Clevo P870TM_TM1 CPU throttling because of battery?

Hello there, first of all I apologize for any stupid question because hardware is not my cup of tea.
I own a 2yo P870TM_TM1, and I'm having severe problems with it.

My problem is that when I use the 1070 GPU (one or both, in SLI) all of a sudden the PC fans start to go to the maximum and after a while the pc simply shuts down.

I tried to investigate the problem during the last weeks and so far what I did understand is:

1. using AIDA64 i was able to stress a CPU throttling problem till 8% (but maybe more, since after a few seconds I decided to stop the simulation)
2. using the same software, if I unplug the battery from the power supplies, the throttling seems not to be present.

Side notes:

1. I'm pretty sure the CPU fan is working properly because with the Clevo Control Center software I can force the fan to go to maximum and I can listen something like a rocket ready to take off, and at the same time I can see within AIDA64 that the temperature goes down a few degrees
2. Overclocking is off, and I never tried to play with it, since I should have enough power to play almost anything
3. I tried disabling SLI and use 1 single GPU at a time and the result is pretty the same.
4. I payed (50 euros) when I bought the laptop to have a proper delid, so I assume it was made professionally, but I cannot be sure 100%.

Any help in these? What do you suggest to do for further investigations and for a fix?
Your experience would be of great help.

Sorry in advance for my english coz it is not my native language, so in case anything is not clear just point it out and I'll try to fix it.
Cheers!

 

Does the machine beep at all before the power off?

 

Hello Meaker@Sager, nope, no beep at all. Whatever is on the screen, it freezes for a bunch of seconds, and than black screen. On a few occasions I also had to unplug both the power supplies to power the laptop, because their green led disappeared. As said, luckily they re-appeared after unplugging and pluggin again.

 

If I am interpreting what you posted accurately, the CPU is not overheating. What temperatures are you seeing from the GPUs under load? If the GPU is overheading, (one or both,) that can cause the fan behavior, drop in performance and unexpected shutdown.

If you remove the battery and the system seems to behave normally, the most logical assumption would be that there is a good possibly something wrong with the battery. It could be shorted. If it works properly with the battery removed, it would be best to leave the battery out until you can purchase a replacement. If it is shorted internally, it could pose a fire hazard.

It is extremely unlikely the CPU delid has anything to do with the issue you are experiencing.

 

As above, it does not sound like an ECU panic so it could be the charging circuitry for the battery shorting.

 

Thank you both, here my thoughts and tries:

"If you remove the battery and the system seems to behave normally ..."
​

Actually I never tried this, but I noticed that without the power supplies (so the power is given by the battery only) there is no issue.
Anyway, I tried to invert the test, and removed the battery as suggested, and tested with only the power supplies attached... black screen in a few seconds. Here AIDA report:

https://ibb.co/7VwR8r3

As seen from the report, the CPU was 86 degrees and the software didn't have the chance to state the throttling problem, so I guess you both are right in saying that this has nothing to do with the cpu (which I am glad).
As from the behaviour with the battery, I guess it is not the problem either, so I am gonna place it in its spot again.

Next suspect that comes to my mind... one or both the power supplies?

Trying to ask your other question: in AIDA64 test I disabled the "Stress GPU" option, and I tried to test again, and the problem still persist:

https://ibb.co/9yWZqxv

I dunno if this is really the right way to exclude the GPUs from the tests. In case, should I use another software?

Thank you both for any help you can give me...
Cheers!

 

I tried the power supplies too:

• one at a time with the Power Converter Box (this thing) connected
• one at a time directly plugged into the pc (without the Power Converter Box)

Same result, the CPU reaches above 90 degrees in less than a minute and AIDA states "CPU throttling, overheating detected".

If I unplug any power supply, and use only the battery, the degrees are always below 60.

 

Put up Hwinfo to follow temp. Hit Fn+1 to force max fans. Will the fans run at max? If they do, then run Aida64 stress test. If you see overheating there is something wrong with paste or heatsink.

Cpu won't run at max speed <if you run on battery - unplugged psu>. Hence it won't overheat.

 

Thanks Papusan, I didn't know what you stressed about cpu max speed with unplugged psu. I searched for more info about that and I must admit I am confused. You can see here why I am confused: https://ibb.co/s97rjws it seems to me that there shouldn't be a difference, but probably I am wrong (sorry it is in italian, but it is pretty clear that cpu power should be 100% on both power scenarios).

I installed Hwinfo. If I press Fn + 1 nothing happens, but I can force fan at full speed via Control Center via performance > system > fan speed.

This way I tested with AIDA. and I noticed something really strange.

1. the first test was without Stress GPU option: https://ibb.co/Yc22ZZf
You can see that the temp is overall below 70 degress. You can also notice that Remaining battery is in state charging, and everything seems ok, at least to me.

2. this second test resulted in a strange behaviour: I did the same as before, with the difference that I enabled also GPU Stress. Once started, the Remainin battery option switched from charging to battery, and indeed my psu seems switched off (no green led on it). I have to unplug and plug again into the power line to make the psu work again and be correctly recognized by the system.

I'm totally lost guys...


edit: thinking again about what happened before... could it be that since the psu are switched off due to something, that my pc simply shuts down after a while because the battery is totally drained...???

 

That or it cant handle a full game load when a brick goes down.

 

Yes, it could be that the system is shutting off because of a drained battery. As Brother @Meaker@Sager pointed out, your AC adapter may be failing and struggling to meet the system's power demands when the CPU and/or GPU are under load. If the power indicator light goes out on the AC adapter when the system is under heavy power load, that is a fairly good indication that the AC adapter needs to be replaced.

As a side note, system performance will be dramatically reduced under CPU+GPU load as well if your system is reliant on battery power, or if it is switching to battery power due to the AC adapter failing.

Simple test: Remove the battery. Connect the AC adapter. Run a stress test. Load test the CPU and GPU simultaneously. If the system turns off and the power indicator on the AC adapter goes out, the AC adapter is junk and needs to be replaced. Your Clevo should function fully and with no adverse symptoms with no battery installed.

 

Thanks to you both, I will try that simple test with both the psu, hoping that at least one is behaving correctly.
Should I also test in a more technical way, somehow? I just watched this video and the guy is testing the voltage with a digital multimeter, but I guess that to do a test like that with this stuff is kinda complicated... or not? I do have a multimeter though...
Thanks again, I will report again asap.

 

The AC adapter green light shutting off happens usually for two reasons:

1) Internal OCP/OVP protection--often related to GPU shorting/failure as IIRC, the CPU and GPU power come from the same source (unlike desktops)
2) AC adapter overload protection.

You said you have TWO AC adapters (this is an SLI system?) and both of them shut down at the exact same time? Or only one of them shuts down?

 

You could do a lighter load on one adapter with SLI disabled to test each brick.

 

Hello guys... ok, well, if possibile, I am even confuser (is it proper english?) than before.

I tried to sum your suggestions up, and first I focused on the psus.

Since I am a dork, and I tried the last tests with the SLI ON but only one PSU connected, and this is the scenario when the PC shuts off after 1 minute or so, I treid to split things up.

Once undestood that, the following tests run without the PSUS to completely shuts down (the green led to go off).

I decided to believe that the psus are ok, and to investigate the gpus, so I disabled SLI from the NVIDIA Panel, and I also disabled CUDA GPU setting from 3D Settings panel in order to have only 1 GPU enabled at a time.

This are the results:

Test 1: SLI Off, GPU2 enabled, only one PSU, stress only GPU in AIDA: as you can see for 5 minutes it works, the temp is high (90°) but stable. No shuts down


Test 2: SLI Off, GPU1 enabled, only one PSU (same ad before), stress only GPU in AIDA: BAM! Just a little more than 60 seconds and the PC shuts down. Not even the temperatures are interesting, from my limited point of view


I tried again Test 1 and it keeps being stable, though I am not so confident that 90° stable are safe, but my know-how on stress test is limited (I even don't know if 5 minutes are enough to draw conclusions)

Then I got bold and brash, so I decided to enable al tests on Test 1 scenario (let's call it Test 3), and than this happened:


As you can see, the pc stays on but the Overheating and CPU throttling shows up again...

I think that definitely GPU1 is faulty, and this, as I can assume now, is ONLY ONE of the problems I'm experiencing on my machine...

Do you agree?
Starting from this, what do you suggest? Do these test give you a different reading key?

I could be satisfied, in this situation (I mean, covid19 and so) to use only 1 GPU without worrying so much about the faulty one, for even a few months, but I have to be sure that the system can work without problems and this overheating is not going to damage the pc itself.

Thank you all for every kind of help that you can give me, it is much appreciated guys!

 

Could also be the VRMs on that card, worth checking they are making good contact.

Also don't run SLI on a single PSU!

 

Yeah, definitely. Running SLI on a single PSU is not going to function properly at all. If one was enough, there would be no reason for the option of two to exist.

 

Hi guys, yes, as I said, as soon I noticed that I had SLI on and only one PSU, I decided to reset all tests, and start from the beginning, and since with SLI on and both the psus connected the pc shut down after a minute, I decided to switch off the SLI and test one gpu at a time, which is the basic scenario of the tests in my previous post. I added "SLI off" at the beginning of the descriptions of the tests for clarity, since I assume you both thought I was testing with the SLI on, which is not true.
Thanks!

 

I am investigating the CPU now. The reseller of the laptop suggested to investigare with Intel XTU software. so I installed it and launched a basic benchmark, which resulted in 2085 Marks.

Here my system:


And here the advanced tuning page:


Power Limit Throttling and Current/EDP Limit Throttling keeps flashing in yellow even doing nothing at all, and I mean running nothing except chrome browser.

Dunno, maybe this can help you to understand better the scenario.

I'm gonna try to learn something about this stuff right now.

Thanks guys!

 

There will always be a reason why the CPU is not running faster most concern is around the temps under load.

 

I guess so. I will try to disassemble the gpus and the cpu, clean, repaste and do some testing.
I already took off the slave gpu from the SLI. In a couple of days I will receive some stuff I ordered from amazon and I will try to do this surgery, hoping for the best. I will report asap.
Thanks again guys.

 

You know both cards are the same right? You can just swap them to test if you like.

 

Yes @Meaker@Sager , that's indeed one of the first things I am gonna do.
At the moment I removed tha slave GPU (which unfortunately is the one that seems to work), I cleaned the mess with thermal past that was on it (al the resistors and capacitors surrounding the processor where full of paste), and I bought some new thermal metal compound (Silver King) and thermal pads (Gelid, 12W/mK). I am waiting for some Scotch 88 electrical tape to proceed with the disassemble of the faulty master gpu, replace with the cleaned one, and since I am already there I want to check out the cpu state and see if it needs some work on it.

 

So long as its not conductive the paste wont be an issue.

 

Thanks fot the info. One question: the heatsink are made of what metal? I know I cannot apply the Silver King paste on aluminum heatsinks, and I am pretty sure mines are not made of it, but I cannot find the exact material anywhere in the net...
Thanks!

 

Bare copper, so liquid metals will alloy over time and need a top up after a while. The dual 1080 machines came with a vapor chamber as default (which is also copper).

 

Hello everybody. It took me some days to experiment but maybe I found out which is the most stable configuration possible so far.
I have to say at first that all your suggestions were really appreciated and I think I can say I learned a few new things, so thank you all.

The things I did so far:

• clean the heatsink
  
• remove the (maybe) faulty master GPU and moved the slave one to its slot
• repaste the GPU with noctua nt-h1 thermal paste
• replace some thermal pads for the surrounding circuits with a few Adwits 6W/mK pads

I tested and the gpu, on stress test, reaches 78 in 3 minutes, and it seems stable, while on game with MSI afterburner it seems a little below that (76-77° yestarday playing 40 minutes on Everspace).
Before this I reached 94-95 easily.

This morning I decided to work on the CPU, so:

• clean the heatsink
  
• repaste the cpu with liquid metal compound (Silver King)
• replace the thermal pad on the surrounding circuits with Gelid 12W/mK

Testing it, the temperature on stress test for 3 minutes are below 80 degrees.
Playing at Everspace the cpu was on average below 60°.

On idle both the GPU and the CPU are below 40°.

The things I did, on the software (CCC):

• undervold the CPU to -100V
• switch the fan setting from "automatic" to "overclock"
• switch from "performance" to "fun" setting (I don't know if it is the right translation in english, since mine is in italian)

I guess the last one, in some way forces the hardware not to reaches its full potential, hence help not to reach high temperatures: I need a stable pc above a pc used at 100%, also because the pc itself is powerful anyway, so it is ok for me.



In the next hours/days I'm gonna test with some game for some time and after that I will give the second GPU a shot, in order to understand if I have to trash it or not.

Thanks!

 

Good luck