Fluctuating GPU temps?

[Gymrat76]

Noticed something strange last night. When my GPU is idle, the temps reported is usually around 34-36C. Yesterday when I rebooted, the temps were in the 60s when I was at desktop (0% load on Afterburner). However the output from the radiator fan was cool, suggesting that it was not really at 60C because when I'm gaming and the temps show 60C there is usually hot air blowing out from the radiator.

I then ran Kombustor Lakes stress test and the temps shot up to 99C, and the radiator fans went to max. Again the output was just slightly warm. After about a minute I turned Kombustor off and let the desktop idle for 30 mins and the temps went back to 34C.

What could be the problem? System is not overclocked, neither CPU nor GPU.

One thing which I did prior to this was to enable XMP in BIOS but I noticed some system instability so I reset back to default. I also have issues when rebooting, the system sometimes doesn't reboot back into the BIOS screen but hangs with the DRAM light on the Motherboard. If have to reset it hold down the power button to turn off and on before it starts normally.

Could a faulty RAM or mobo cause the funny temp readings it more likely to be the GPU?

 

[jedi5diah]

try this first
http://openhardwaremonitor.org/

if persist then
unplug ur gpu n put back

 

[wacko]

if you did not enable XMP, are you running your RAM as 2133MHz? or did you set the timings manually? if you enable XMP, you may need to adjust the DRAM voltage manually to 1.35v, and possible also adjust SA and IO voltage to about 1.12v.

your GPU temp issue sounds like a thermal sensor glitch, or worse, a pump issue.

 

[lcheowl]

When you install your win 10, did you turn on the xmp?

Did you test your ram using memtest86+?

At 99 degree, the air should be very warm, try another temp monitoring s/w.

 

[Gymrat76]

THanks guys.

I reseated the card just now and rebooted the PC.

This is what I typically see when idle:

When I ran the Lakes stress test this time, the temps were well under control:

Will try and see if I can replicate the problem yesterday. Scared me when I saw the GPU temp hit 99C and load 100%

 

[Gymrat76]

if you did not enable XMP, are you running your RAM as 2133MHz? or did you set the timings manually? if you enable XMP, you may need to adjust the DRAM voltage manually to 1.35v, and possible also adjust SA and IO voltage to about 1.12v.

your GPU temp issue sounds like a thermal sensor glitch, or worse, a pump issue.

This is my RAM with default settings, XMP disabled in BIOS:

These were the XMP-enabled settings, including voltage to 1.35v

 

[Gymrat76]

When you install your win 10, did you turn on the xmp?

Did you test your ram using memtest86+?

At 99 degree, the air should be very warm, try another temp monitoring s/w.

Windows was installed without XMP enabled. I only enabled XMP later on to try.

Haven't tried memtest86+ yet. Will be bringing the RAM down to Corbell to test later just to be sure.

 

[lcheowl]

Windows was installed without XMP enabled. I only enabled XMP later on to try.

Haven't tried memtest86+ yet. Will be bringing the RAM down to Corbell to test later just to be sure.

Since you going to Corbell, check with them that will there be any issue to enable xmp after you finished install win 10.

I was told that especially on MSI mobo, if using oc ram, best is to enable the xmp before you install win.

 

[Gymrat76]

Since you going to Corbell, check with them that will there be any issue to enable xmp after you finished install win 10.

I was told that especially on MSI mobo, if using oc ram, best is to enable the xmp before you install win.

Actually Corbell just told me to wait for the next version of BIOS to try, so saves me the trip down. However, I'm still having the issue with restarting and hanging with the DRAM light on, and only when I turn off and on the system will it boot normally in BIOS. This happens 3/5 times I restart my PC

On the other hand, the GPU temps seem to be behaving normally now. Tried 5 mins of Witcher 3 (ambient temp 30-31) following by 2-3 mins of idle to see how fast the temps come down, looks OK.

An interesting comment on the XMP when installing Windows 10. Unfortunately, my Win 10 is an upgrade from Win 7, so I've had to install 7, followed by the upgrade to 10.

 

[Op_Valkyrie]

Actually Corbell just told me to wait for the next version of BIOS to try, so saves me the trip down. However, I'm still having the issue with restarting and hanging with the DRAM light on, and only when I turn off and on the system will it boot normally in BIOS. This happens 3/5 times I restart my PC

On the other hand, the GPU temps seem to be behaving normally now. Tried 5 mins of Witcher 3 (ambient temp 30-31) following by 2-3 mins of idle to see how fast the temps come down, looks OK.

An interesting comment on the XMP when installing Windows 10. Unfortunately, my Win 10 is an upgrade from Win 7, so I've had to install 7, followed by the upgrade to 10.

Bro not true with the latest build of windows 10.... Clean install of full windows 10 directly...

http://m.winsupersite.com/windows-10/windows-10-november-update-activate-os-using-your-windows-7-and-8-product-keys#slide-0-field_images-61741

 

[lcheowl]

Actually Corbell just told me to wait for the next version of BIOS to try, so saves me the trip down. However, I'm still having the issue with restarting and hanging with the DRAM light on, and only when I turn off and on the system will it boot normally in BIOS. This happens 3/5 times I restart my PC

On the other hand, the GPU temps seem to be behaving normally now. Tried 5 mins of Witcher 3 (ambient temp 30-31) following by 2-3 mins of idle to see how fast the temps come down, looks OK.

An interesting comment on the XMP when installing Windows 10. Unfortunately, my Win 10 is an upgrade from Win 7, so I've had to install 7, followed by the upgrade to 10.

You try to disable your xmp and see whether you still have the rebooting issue.

I was told that XMP is not plug and play so changing the memory timing may cause problem on certain mobo.

Asus mobo is more friendly but there are incidents that in order enable xmp on MSI mobo, they re-install the os.

I'm also eager to know whether it is true.

 

[Gymrat76]

XMP has bee disabled. The problem still there

 

[lcheowl]

XMP has bee disabled. The problem still there

oh, look like the bios cannot detect your memory.

you go download memtest86+ and run overnight, or at least 1 pass. (It will take around 3 hours to run one cycle.) to check your memory.

 

[wacko]

This is my RAM with default settings, XMP disabled in BIOS:

http://i.imgur.com/WBHgLdC.jpg[IMG]

These were the XMP-enabled settings, including voltage to 1.35v

[IMG]http://i.imgur.com/eEFz5Lo.jpg[IMG][/QUOTE]

you are running your RAM as 2133MHz, which is a bit of a waste since they are rated for 3000MHz. you can try adjusting your "CPU SA Voltage" and "CPU IO Voltage" to 1.12v. these settings can be found under Advanced Mode, under OC section. if 1.12v doesn't work, you can try going up to 1.2v, should still be safe.

high speed DDR4 stability on the Skylake platform is a lot about VCCSA and VCCIO voltages.

 

[Gymrat76]

oh, look like the bios cannot detect your memory.

you go download memtest86+ and run overnight, or at least 1 pass. (It will take around 3 hours to run one cycle.) to check your memory.

Will run that tonight and see, thanks

 

[Gymrat76]

you are running your RAM as 2133MHz, which is a bit of a waste since they are rated for 3000MHz. you can try adjusting your "CPU SA Voltage" and "CPU IO Voltage" to 1.12v. these settings can be found under Advanced Mode, under OC section. if 1.12v doesn't work, you can try going up to 1.2v, should still be safe.

high speed DDR4 stability on the Skylake platform is a lot about VCCSA and VCCIO voltages.

I may run it a 2400MHz first and wait for the newer BIOS release. Let's see

 

[wacko]

I may run it a 2400MHz first and wait for the newer BIOS release. Let's see

chances are a BIOS update is going to set default SA and IO voltages higher for higher speed RAM, something that you can already control right now. by the way, MSI recommends running in DIMM slot 2 and 4 if you are only using 2 sticks of RAM, is that your configuration?

 

[Gymrat76]

chances are a BIOS update is going to set default SA and IO voltages higher for higher speed RAM, something that you can already control right now. by the way, MSI recommends running in DIMM slot 2 and 4 if you are only using 2 sticks of RAM, is that your configuration?

Yes, the RAM are in slots 2 and 4

 

[Gymrat76]

Completed 1 cycle (100%) of MemTest86+ with no errors. Will repeat again later.

The first time I ran it, I turned off my monitor and went to take a shower. When I got back and turned on the monitor I saw "no signal" displayed. My keyboard backlight and mouse lights were also off, though the PC was on (Mobo + GPU lights etc. all on). Same as when I reboot and it hangs. Weird.

Then I ran MemTest86+ a second time and it completed without errors.

Last night had 2-3 hours of gaming and everthing worked well, no weird temps etc.

 

[lcheowl]

Completed 1 cycle (100%) of MemTest86+ with no errors. Will repeat again later.

The first time I ran it, I turned off my monitor and went to take a shower. When I got back and turned on the monitor I saw "no signal" displayed. My keyboard backlight and mouse lights were also off, though the PC was on (Mobo + GPU lights etc. all on). Same as when I reboot and it hangs. Weird.

Then I ran MemTest86+ a second time and it completed without errors.

Last night had 2-3 hours of gaming and everthing worked well, no weird temps etc.

so nothing is wrong with the ram, likely is the bios issue.

while waiting for the new bios, you may give a last try by clearing the CMOS and set everything to default.