Overclock.net › Forums › Intel › Intel CPUs › Weird problem while stress testing i7 920
New Posts  All Forums:Forum Nav:

Weird problem while stress testing i7 920

post #1 of 11
Thread Starter 
Hello all,

I've been a reader of overclock.net forums for a long time but I just signed up to ask about a problem I encountered with my PC. First of all, let me give you more detal about my system:

MOBO: Gigabyte X58-UD3R
CPU: Intel Core i7 920 D0
RAM: 3x 2GB Team Xtreem 1600mhz
GPU: Gigabyte G1 Gaming GTX1070 8GB (new)
PSU: High Power 750W 80+ Bronze (new)

So, I've first noticed the problem after I recently replaced my faulty ASUS R9 290X TOP graphics card with a brand new Gigabyte G1 Gaming GTX1070. I am aware that my system components are pretty old, but I thought hey, it won't hurt to renew the parts one by one. Anyway, since the gpu is quite a beast (I am playing games on a 4K monitor, that's reason I went with this card also considering my budget), I've decided to overclock my CPU, at least to improve some potential bottlenecking.

I've done everything according to the book; seperated the cpu clock, ram modifier and bclk. First increased bclk, nudging the QPI/VTT voltage up a notch if unstable. Then adjusted ram modifier, again adjusting voltage. Finally, increased clock rate and played with vcore until system was stable @3.8ghz. Of course, I used intel burn test to stress test in every step, made it run for hours etc.

That's where the problem came up! If the system crashed to a BSOD or just froze or anything like that during the test, playing with voltages or decreasing the clock etc. fixed the issue. However, my problem is different and I've stress tested even on stock settings (2.67ghz), the problem is still there!

PROBLEM: During stress testing, the monitor goes black for a second or two, then comes back on again. It writes "DisplayPort" on the upper left of the screen, as if the signal was just received. There is no problem with image and stress testing continues, PC doesn't crash. There are no error messages, no driver crash notifications, no nothing. This continues to happen in random time intervals; sometimes twice in a minute, sometimes it just happens after 10-15 minutes. Temps are also fine for every component while this happens.

So I lose visual for 1-2 seconds, the screen goes black as if signal lost, then it comes back on and nothing seems interrupted or corrupted, the test goes on as usual. This happens only when stress testing; icluding testing at stock clocks mind you, so it's definitely not about any overclock.

I opened Gpu-Z during stress testing my CPU a couple of nights ago and saw that something weird happens during these 1-2 second blackouts. As you can see below, there are two spikes measured. The second one was a black screen (first one didn't cause anything). The voltage increases to 0.850 levels, the fan rpm spikes, power consumption increases a little, memory controller load (whatever that is) touches zero etc. etc.



I have no idea what is causing this problem, and it is worrying that it is also happening during stress testing at stock clocks. Any ideas, suggestions? I am really in need of your assistance.

Thanks.
post #2 of 11
Thread Starter 
Anyone? Have I posted in the wrong section?
post #3 of 11
Welcome to OCN! Long posts tend to intimidate members, reading and such. rolleyes.gif
Add some North Bridge voltage. IIRC you can go up to 1.25 safely. I think the quick and dirty 4.0Ghz guide tells you to set 1.2. What other voltages are you running?
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
post #4 of 11
Thread Starter 
Hi Klue22, I need to emphasize that this problem occurs even on stock cpu clock and default voltages while testing with intel burn test.




When I overclocked, other than increasing QPI/VTT and Vcore, I also increased QPI PLL to 1.2v, IOH Core to 1.2v and DRAM voltage to 1.64-1.66v. Unfortunately my PC doesn't boot if I both increase IOH Core and QPI PLL voltages above 1.2v.

But again, I have no idea why such blacking outs for 1-2 seconds would happen even when stress testing on stock settings.
post #5 of 11
What are the results of stress testing the graphics card only?
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
post #6 of 11
Thread Starter 
There are no problems while stress testing just the graphics card; running Heaven or Furmark for more than an hour without any glitches or blackouts. Same goes for playing games for hours.

Additional info which might be useful to determine the cause of the problem: My GTX1070 has Micron ram rather than Samsung, and these rams are known to encounter checkerboard effects when memory is overclocked; these vram models usually can't overclock as high as Samsung ram GTX1070 cards. This is theorized to be caused by memory voltage control in the vbios, not being able to push enough voltage to the vram in the required time.

Anyway, I am wondering if the gpu core clock and mem clock spike observed during these blackouts can cause such a blackout given the above info. Maybe it is not anything related to cpu clock or mobo, just that the increase in gpu clocks can't be feeded with enough voltage in that moment. Is this possible?

I will try locking the core and mem clocks of my gpu to their boost clocks and then run intel burn test at cpu stock settings to see if it changes anything, and of course, I'll report back immediately smile.gif
post #7 of 11
The spikes are probably just the graphics card resetting itself. Have you made sure you have the latest BIOS for your motherboard?
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
i7 970 4.15 @ ~1.39v HT ON, Turbo Off EVGA x58 3x SLI 2x EVGA GTX 980 SLI (watercooled), 1x EVGA GTX ... 6 GB 1600 OCZ DDR3 Gold Edition 7-7-7-18 @1475Mhz 
Hard DriveOptical DriveOSMonitor
Samsung 840 SSD 250GB, 2xSamsungF3 1TB (Raid0) 22x Super Multi, 8x Blu-ray Reader Windows 7 Ultimate x64 2x Yamakasi Catleap Q270s (2560x1440) 
PowerCaseMouseAudio
Kingwin 1000w Platinum HAF 932 Black Interior Logitech G500 Logitech Z5500 
  hide details  
Reply
post #8 of 11
I would manually set the CPU VTT to 1.2v, the ICH to 1.2v, and manually set the RAM to the specified voltages
     
CPUMotherboardGraphicsRAM
i7 970 between 4 - 4.4GHz Gigabyte UD7 Rev 2.0 EVGA GTX 1070 FE 2000MHz CL8.8.8.24 1T tRFC 88 
Hard DriveHard DriveCoolingOS
Samsung Evo 128Gb OCZ VertexLE 100Gb Megahalems  Windows 10 Pro 64 
MonitorKeyboardPowerCase
Asus VG248QE Corsair K70  EVGA Supernova T2 1000W Lian Li PC-A20 
MouseAudio
Sidewinder MX5 Creative Zx 
  hide details  
Reply
     
CPUMotherboardGraphicsRAM
i7 970 between 4 - 4.4GHz Gigabyte UD7 Rev 2.0 EVGA GTX 1070 FE 2000MHz CL8.8.8.24 1T tRFC 88 
Hard DriveHard DriveCoolingOS
Samsung Evo 128Gb OCZ VertexLE 100Gb Megahalems  Windows 10 Pro 64 
MonitorKeyboardPowerCase
Asus VG248QE Corsair K70  EVGA Supernova T2 1000W Lian Li PC-A20 
MouseAudio
Sidewinder MX5 Creative Zx 
  hide details  
Reply
post #9 of 11
Which 4k monitor? The u590e28?
post #10 of 11
Thread Starter 
@Klue22 Yes, it is the latest BIOS for my mobo availabe on Gigabyte's website. What would it mean if those spikes were the graphic card resetting? Would it cause blackouts such as I am experiencing?

@Gamester3333 Yes, my monitor is Samsung U28E590. Do you think its the culprit? smile.gif I never thought about it before smile.gif

@OCmember I will definitely try that, setting everything manually even on stock clocks.

Now, about the test. I tried increasing IOH core (northbridge) voltage to 1.34, still blackouts.
I tried QPI/VTT over 1.32, still blackouts.

Then I remembered that I was going to test locking the gpu voltage via afterburner. I locked the voltage to 1.05v, so the gpu didn't go idle. I started intel burn test, and amazingly, there were no blackouts. I tested this scenario with cpu stock clocks so no overclock, but comparing to previous observations, it seems locking the gpu voltage fixes the blackout problem.

So, I assume we can narrow down the problem to the graphics card? That the sudden increase in voltage (as seen in gpu-z data) when idle leads to blackout either because of Micron ram issues or that every card experiences this?

Any comments? smile.gif
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Weird problem while stress testing i7 920