Overclock.net › Forums › Components › Memory › Pass Memtest86+... Windows BSOD?!
New Posts  All Forums:Forum Nav:

Pass Memtest86+... Windows BSOD?!

post #1 of 7
Thread Starter 
I have the G.SKILL ECO F3-12800CL7D-4GBECO 2x2GB kit.

At 1333MHz 6-7-7-20-1T-26 it is very stable.

At 1333MHz 6-6-6-18-1T-24 I ran Memtest86+ with no errors. However, Windows booted up very oddly and slowly until it finally just gave me a BSOD. I went back to the former stable timings, and everything seems rock solid.

Why would this happen? I'm guessing maybe the memory controller can't handle such a low latency, since Memtest+ doesn't really stress it very much. Would more voltage to the CPU-NB perhaps be needed to make it stable?
post #2 of 7
more ram voltage will make it stable

increase your imc or vvt is just going to alot you to lower the timing, but doesnt give more stability.
Casmatic
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k ASRock Z77 Extreme4  PNY GeForce GTX 1080 PNY XLR8 16gb 
Hard DriveHard DriveOptical DriveCooling
samsung 850 Seagate 5tb LITE-ON Black 12X Blu-ray Scythe Mugen 5 
OSMonitorKeyboardPower
Windows 10 ASUS VE258Q TeamWolf Zhuque + (Zealios switches) Corsair HX750 
CaseMouseMouse PadAudio
Cooler Master 690 II Advance G900 Corsair - MM300 Sound Blaster E5  
Other
Sunbeam Rheobus Extreme 
  hide details  
Reply
Casmatic
(17 items)
 
  
CPUMotherboardGraphicsRAM
Intel i7 3770k ASRock Z77 Extreme4  PNY GeForce GTX 1080 PNY XLR8 16gb 
Hard DriveHard DriveOptical DriveCooling
samsung 850 Seagate 5tb LITE-ON Black 12X Blu-ray Scythe Mugen 5 
OSMonitorKeyboardPower
Windows 10 ASUS VE258Q TeamWolf Zhuque + (Zealios switches) Corsair HX750 
CaseMouseMouse PadAudio
Cooler Master 690 II Advance G900 Corsair - MM300 Sound Blaster E5  
Other
Sunbeam Rheobus Extreme 
  hide details  
Reply
post #3 of 7
Thread Starter 
Quote:
Originally Posted by lilraver018 View Post
more ram voltage will make it stable

increase your imc or vvt is just going to alot you to lower the timing, but doesnt give more stability.
But why would Memtest+ run with no errors but Windows be clearly unstable?
post #4 of 7
memtest86+ is not a stability test for overclocked memory; as you can see, the results between memtest and Windows are different. Memtest is for checking whether memory is defective or not. Try increasing the DRAM voltage also, you're good up to around 1.6V on the ECOs.
post #5 of 7
Thread Starter 
Quote:
Originally Posted by xd_1771 View Post
memtest86+ is not a stability test for overclocked memory; as you can see, the results between memtest and Windows are different. Memtest is for checking whether memory is defective or not. Try increasing the DRAM voltage also, you're good up to around 1.6V on the ECOs.
I thought Memtest was good for both.

When I first installed the memory, I set the speed and timings to the advertised spec: 1600Mhz 7-8-7-24-2T. I got a number of errors on test #5.

Because I was eager to have a working system, I decided to use JEDEC spec's for the time being and tweak at a later date. I then ran memtest and it passed three times with no errors. So it definitely does help indicate instability.

Lsdmeasap, a Gigabyte/Mushkin overclocker, says this about Memtest:

"When doing basic tests on new ram or just for general testing purposes you should let the test run on it's own for a few hours up to 24 if you have the time and don't mind.

When you are tweaking timings and just need some quick answers you can quickly see if your recent changes have made your ram unstable by choosing certain tests. Some say their own set of test #'s to use and they may differ from my own, so if you see someone else suggesting test #'s they that would be fine as well to, it is up to you.

What test #'s I use for basic speed testing to see if recent changes are causing errors quickly are #'s 2 & 5. I run test #2 for about 10 minutes and if I see no errors I will then run test #5 for a hour or so. If no errors are found then your changes are likely ok for the time being. Longer tests will let you know for sure though, but like I said here we are speed testing while doing a little bit of tweaking. If it is going to fail you think and are not so sure let #5 run for a while. It will fail on #5 if it is going to fail."
post #6 of 7
SuperPi 32M is more suitable as a memory stability test. You may need slightly more than stock 1.35V to run 1600 7-8-7.
post #7 of 7
Thread Starter 
Quote:
Originally Posted by xd_1771 View Post
SuperPi 32M is more suitable as a memory stability test. You may need slightly more than stock 1.35V to run 1600 7-8-7.
I'll check that out, thanks. I still think memtest is a good initial indicator of memory stability, but obviously it's not sufficient. Memtest is very light on the cpu, I believe, and adding processor load probably uncovers problems that memtest doesn't address.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Memory
Overclock.net › Forums › Components › Memory › Pass Memtest86+... Windows BSOD?!