39 replies to this topic

[Flapdrol]

My brick wall is 4.9, it'll boot most the time, but just cant be stable whatever the voltage it seems.

Anyway, I have similar results on all medium, but I usually run with most stuff on low. Keep environment and textures on medium, keeps the cpu time available for the actual gameplay code instead of wasting it on particles

Edited by Flapdrol, 12 October 2014 - 04:37 AM.

[Oderint dum Metuant]

Flapdrol, on 12 October 2014 - 04:35 AM, said:

My brick wall is 4.9, it'll boot most the time, but just cant be stable whatever the voltage it seems.

Anyway, I have similar results on all medium, but I usually run with most stuff on low. Keep environment and textures on medium, keeps the cpu time available for the actual gameplay code instead of wasting it on particles

For the graphs and stuff i decided to keep it simple on all medium, i will probably even try out the borked DX11 at some point.

[Flapdrol]

I've been using dx11 for a while, seemed smoother for some reason.

[Catamount]

DV, Are you just increasing the core voltage, or are you playing with anything else? I doubt loadline calibration is your problem if it insta-locks once you hit Windows, but have you been keeping your input voltage high enough?

Have you played with Uncore at all? Most OC guides seem to say to leave it at stock and stock voltage because it does **** all for performance to OC it while messing with it can influence stability.

Edited by Catamount, 12 October 2014 - 07:21 AM.

[ninjitsu]

That FPS is pretty good. There are some AMD chips that struggle to get what you're achieving with a dual core!

[Oderint dum Metuant]

Catamount, on 12 October 2014 - 07:20 AM, said:

DV, Are you just increasing the core voltage, or are you playing with anything else? I doubt loadline calibration is your problem if it insta-locks once you hit Windows, but have you been keeping your input voltage high enough?

Have you played with Uncore at all? Most OC guides seem to say to leave it at stock and stock voltage because it does **** all for performance to OC it while messing with it can influence stability.

I'll be honest i have kept core voltage up and increasing as needed, also input voltage (VCCIN on my board) i have been boosting the uncore and voltage here too, but i may go back to stock uncore and see what happens.

I've been using this as a guide

http://www.overclock...with-statistics

Edited by DV McKenna, 12 October 2014 - 10:52 AM.

[Goose]

Flapdrol, on 12 October 2014 - 04:35 AM, said:

… I usually run with most stuff on low.

And that was the line that'd coffen-nailed the subject for me; Descent framerate at all Medium is a Revelation.

When do we get to the user.cfg set of tests?

Good work.

[Goose]

Gwad

Where did I

Is this it?

[ninjitsu]

DV, if you record your system usage data with Afterburner log to file and send me the HardwareMonitoring.hml, I can make a proper graph.

[Oderint dum Metuant]

ninjitsu, on 14 October 2014 - 09:01 AM, said:

DV, if you record your system usage data with Afterburner log to file and send me the HardwareMonitoring.hml, I can make a proper graph.

I actually turned that on today to troubleshoot, so it's gathering data atm (i'll reset it before playing MWO however)

Randomly twice in 2 days my monitor has gone black - no signal, nothing brings it back save a reboot, where it comes back straight away.
Getting back into windows, CAM tells me the GPU hit 96C, im assuming thus the monitor went black.
Problem is, both times this has happened the computer has been idling at desktop/web browsing, i've been playing some MWO/WoW/Shadows of Mordor over the last two days up to 8 hours a session (god bless days off) and the GPU never went above 49.

GPU is stock still no OC yet,need to puzzle this out first.

[Flapdrol]

I had 2 black screens with my gtx 470 when I was running arma 3 with downsampling and antialiasing. Never seen it agian though. There's an accelero extreme on it, so I don't think it was heat.

[Catamount]

I can't imagine the GPU hitting 96C at idle. OC or no something has to actually be driving it to sufficient load, and from the sounds of it, that's a lot.

Did you disable C3 and C6 by chance when you overclocked the CPU? Those can cause system crashes (including black screen lockups) during idle.

[Oderint dum Metuant]

Catamount, on 14 October 2014 - 10:13 AM, said:

I can't imagine the GPU hitting 96C at idle. OC or no something has to actually be driving it to sufficient load, and from the sounds of it, that's a lot.

Did you disable C3 and C6 by chance when you overclocked the CPU? Those can cause system crashes (including black screen lockups) during idle.

No I didn't disable C states but to rule the CPU stability out I might reset the bios to default for a day or 2 and see if it occurs again.

[Oderint dum Metuant]

Right it is a GPU area of problem.

Quote

14-10-2014 21:08:29, 101.000 ,12.000
14-10-2014 21:08:30, 100.000 ,18.000
14-10-2014 21:08:31, 101.000 ,17.000
14-10-2014 21:08:32, 101.000 ,18.000
14-10-2014 21:08:33, 101.000 ,18.000
14-10-2014 21:08:34, 101.000 ,19.000
14-10-2014 21:08:35, 101.000 ,18.000
14-10-2014 21:08:36, 101.000 ,19.000
14-10-2014 21:08:37, 101.000 ,17.000
14-10-2014 21:08:38, 101.000 ,18.000
14-10-2014 21:08:39, 101.000 ,19.000
14-10-2014 21:08:40, 101.000 ,19.000
14-10-2014 21:08:41, 102.000 ,18.000
14-10-2014 21:08:42, 101.000 ,19.000
14-10-2014 21:08:43, 101.000 ,17.000
14-10-2014 21:08:44, 102.000 ,17.000
14-10-2014 21:08:45, 102.000 ,17.000
14-10-2014 21:08:46, 102.000 ,19.000
14-10-2014 21:08:47, 102.000 ,19.000
14-10-2014 21:08:48, 102.000 ,17.000
14-10-2014 21:08:49, 102.000 ,19.000
14-10-2014 21:08:50, 102.000 ,20.000
14-10-2014 21:08:51, 102.000 ,3.000 ,
14-10-2014 21:08:52, 102.000 ,1.000
14-10-2014 21:08:53, 102.000 ,1.000
14-10-2014 21:08:54, 103.000 ,18.000

First number Temp, second column is GPU usage at the time.

However while it was doing this i left it and had a look around, it seems to the fan that keeps the VRAM and Moffs cool had stopped spinning.

The GPU pump connects to the motherboard and gets its power from there, the fan connects to a connection on the G10 bracket, i disconnected this one and plugged into a PSU splitter for power and it started spinning again.

Guess it's back to wait and see, either a dodgy power connector (but then why is it ok during an entire day or two of gaming?) or a dodgy fan.

Final option is that something is turning the fan off when it detects low GPU usage and then it's not coming back on and the temp ramps up.

(The log is 10 mins longer i cut it for ease of use.)

Edited by DV McKenna, 14 October 2014 - 12:40 PM.

[ninjitsu]

It's been a while since I ran a Nvidia gpu but with AMD I know it ramps the fan up and down based on gpu usage. Are you able to manually set fan speed in the driver?

[Oderint dum Metuant]

ninjitsu, on 14 October 2014 - 01:13 PM, said:

It's been a while since I ran a Nvidia gpu but with AMD I know it ramps the fan up and down based on gpu usage. Are you able to manually set fan speed in the driver?

Short answer not through the Nvidia drivers as it's an aux fan on the Kraken G10 bracket. But i can set it through CAM and MSI Command Centre which i already had done to run at 100% and the problem still occurred.

The liquid cooler, and the fan attached to the radiator work fine and never stop.
It's just the aux fan over the VRAM and mofs.

[Oderint dum Metuant]

Quick update

Bios at stock; played last night for about 4 hours no issues.

Left it on overnight and about 4 hours later the GPU secondary fan stopped again and the GPU overheated even with the fan plugged into the PSU via splitter.

Gonna swap the fan and uninstall CAM next

[Lord Letto]

Any Updates?

[MechWarrior4172571]

DV McKenna, on 14 October 2014 - 01:22 PM, said:

Short answer not through the Nvidia drivers as it's an aux fan on the Kraken G10 bracket. But i can set it through CAM and MSI Command Centre which i already had done to run at 100% and the problem still occurred.

The liquid cooler, and the fan attached to the radiator work fine and never stop.
It's just the aux fan over the VRAM and mofs.

DV McKenna, on 15 October 2014 - 10:54 PM, said:

Quick update

Bios at stock; played last night for about 4 hours no issues.

Left it on overnight and about 4 hours later the GPU secondary fan stopped again and the GPU overheated even with the fan plugged into the PSU via splitter.

Gonna swap the fan and uninstall CAM next

DV McKenna, on 11 October 2014 - 05:29 AM, said:

To extend my experiment i decided to replace the good Asus Direct II cooler on the GPU with an AIO liquid solution using an NZXT G10 and what should have been a compatible NZXT X40 cooler.
I say should have been, because i made a mistake with my order and ended up with an NZXT X31 instead (not compatible with the G10......at least officially.)

And here it is with my brute force bodge job G10 :



I wasn't sold this would work, i had to exert some excessive pressure on the board to get the mounting screws on because the contact plate on the X31 im guessing is thicker which left virtually no screws coming through to secure it in place.

The answer maybe closer than you think--you strained the fan's frame so that there is pressure on the motor. Sooner or later the grinding forces flags to go off and the fan and gpu shuts down to prevent overheating. I recommend you install custom long screws you might have laying around and run with them in loose configuration for the VRAM fan. See if that helps. The fan might be fine but you overtightening it might have created an anomaly. Also, you can show alot more graphs with Afterburner 4.0 than what you have, things like GPU Temperature and bus usage, etc.--just a heads up. What version of Afterburner are you running? You can right click on the display graph and choose properties in order to select some exotic measurements to be shown.

Edited by Jesus DIED for me, 09 February 2015 - 04:43 PM.

[Oderint dum Metuant]

Jesus DIED for me, on 09 February 2015 - 04:41 PM, said:

The answer maybe closer than you think--you strained the fan's frame so that there is pressure on the motor. Sooner or later the grinding forces flags to go off and the fan and gpu shuts down to prevent overheating. I recommend you install custom long screws you might have laying around and run with them in loose configuration for the VRAM fan. See if that helps. The fan might be fine but you overtightening it might have created an anomaly. Also, you can show alot more graphs with Afterburner 4.0 than what you have, things like GPU Temperature and bus usage, etc.--just a heads up. What version of Afterburner are you running? You can right click on the display graph and choose properties in order to select some exotic measurements to be shown.

That's not quite correct, the AIO is what took effort to fit as the supplied screws don't reach very well through and the contact plate is thicker.

The fan for the VRAM is entirely separate and not connected to the AIO. So the issue was unrelated.

I say was because the actual issue was power and the NZXT software. Removing the NZXT software and swapping the cable power adapters removed the issue.

If you check the original date this was done some time ago. I do have plans to update the thread probably on my next set of days of using afterburner, Fraps and the graph suite from the performance thread