[SOLVED] Old problem , just want to know

[toniplavna]

Hello guys,

I have a long problem from time to time and I will write whole story.

Short: problem is that in certain games my monitor goes black and message is check signal cable, gpu fans slow down and I need to restart.

After few years with different components, OS etc. I think its the drivers, but can be gpu or psu.

Lets start. I had old PC with : Asus p5k se, q 9300, 4gb ddr2 ram, hdd, 550 w PSU rasburo, MSI hd 6850 OC...

I remember that I had kernel power 41 error and when I got BSOD there was atikmdag.sys. I tried many thing with different drivers, OS etc. but no avail.

Problem was in GTA4, in one part of the Witcher 2... But for example I played Witcher 3 with no problems at all. Then I decided to go with GPU benchmarks and in the end with Furmark. Under them, all was stable, no blackscreen, nothing. So I think its not PSU and not GPU probably but drivers with certain games.

Now I updated my PC, I got i7 3770k, gigabyte z77mx d3h, 16gb ddr3 ram, GPU and PSU are the same. I was playing gta 4 before few days and same thing happened twice.
When Im playing Rome 2 Total war and Cities skylines, gpu is at 99% , sometimes my temps reach 80c in skylines when is hot outside but I have no problems at all.

Yesterday I installed Kingdoms come deliverance, and I get same error before I even hit the main menu. How it can be PSU, my GPU is not even under load at all. So I guess I can rule that out. So it happens only with certain games.

GPU is quite old and latest drivers are from 2015. no new ones, so I guess I cant do anything about it.

Do you know anything that I can try? I can't try spare GPU cuz I don't have one, or PSU. Only thing I can do is to try something with OS, drivers and similar thing.

 

[Colif]

PSU voltages are within range but 12V is pretty low in that range. Is at 11.52V when lower spec is 11.4. So all in all probably nothing to worry about - my concern would be that most reporting software can be quite inaccurate with voltage readings, they're just good for an indicator, so if the voltage reading is actually higher than actual, it could well be a problem.

3.3V is in range and no 5V is recorded for whatever reason.

However one thing I do notice is the GPU thermal diode shoots up to 510 degrees (which can't be right) as the fan RPM and % usage on it stops registering even though there is memory clock and fan % usage still active - not sure if this is when the crash occurs though and sensors just go a bit haywire.

open for interpretation, no 5v is interesting but then so is the temp peak on GPU. I don't want to jump to any conclusions

 

[toniplavna]

That must be error ofc, 500 celsius lol, it would cause fire... So something going on before that...

 

[toniplavna]

Can it be that sensor is bad so it shows high temp and gpu shut down to prevent damage?

But, still I can do Furmark test etc... still I can't dismiss drivers or OS...

 

[Colif]

its times like these you need a spare GPU to swap in to see if errors continue
or a spare PC to put GPU in and see if it still happens.

Can it be that sensor is bad so it shows high temp and gpu shut down to prevent damage?

Yes

hence the need for spare parts. I would perhaps get PC looked at by a repair shop, they should have spare parts for both slots and might resolve it without guessing.

 

[toniplavna]

Well, Im living in a small village no that kinda shops here only thing I can do is try to find spare GPU or PSU and try that.

Well I guess I will try to buy other gpu anyway...

 

[Darkbreeze]

PSU voltages are within range but 12V is pretty low in that range. Is at 11.52V when lower spec is 11.4. So all in all probably nothing to worry about - my concern would be that most reporting software can be quite inaccurate with voltage readings, they're just good for an indicator, so if the voltage reading is actually higher than actual, it could well be a problem.

3.3V is in range and no 5V is recorded for whatever reason.

However one thing I do notice is the GPU thermal diode shoots up to 510 degrees (which can't be right) as the fan RPM and % usage on it stops registering even though there is memory clock and fan % usage still active - not sure if this is when the crash occurs though and sensors just go a bit haywire.

open for interpretation, no 5v is interesting but then so is the temp peak on GPU. I don't want to jump to any conclusions

Where did this come from, because I looked all through this thread and I don't see that having been a post anywhere in it. Is this from a different thread?

In fact, I don't see ANY posts by PC Tailor in this thread at all.

If that IS relevant to this thread, then the PSU needs to be changed. 12v voltage at 11.52, the unit should be discarded, because if that is a software reading it's probably not fulling showing how bad it actually is and if that's a volt meter reading it's not while under a load and is actually probably much lower than that under a load.

No PSU should ever have any reason to bet at 11.6v or lower unless it is either weak or failing, and in both cases replacement would be the recommended course of action. On my own system, if I ever saw a 12v reading that was less than 11.8v, I'd warranty or replace it. I've never seen any power supply in my own machines or any client builds, that were below 11.8v even under a load unless there was an issue, regardless of whether or not that is not technically low enough to commonly be considered discard or not.

 

[Colif]

I asked him to look at them via PM

I have requested someone look through the CSV file, and see if it reveals anything. I don't know when they will reply. Hopefully soon.

He was quite able to reply but I quoted his answer from said PM

We did question whether it could be both, as that sensor temp, though improbably high, could have led to screen turning off.

 

[toniplavna]

Where did this come from, because I looked all through this thread and I don't see that having been a post anywhere in it. Is this from a different thread?

In fact, I don't see ANY posts by PC Tailor in this thread at all.

If that IS relevant to this thread, then the PSU needs to be changed. 12v voltage at 11.52, the unit should be discarded, because if that is a software reading it's probably not fulling showing how bad it actually is and if that's a volt meter reading it's not while under a load and is actually probably much lower than that under a load.

No PSU should ever have any reason to bet at 11.6v or lower unless it is either weak or failing, and in both cases replacement would be the recommended course of action. On my own system, if I ever saw a 12v reading that was less than 11.8v, I'd warranty or replace it. I've never seen any power supply in my own machines or any client builds, that were below 11.8v even under a load unless there was an issue, regardless of whether or not that is not technically low enough to commonly be considered discard or not.

Where did I post that my PSU is 11.52? Or anyone in this thread? There is 11.880V in my picture, this info isnt from my thread....

 

[toniplavna]

I asked him to look at them via PM


He was quite able to reply but I quoted his answer from said PM

We did question whether it could be both, as that sensor temp, though improbably high, could have led to screen turning off.

Ok so its PM.

 

[Colif]

That score is in the CSV file that we got from you that happened during the game test. I cannot prove it for you but if you have excel, you can find it yourself in that log file.

your PSU might be fine while not under load but during a game under load, its another story

 

[toniplavna]

Guess I will first buy some GPU below 100w, and try with that first. Second, buy some 700w no brand PSU, don't have money for anything more expensive. Will be enough for that system...

Power Supply Voltage Tolerances (ATX v2.2)

PSU Tolerance Table
Voltage Rail	Tolerance	Minimum Voltage	Maximum Voltage
+3.3VDC	± 5%	+3.135 VDC	+3.465 VDC
+5VDC	± 5%	+4.750 VDC	+5.250 VDC
+5VSB	± 5%	+4.750 VDC	+5.250 VDC
-5VDC (if used)	± 10%	-4.500 VDC	-5.500 VDC
+12VDC	± 5%	+11.400 VDC	+12.600 VDC
-12VDC	± 10%	-10.800 VDC	- 13.200 VDC

Thanks for the effort.

 

[toniplavna]

That score is in the CSV file that we got from you that happened during the game test. I cannot prove it for you but if you have excel, you can find it yourself in that log file.

your PSU might be fine while not under load but during a game under load, its another story

Still under 5% rule, but ok will change it anyway, thanks.

 

[PC Tailor]

Still under 5% rule, but ok will change it anyway, thanks.

That's the thing, it likely isn't in 5%. This is what I was saying to Colif in the quote, whilst it is within 5%,there, its low which theoretically "nothing to worry about" , software readings are notoriously inaccurate, so it's likely much worse than that, especially when under load.

This is partially also what Darkbreeze is saying, if inaccurate software is reporting voltages that low, it's a greater cause for concern as it is potentially way out of 5% under load.

The only thing completely accurate is a multimeter or direct reading from. The PSU itself, which even then won't often simulate high loads.

Then replacing it with a no name cheap PSU will likely yield the same results.

 

[Darkbreeze]

Getting a no name PSU is the absolute best way to stay in the same situation you are in now, while spending money to do it. More likely than not any no name 700w PSU will be lucky if it can do 350w and sustain it.

 

[toniplavna]

Getting a no name PSU is the absolute best way to stay in the same situation you are in now, while spending money to do it. More likely than not any no name 700w PSU will be lucky if it can do 350w and sustain it.

I never had some top gaming PC and always used average PSUs, never had a problem until now. I know, ofc it is better to have some brand one, more efficiency etc. but really can't spend any more at this point. And Im fine with 2-3 year warranty for that price.
I will first change GPU when I can, cuz probably that will fix the problem even without changing the PSU cuz as I said Im planing to get lower wattage GPU.

These are the options:
https://www.inter-tech.de/en/products/psu/atx/sl-700-plus

and
https://www.inter-tech.de/en/products/psu/atx/eps-650

Would be helpfull if someone can check these and see the spec and detail page on these products and tell which one is better.

My current spec is:

Intel i7 3770k
Gigabyte z77mx d3h
16gb 1600mhz ddr3 kingston hyper fury
1 hdd sata 7200 500gb
1 ssd 256gb A DATA
MSI Radeon hd 6850 OC 1gb
22.5 inch monitor
keyboard and mouse and thats it.

I will plan to update GPU for something that is less than 100W TDP like RX 560 4gb.

 

[toniplavna]

I got it to work, for now...

What I have done: Opened MSI Afterburner and uncheck my custom fan profile, and backed up a bit core and memory clocks.
Because this GPU is OCed by MSI its set to 820mhz and 1100mhz. I backed it up to 800 and 1050.
Default Radeon hd 6850 is 775mhz and 1000mhz memory.

I didn't unlock the vcore yet, maybe I can back it up a bit. Reference 6850 vcore is 1.094, MSI and other OCed are 1.148.

I will test a bit more and see...

 

[toniplavna]

Now Im from 1.148V to 1.040V and still no tearing or artifacts in furmark and valley... what the hell...

 

[toniplavna]

At 1.000V and still stable... Max temp in furmark dropped for more than 11 celsius...

 

[toniplavna]

Got it stable at stress test at 0.950v.

 

[toniplavna]

No crashing so far.

 

[toniplavna]

I installed Kingdom come again, and got to the main menu finally! Then black screen appears with No video input!
At least managed to the main menu xD

edit>
Back it up to 775mhz and 1000mhz and worked.