ThinkPad P50 blackscreen

[thessarb]

hi!
Got an issue with my P50. If i run tpfc on discrete gpu mode the screen goes black and never recovers. If i run it in hybrid mode the discrete gpu crashes and completely disappears from device manager leaving no trace. If i disable dgpu in devmanager, run tpfc and then try to enable dgpu i get error 31 (windows cannot load drivers for this device). This issue persist with every dual fan version of tpfc i found on github and regardless the OS version. tvicport is installed, BIOS is the latest version (1.67), and ive tried OS versions from W10 1507 up to the latest insider W11 22H2. Im running W11 insider 22H2 currently.
Thank you in advance

[Shuzhengz]

Hmm that's weird, the program doesn't actually access the GPU
Do you have a Xeon CPU or an i7? This might be an issue with how the CPU communicates with the GPU. I don't think this is an issue with the Quadro because the program works with my W540 (i7 X series and K2100M).

But the Nvidia driver (and intel driver) might be a cause so it's probably a good idea to update to the latest version (I also have nvidia's studio driver instead of game ready driver installed).

Also do the single fan versions also have the same issue? (if not then it might be an EC issue) You can find a pretty good list here

Thank you!

[thessarb]

thank you for your reply
i have the i7-6820HQ with the M1000m. single fan versions of tpfc work fine. the drivers are the latest ones from both intel and nvidia respectively (nvidia-studio drivers).

[Shuzhengz]

hm okay, I'll try modding one of the single fan versions into dual fan then

[thessarb]

Thank you

[Remalisa]

hm okay, I'll try modding one of the single fan versions into dual fan then

Hello! I have the same problem with black screen on P50. My P50 is Xeon/M2000M.
I took a lot of time to solve the problem and find a program to control two fans. I tried to diagnose the fans for problems through the official Lenovo Diagnostic Program. When im trying to check the fans I also got a black screen, I hope this information helps to solve the problem.

I really hope for you!

[Shuzhengz]

Thank you!

[Shuzhengz]

tpfc.zip
try see if this one works
code on #25

[Remalisa]

Hello, I will try it at Monday.

Thanks a lot in advance!!

[Shuzhengz]

yeah np!

[thessarb]

hi!
just tried it and it still blackscreens. this time windows doesnt seem to freeze on dgpu only mode tho as this time i can actually blindly shut it down via cmd. here are ythe logs captured by the program:
TPFanControl.log
TPFanControl_csv.txt

[Shuzhengz]

hmm from the logs it seems like the program crashed while trying to access the fan EC
i think it might be a problem with the the second fan address

[thessarb]

thanks for the reply!
is thre anything i could do about it or something you would suggest? (fan can be controlled under linux)

[Shuzhengz]

yea np!
is it linux mode for the laptop or running GNU/Linux? if it's the ladder do you have the source code for it? the two addresses might be similar
thank you!

[thessarb]

it is running GNU/Linux (PopOS 20.10) in tripple boot (also got MacOS Monterey beta). Here is what i use to control the fans in linux

[Shuzhengz]

oooh yeah i almost forgot that linux doesn't use address to control the EC. I wonder if it's a kernel level thing

[Remalisa]

@thessarb Hello! Im sorry that im very noob at programing and understanding of all of this stuff.
I correctly understand that the problem is not solved at the moment?

[Shuzhengz]

yeah, I'm currently working on resolving issue #20 with PR #31, maybe you can try that build out and see if it works

[thessarb]

@Remalisa np i am no different
@Shuzhengz thx ill try it out later

[Shuzhengz]

heres the link to a wip build
tpfc.zip

[thessarb]

hi!
just tried the wip build and tpfc launches finally, but it lacks cpu temps and cant control the fan. also it shows the fans as if they were stuck at 5120RPM. here is the log:

TPFanControl.log

[Remalisa]

@thessarb Hi! Can you please explain to me how did you start the program? I was trying, but I failed ;(

[thessarb]

@Remalisa i didnt do anything different, just extracted the zip and ran fancontrol.exe (W11 Insider Dev Build 22593)

[Remalisa]

Hello!
I have finally opened the program. I also observe that the temperatures and fan speeds do not change in numbers. I also noticed that when the manual mode is selected from 0 to 7, the fan speed simply does not change. Moreover when you select manual mode 7, it automatically changes to program mode .ini

[Shuzhengz]

hmm seems like the EC controller is in the wrong address. Try downloading RW everything, then open EC Controller and search for your sensors

[Remalisa]

@Shuzhengz trying to install rw apears the same problem as in this link: https://www.bleepingcomputer.com/forums/t/709973/rw-everything-not-working-after-windows-build-update-to-the-last-version-1903/

maybe i can resolve this problem using other program?

[Shuzhengz]

hmm are you on windows 10 or 11?
you can try using another program, any that reads the EC (maybe MSR as well) should work, I just haven't used those other than rweverything on windows

[Remalisa]

I'm using windows 10 21H2. Ok, I will try something else, thanks!

…

On Fri, Apr 8, 2022, 11:26 PM Tom Zhang ***@***.***> wrote: hmm are you on windows 10 or 11? you can try using another program, any that reads the EC should work, I just haven't used those other than rweverything on windows — Reply to this email directly, view it on GitHub <#23 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AYF5QGIKBALBARK3DB2VIL3VECI7NANCNFSM5QF5RKSA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[Lego737]

Unfourtnately, it sadly didn't work for me. I tried both builds on dGPU mode and the screen still went black.

[Shuzhengz]

hmm does it go black immediately or after a few seconds?

[thessarb]

@Shuzhengz thank you, ill try it as soon as i get home

[thessarb]

@Lego737 boot in hybrid mode, disable dgpu from dev manager so you dont BSOD in a few minutes, open rweverything and choose Embedded Controller, run tpfc. Look at the value at the row 30 and column 01. The value shpuld be 41 at the start, check the value that is input by tpfc.

[Lego737]

Yes, it goes black instantly.

[thessarb]

@Shuzhengz hi again! im happy to report that tpfc doesnt cause black screen for me anymore. also open hardware monitor seems to work and the values match both hwinfo64 and hwmon

[Lego737]

@thessarb, how did you fix the issue in dGPU mode?

[thessarb]

I did absolutely nothing. Just tried it in hybrid, dgpu didnt disappear. So i tried it in discrete mode and didnt get a black screen. The temp readout is kinda funky (the degree symbol is weird) but other than that it seems to be working fine for me

[Lego737]

I tried his latest build in dGPU mode and my screen still went black.

[thessarb]

Try the one Tom uploaded here, thats the one im using

[Shuzhengz]

@Lego737 this one

[Shuzhengz]

@thessarb okay so using C# is giving me a ton of problems, I think I'll try another method by reading the msr

can you try and see from rweverything's cpu msr page if address 0x019C (the temperature of each core) and address 0x1B1 (package temperature) reports the cpu temperature?

the values would look something like 0000000088260a82, you can ignore everything else and look at the "26" part (decimal digit 16-23 in the more detailed panel), the value when converted to decimal should be half of the cpu temperature from what i can tell

thank you!

[thessarb]

I will look at it tomorrow after reinstalling windows. Today macos just randomly decided to wipe my windows install clean off the ssd.

[thessarb]

hi! sorry for the delay. i opened the msr page but i cant seem to find the said addresses. the list is pretty short and the values dont seem to change at all. a considerable amount of the values are just strings of 0-s.

[Shuzhengz]

no worries
the MSR register is actually in the user value section (theres a button on the top right of the window next to the save button to add the addresses), and there should be a value for each of the CPU cores. For 0x019C they should be slightly different, but for 0x1B1 they should all be the same
thank you!

[Shuzhengz]

[thessarb]

this is all i get there

[Shuzhengz]

yeah so you have to add the values in the pop up window

[Shuzhengz]

just type in the addresses, click add, and it should show up

[thessarb]

just looked at it and i cant relate the values to the core temps in a meaningful way. when the cpu was at 36 degrees, the value there was 42, when the temp was at 60 the value was 3A and when it hit the thermal throttling point it went crazy cycling thru all possible numbers and letter up to F

[Shuzhengz]

yeah in Intel's manual there is supposed to be a bitfield modifier for the values to convert it into hex (I believe it's 15:0), I can add that in the code, but I have no idea what it means or actually does.

[Shuzhengz]

page 20 of this

[Lego737]

@Shuzhengz

I love that you were able to fix this app as I can now use my P50 without it overheating a ton! So, thank you for that!

However, I've come accross a problem that I'm currently looking into. When I launch a game, that being Fortnite, and have the app running in the background, Fortnite ends up crashing after 10 or so minutes. I did, at one point, reinstall the dGPU driver which seemed to fix it but in the end, it still crashed. Do you think that this is the fan app on my machine or could it be something else?! I hope to hear from you soon :)

Lego737

[Shuzhengz]

hi @Lego737
thank you! can you try using a CPU stress test tool like cinebench CPU or blender CPU render, with the fan control set to manual 80 and see if it crashes? this will narrow the issue down to CPU and fan contol or GPU.

also currently I'm still working on temperature for the p50 because it's not in the EC, so it would probably be a good idea to avoid using smart mode for now and stick to manual mode to avoid overheating

[btom1970]

I have a P-50 Xeon M2000. TpFanControl only show CPU and APS temp. This is normal? The others sensors can't be displayed? GPU and others?
I must use the program in manual mode? What values are recommended between 0 and 7?

[Shuzhengz]

I have a P-50 Xeon M2000. TpFanControl only show CPU and APS temp. This is normal? The others sensors can't be displayed? GPU and others? I must use the program in manual mode? What values are recommended between 0 and 7?

unfortunately yes, temps are broken right now as we can't find the address for that. I recommend setting the fan as high as possible (7 or 80) to avoid damaging the device, but it shouldn't be an issue as the CPU will slow down if it overheats.

Edit: typos