer had BSOD when launching AIDA64 as well and they said the fix is to have more Intel programs installed (i.e. Intel Rapid Storage app, which I do not have currently). I'll try installling this first and see if it resolves the blue screen so I can provide some data from AIDA64.

 

did you have this problem on a previous BIOS version?

 

Yes. I had the same stuttering problem w/ the previous BIOS (the one that was factory installed) which is why I updated it (along w/ the firmware) to see if it would fix the problem (which it did not).

I have never used AIDA64 before, so please pardon my ignorance if I'm not running it right or with the correct Options/Preferences, etc. I will say that I could not find the dial at the bottom to stretch out the time (X-axis). There was only a slider at the left side to expand out the Y-axis.

After installing the Intel Rapid Storage Driver, I was able to get AIDA64 to run just fine on my computer without BSOD'ing - here's the results from a 10 minute stress-test.

Statistics Summary - CPU Throttling!!!



Temps:



Voltages:



Voltages (zoomed in):




Power:



Clocks:



Unified:



Thanks.

 

very well. The only thing is, it's a pity that you did not attach the OHM data, and we cannot see the readings of the GPU 2080 during the test.
Anyway, unclewebb said 130-140 watts is too much for your PSU. But we can see that your processor is quietly consuming 173 watts under heavy load. And the laptop does not shut down during the 10 minutes of the stress test. In addition, the processor frequency is stable at around 4.3-4.7 GHz, except for a frequency reduction only once at the beginning of the test. The rest of the time it holds the load stably. Means he does not have a lack of power supply.

 

In fact, this is not the first time this problem occurs in MSI laptops. I've seen complaints from other users on other forums as well.
Perhaps you need to contact Falkentyne, he really understands all the intricacies of BIOS and power supply settings. There is more information in this thread
http://forum.notebookreview.com/thr...-cmos-and-prevent-common-issues.812372/page-8

 

I'm trying to ascertain whether you're having the same issue I was.

What would really settle thing would be if you would run MSI Afterburner's in game overlay while you run a game benchmark. while displaying the CPU and GPU frequencies.

If both the CPU and GPU are constantly bouncing up and down, nonstop, it probably is the AC adapter problem. I'd also like to know if, while this occurs, does the laptop ever show as going from AC power to unplugged, back and forth.

 

Agree, this seems like a hardware problem.

 

EC controller firmware, BIOS, Windows and Dragon Center have access to laptop power management. So each of these variables can cause this problem.
But yes, there is a possibility of a hardware problem too.
And in order to determine this, you need to exclude all the possibilities of the program reason.

 

When I do this, do you want the "hml" files saved and provided or do you want me to convert them to a .txt file instead? If you want a .txt file, which metrics do you want to see (Just "CPU Clock" and "GPU Core Clock")? Just curious because if too much data is logged (via the "Global monitoring/Begin logging" function), the lines span multiple lines and it gets real confusing which number is which.

 

try to disable the CPU C-State on BIOS, and see if this changes the situation

 

So after downloading and installing MSI Afterburner, I went ahead and did benchmarks w/ both Warzone and Overwatch and here are the results.

Overwatch Benchmark
Overwatch (which I did in a Custom Game w/ a bunch of Bots, and I play on the lowest settings, so these are quite horrible for my GT76 Titan)
14-08-2020, 15:47:33 Overwatch.exe benchmark completed, 30284 frames rendered in 333.671 s
Average framerate : 90.7 FPS
Minimum framerate : 27.7 FPS
Maximum framerate : 190.1 FPS
1% low framerate : 24.2 FPS
0.1% low framerate : 6.7 FPS


Sorry about the really bad quality graphics here, but it is what it is.

So for this benchmark, the GPU1 (RTX 2080) Core Clock did dip as low as 555 MHz, while the CPU Core Clock dipped as low as the good 'ole 800 MHz.
All Temperatures seem fine to me - nothing anywhere close to overheating; but I'm guessing the way Thermal Throttling works is that if any component gets an "Instantaneous spike" over some threshold, it sends some global warning signal which causes the throttling to briefly occur.
So the RTX 2080 seems to be doing it's job, but the CPU Usage % for Overwatch is only around 25% (during gameplay)... wtf !?!

Warzone Benchmark
On to the Warzone benchmark (which I did in BR Solos mode and also on the lowest graphics settings possible to try to get the most FPS).
Warzone (I also play w/ everything disabled and on the lowest of settings to try to maximize FPS):
14-08-2020, 15:23:31 ModernWarfare.exe benchmark completed, 53138 frames rendered in 822.938 s
Average framerate : 64.5 FPS
Minimum framerate : 2.9 FPS
Maximum framerate : 154.3 FPS
1% low framerate : 15.3 FPS
0.1% low framerate : 1.0 FPS


Sorry about the really bad quality graphics here, but it is what it is.

GPU behaved better here, well, at least according to this data... GPU Frequency didn't seem to dip as far down as with other game (see above). But yea... the CPU Clock went back down to 800 MHz again.

One other tidbit that seems fishy is that if you inspect the CPU usage % - it was way higher in Warzone (hovering around 50%), whereas in Overwatch, the CPU Usage % was pathetic ( ~25% during gameplay... NOT menus).

Now as far as seeing the battery go from Charging to Not Charging, I did not see the icon ever toggle. However... before I updated to the Firmware/BIOS to 17H1EMS1.103/E17H1IMS.10D, even with No MSI Dragon Center installed on my PC, and also with Windows 10 on "High Performance" (and no Option for "Adaptive Brightness" anywhere to be found anywhere), I do recall while the PC was plugged into the wall on AC power, seeing the screen constantly toggle from regular brightness to dim, regular brightness to dim.... repeatedly.
On the new Firmware/BIOS, I don't see this any more and also the Battery is finally showing full-charge (after I "Calibrated" it).

Hopefully, this data helps.

I will look into trying this, but I gotta be careful as I am an noob (idiot) when it comes to that stuff and I don't want to brick my PC.

 

there is nothing difficult in this, it will just turn off the power saving of the processor

 

So I Disabled C-States in the BIOS, restarted, and reran another Benchmark test on Overwatch - I made a mistake w/ the data I provided above, with the original MSI Afterburner benchmarks in that I had forgot to disable the Frame Rate Limiter, but I did that this time.

Basically, the RTX 2080 core clock dropped as low as 300 MHz (multiple times) during gameplay, but the RTX 2080 "Memory Clock" only dropped to 5001 MHz (also repeatedly).
The CPU Core Clock continuously cycled from 4700 MHz to 800 MHz (again, during gameplay) and the CPU Utilization was, on average, a measly 25-30% whereas the GPU Utilization was at its limit the whole time (95%+).

Thanks for everyone's help with providing feedback with this.

 

This behavior is similar to what i see when i make my GS75 draw more power than is allowed by the EC.

I think the EC is broken on your machine.

 

And what are your indicators of 2080 in the Aida64 test. Does the video card hold the frequency stably, or does it also have a drop?
5 minutes will be enough. You can also check it in the FurMark test. It is advisable to take measurements with an open OHM

 

I guess the next step is for me to downgrade both the BIOS and EC Firmware back to its factory-shipped state and rerun benchmarks again. I'll try that next.

I'll try to rerun the Aida64 test for 5 min. with the OHM open and provide that data (w/ C-states disabled).

 

I ran Aida64 w/ OHM (and C-states disabled in the BIOS) and here's the results of the Stability Test.

Temps:



Voltages:



Power:



Clocks:



Unified:



Statistics:



Other (bottom) OHM values (after the 5 min. test was complete):



Thanks.

 

So the GPU load and clocks were stable during the test? was there a drop in frequency during the test?

 

I may be wrong, but it doesn't seem like that to me.

In the Aida64 5 min. Stability Test, both the CPU and GPU frequencies dipped really low.


This exact same behavior (both CPU and GPU frequencies being Throttled) was seen while playing and/or benchmarking Overwatch and Warzone.

When doing things that are not very computationally expensive (i.e. playing chess on the internet, there was no Frequency Throttling of any kind).
Edit: After doing yet another test: Staring at OHM while just typing this post, I see that the GPU Frequencies are low (300 MHz) to begin with - so I can't draw any conclusions about the "Lowest" GPU Frequency value at this time (that is apparently "Normal" to have a low GPU Frequency value). Or maybe the better process would be to reset the low frequency while playing a game and see if it drops back to 300 MHz while the game is running. I am going to leave OHM up for a while and see if the CPU Frequencies dip to 800/798 MHz, and will add that info here, but as of now, they haven't dipped much with just typing this post.

So, to me, it seems like the same [undiagnosed] root cause (EC, BIOS, Power Adapter/Battery, or God forbid a bad Mobo) is the culprit to both problems (the Aida64 test results and the game playing/benchmarking). I'm new to this stuff, so hopefully it was a bad mobo or popped capacitor, the system wouldn't even boot (which isn't happening here).

I'm going to do more research into what control I have over EC settings or customization. I may or may not peel the EC version back to how it was when it was shipped to me last year, but in any event, I already know based on previous testing using TS9.2 that there are tons of "POWER STATUS CHANGE" events logged while playing games.

 

The graph shows that the processor resets the frequency only 1 time, this does not look like a problem with the power supply. Otherwise, the frequency would have been low throughout the entire test, or the processor would have had a lot of frequency drops.
As for the GPU, this low indicator could have been already at the end of the test or before it, unfortunately we do not see a graph for the GPU to determine this exactly.
In addition, we see that the total consumption of the cpu + gpu system was 176 + 162 watts = 338 watts, and 100% utilization of the cpu and gpu.

 

Yes, it may be incorrect operation of the EC firmware, incorrect operation of the Windows performance mode. Recent versions of Windows are very unfortunate. It may also be the wrong work of the Dragon Center.
I had several versions of the Dragon Center, and on one of them the processor had the same problem with a drop in frequency to 0.8 GHz, despite the fact that the performance mode would be in the maximum position. Still a software problem.

Also, some games may have poor optimization and may not load the GPU properly. Or it may be an unsuccessful version of the nvidia driver, when the v-card does not know how to behave. It can continuously turn on power saving when not needed.
Modern Windows, drivers and software are configured in such a way as to save energy, sometimes even when it is not required. And this is a big problem for laptop owners.

 

I think you need to contact the official support of msi, or the seller, and say the laptop has a defect with the power or the software that controls it. And it has regular performance failure

 

normal get 100C cpu --800mhz drop
and enable c-states ,,this config need sleep cores in lower load
set better pl limit 90-94C max "no drops down"

edit.. gpu ?? min -normal ,,check max or "current" in 3d stress "adaptive clock" BASE
+ boost clock works only in games or need lock "afterburner"

 

it gets drops to 0.8 GHz even when the temperature is not high in games. This test was to test stability for maximum performance and power supply

 

https://www.hwinfo.com/
sensor -create log from games ---send me this file

ec reset
https://www.xoticpcforums.com/forum...resetting-your-laptops-ec-embedded-controller
1x 30s msi brand

log guide
https://support.lumion.com/hc/en-us/articles/360021829573-How-do-you-provide-a-log-file-from-HWiNFO-

some new nb ,,check input voltage vs battery voltage //vs wear level
need 100./. recharge in battery master "dragon center" for full performance

and same voltage pp input vs battery or crazy SPIKES --bad AC adapter --example
send file
big wear level --battery calibration ,,still get big wear "after calibration" --bad battery
https://forum-en.msi.com/index.php?attachments/bez-názvu-png.140364/

sorry translation

 

I think this is too difficult for a user like jpjones69. Having paid that kind of money for the Top Notebook, the user should not do such things. He must receive a finished working product with Top performance corresponding to his price

 

same like get 100C on easy load AIDA64

edit,,ec reset is normal way "power problems" after update bios/EC
,,battery calibration same "check wear level" or after 3 months is recommended

my recommended
need all time + AC + 90-100./. battery level for normal performance
MSI is complicated
and any overheat on GT76 is absurd ,,Top price
crazy see any limit on this NB ,,yes is absurd ,,but all this NB is same in stock

 

people get 100 celsius even on mobile processors, sometimes even 110 celsius with Reisen 4xx processors. Right there we have a desktop processor running at 4.7 GHz, no surprise

 

problem is PROCHOT and Tmax ,,hw slowdown ,,not possible cooling
9900k -100C max t-max

edit aida stress is normal ,,test prime95 -burn
this result is not ideal ,,just edit the video or similar work
and this config -- useless

but problem here,,heavy limit ,,HWINFO is better tool -check cpu limit reason from domains
or send file

 

Of course, we can fix the frequency at 4.3 - 4.2 GHz, which will be more than enough for gaming tasks. This will significantly reduce its consumption and, accordingly, heating. In addition, you need to adjust the speed of the fans, adjust the curve. And also do undervolting. And in the end we will get good temperatures.
But his problem is not the temperatures.

 

if you have a warranty ..I will use it
test only reset EC
load bios defaults ,.,nothing more

 

It would be great if that solves his problem. So we are waiting for jpjones69

 

need log ,,nobody knows ,,cpu is OK temperatures in games-example
and "ring" all time overheat --HWINFO send log

if you want more information ??
hwinfo --complete info from HW
need file

there is no point in figuring out what the problem is without LOG
or use warranty

translation

 

I think many PC users know this. Up to a certain point, the frequency threshold Intel processors consume a little power. And after a certain value, their consumption grows disproportionately to productivity. Along with this, heating also increases

 

Really dude...?
I thought I made it clear when I said (more than once) that I was a complete degenerate idiot, but I guess not.

 

he means well
the user should not solve anything on gt76 "top product"

,...
send file for more info ,,real limit reason
from gaming or when get problems
translation sorry
edit
or teamwiever -faster help resuts --send PM
and install MSI kombustor + hwinfo

 

ANY gaming laptop needs a repaste done with care, be it an MSI,Alienware, Asus, TongFang chassis, Gigabyte, Clevos, etc, etc, etc.
You either do it yourself with good paste and ensure proper heatsink fit, or it will lead to 100ºC under load, that and undervolting, both a must on any recent laptop..

I paid 2.6K€ for my GT75 and first thing I did was lap the heatsink and repaste.
If you dont want to do, either pay a shop to do it, or just dont buy a gaming laptop at all, because they all need tinkering.

 

Yes you are right. Unfortunately this is the case, they require improvement. And as far as I know, replacing the thermal paste voids the warranty, as the MSI representatives said. Maybe it depends on the country. So not every user will risk the warranty. And not everyone can do it. Some, after disassembling the laptop and changing the thermal paste, get a short circuit or damage the chip

 

The power adapter overloading should cause the power light on the adapter to actually turn off completely, and remain off until it's unplugged.

It's the Embedded Controller triggering this throttle. I saw another user who had this same problem but I don't remember what laptop he had, but I know it was a RTX MSI model.
I actually think it was...I forgot...it was someone on the Pascal TDP tweaker thread who had this problem, but he was trying to shunt mod his video card and ran into this exact same issue.
Maybe it was @senso I honestly forgot. But that was on a TDP modded video card. And I do not believe that model has Optimus.

@seanwee ? who was it.

I'm assuming a "master" total system power limit is being exceeded somewhere, but I only know how this works in the GT73VR and GT75VR. I can't help with the GT76.
You can download RW Everything and take a capture of the "EC" RAM tab (Embedded controller) and look very carefully at EC RAM Registers 30 and 31 (it's a simple X/Y grid, so the very first row at the top is values 00-0F, then the next row is 10-1F, etc), and look at EC RAM registers C5 and C6. See if they are changing values repeatedly. Pay very close attention especially to register C6.
"40" means you are being power throttled and C1 means you are getting full power. If it's switching rapidly between 40 and C1 (in register location C6), you have a problem. And I cannot assist with laptops I do not own.

 

Thank you for this information! I've tried googling more about this (for my case), but I haven't had much luck. Would you mind, please sending me some additional sources on what a particular hex value in each register means?
How did you know that "40" means power throttled ?

 

no need mod ,,not using 460w - 500w
170w cpu 160w gpu 30w system--aida64 stress
330w can still be used for this ,,not dualbox 2x 230w
send file
or use warranty ,,not modding EC ,,why

file guide
https://support.lumion.com/hc/en-us/articles/360021829573-How-do-you-provide-a-log-file-from-HWiNFO-

 

I can't give you ANY other information than what I gave you. You're on your own. Sorry.

 

I'll continue to google it until I find it then.

 

By trial and error and asking other members of this forum to try those values.
You wont find any other info about the MSI EC besides what @Falkentyne has found over time.

 

Especially because I don't have that laptop and I don't want to be responsible for blowing it up if you wind up plugging in random numbers into something.
The only thing I've seen consistent on GT76 / GS75/GE75/GT75/GT75VR etc is how to remove the throttle when the internal battery is disconnected, which I've posted here, the Pascal TDP tweaker thread (a few pages back on there) and the GT73VR thread, etc.

EC RAM register 31-->09
EC RAM register 42-->64 (these two are enough to stop limited power throttle from disconnected battery).

EC RAM register E3 ->Set to the 330W AC power limit value (from what I saw in GT73VR 1070 SLI and 1080 SLI, they use the base 230W or 330W values then they "Double" them if a second PSU is connected via a dongle or if the slave video card (SLI) is installed--there's no direct way to double a 1070 system power limit from 230W to 460W if you have a single 1070, but you CAN Change the power ID to 330W by using the GTX 1080 power ID instead.

I don't know the power ID's of anything past the GT75VR. A few (like senso and a few others) have posted GTX 2070 (330W) and GTX 2080 super (2x230W or 2x240W i forgot) power ID's in the Pascal TDP tweaker thread, etc.

But the throttle issue on the GT76 I have no idea if it's related to that. I don't even know if there is a >330W power ID or what the GT76 uses.

EC RAM register C6 is read only. You can't just plug in values in there and expect it to stick.
EC RAM register C5 can be written to, but using a value too small in it seems to force C6 to "40" for power throttling or it repeatedly toggles throttle on and off. 9E is the default value. I remember being bored and entering "80" here.
I also remember using a value too small will make the laptop repeatedly toggle the value in register C6 between values "40" and "C1". So that's pointless.

 

Thanks. And yea I understand you don't want to be responsible for me blowing up my machine. I actually hadn't planned on changing any of the values in any of the registers, I just wanted to know what the current values meant.