BSOD after adding memory - ntoskrnl.exe+7efc0

J

JustinSane

2[H]4U
Joined
Jun 29, 2005
Messages
2,425
I had one 2gig stick of G.Skill ram in a machine I'm working on. Got some new ram today 2x4 gig sticks of G.Skill and now I'm getting BSOD randomly. ntoskrnl.exe+7efc0 was in the error report. I have no way of running memtest86 but the memory did pass window's memory diagnostic.

i5 3750k
ASRock B75 Pro3
Corsair 750
G.Skill DDR3 1600 2x4gig

It ran for a solid 8 hours before any of this stuff started happening. Any ideas on what is going on?
 
JustinSane said:
I had one 2gig stick of G.Skill ram in a machine I'm working on. Got some new ram today 2x4 gig sticks of G.Skill and now I'm getting BSOD randomly. ntoskrnl.exe+7efc0 was in the error report. I have no way of running memtest86 but the memory did pass window's memory diagnostic.

i5 3750k
ASRock B75 Pro3
Corsair 750
G.Skill DDR3 1600 2x4gig

It ran for a solid 8 hours before any of this stuff started happening. Any ideas on what is going on?
Click to expand...

might need to adjust timings and/or voltage
 
If you can't run memtest86, run HCI Memtest. IMO it's superior to memtest86.
 
Eliminate the first potential point of failure by trying the sticks in the two other currently unused slots. If the error goes away, then you can narrow it down to a hardware problem with one of the slots.

If the problem persists it could very well be a board incompatibility issue. Most boardmakers have a "recommended RAM list" for their boards - check to see that your G.Skill model # is on there, or at the very least, figure out what kind of chip the RAM uses to see if it checks out that way.
 
Also Id like to add test the DIMMs individually rather than together whether you run memtest86 or hci memtest
 
larrymoencurly said:
Does that mean Memtest86 has been reporting a lot of false positives?
Click to expand...

From my experience (at work and home on dozens of machines) it takes memtest86+ much longer to find memory bad ram than hclmemtest even though technically memtest86+ is a superior program.
 
Hmmm. Did some tests with 1 stick of ram. Ran great for 6 hours but after sitting idle for a little bit the computer locked up and now it wont post. About to loose my mind with this mobo.
 
Ran great for 6 hours
Click to expand...

Remember that 6 hours may not be enough. At work have seen cases were it took memtest86+ 72 hours to find a problem. The issue happens when each cell of the ram works correctly most of the time however there are one or more marginal cells that exhibit bit flips. In these cases from my experience running multiple instances of hclmemtest ( so that you have nearly 100% of your ram being tested) seems to catch the problem a lot sooner.
 
drescherjm said:
Remember that 6 hours may not be enough. At work have seen cases were it took memtest86+ 72 hours to find a problem. The issue happens when each cell of the ram works correctly most of the time however there are one or more marginal cells that exhibit bit flips. In these cases from my experience running multiple instances of hclmemtest ( so that you have nearly 100% of your ram being tested) seems to catch the problem a lot sooner.
Click to expand...

Gonna try HCI after a reformat. I tried 1 stick and it got to load windows then BSOD. It shut off and started back up fine. So random and weird, gonna give it a reformat.

Here were the errors I was getting.
MEMORY_MANAGEMENT ntoskrnl.exe
PAGE_FAULT_IN_NONPAGED_AREA ntoskrnl.exe
 
You might need to clear the CMOS settings via the jumper or by removing the battery for a while.

Sometimes when switching out RAM, the motherboard will not pick up all timings correctly (not just the main ones - the sub-timings can affect stability as well).

I have had theis happen on more than one occasion, and it kinda sounds like that what it is.

If that doesn't fix it, I would try upping the RAM and/or IMC voltage a tad.
 
cyclone3d said:
You might need to clear the CMOS settings via the jumper or by removing the battery for a while.

Sometimes when switching out RAM, the motherboard will not pick up all timings correctly (not just the main ones - the sub-timings can affect stability as well).

I have had theis happen on more than one occasion, and it kinda sounds like that what it is.

If that doesn't fix it, I would try upping the RAM and/or IMC voltage a tad.
Click to expand...

Cleared CMOS earlier and no luck :(. If the RAM voltage is at 1.585 right now what is safe to try next?

Also there is a Windows.old folder in the C:/. I don't believe the drive was formatted fully when this copy of windows was installed. Could that be causing a lot of these issues?
 
Was a BIOS update done on this ASRock board? It's something I recommend doing to iron out any instability problems common to ASRock boards when they are first released. I'd also test the board with RAM that doesn't require any voltage tweaks or manual timing adjustments.
 
JustinSane said:
Cleared CMOS earlier and no luck :(. If the RAM voltage is at 1.585 right now what is safe to try next?

Also there is a Windows.old folder in the C:/. I don't believe the drive was formatted fully when this copy of windows was installed. Could that be causing a lot of these issues?
Click to expand...

Was this install a upgrade? I believe the windows.old is generated when you upgrade windows....
 
jmilcher said:
Was this install a upgrade? I believe the windows.old is generated when you upgrade windows....
Click to expand...

Negative. The HDD has an old Windows 7 install on it when it was using an older board. Windows 7 was reinstalled when the new CPU and mobo were put in.
 
windows.old is generated if you do a new install on the same drive that has windows installed. This way the previous windows is renamed to make way for the new install with the same folder names. Also in this windows.old I believe profiles and program files get moved depending on what version of windows you are installing.
 
NiHiL8 said:
Was a BIOS update done on this ASRock board? It's something I recommend doing to iron out any instability problems common to ASRock boards when they are first released. I'd also test the board with RAM that doesn't require any voltage tweaks or manual timing adjustments.
Click to expand...

It's using 1.4 right now. There are 2 new updates you think I should try them?

1.5
1. Support Windows 8.
2. Support write OA 3.0 key by afudos/afuwin.
3. Update new CPU code.
4. Improve discrete VGA card vBIOS compatibility.

1.6
1. Support UEFI PXE boot.
2. Improve monitor compatibility.
3. Update CPU code.
4. Improve discrete VGA card vBIOS compatibility.
5. Auto detects PCIE link speed.

http://www.asrock.com/mb/Intel/B75 Pro3/?cat=Download&os=BIOS
 
System will not POST at all with 2 sticks in the mobo now. It did this with another pair of RAM as well. Board number 2 is bad? That's crazy. :(

Why would it boot for a day with 2 sticks then give me this? So weird. So tired of this. God help meh.

EDIT: Alright I'm going crazy off no sleep but hear me out. I put 1 stick in. No boot. Put the other stick in and it boots. Did I really just get bad RAM again or is this board KILLING sticks of RAM? Is that possible?
 
Last edited:
You may have just got bad RAM, but the BIOS update should be done first. ASRock will list only the major changes done to the new BIOS, but it can fix other issues as well.
 
Tsumi said:
If you can't run memtest86, run HCI Memtest. IMO it's superior to memtest86.
Click to expand...
larrymoencurly said:
Does that mean Memtest86 has been reporting a lot of false positives?
Click to expand...
drescherjm said:
From my experience (at work and home on dozens of machines) it takes memtest86+ much longer to find memory bad ram than hclmemtest even though technically memtest86+ is a superior program.
Click to expand...
I've never seen HCImemtest report an error with RAM that failed MemTest86 or Gold Memory 5.07. I've also had no luck with MemTest86+ or newer versions of Gold Memory with any memory that passed the slow boot-up RAM test.

JustinSane:

Why can't you run any memory diagnostic except the one included with Windows? MemTest86+ can run from a USB flash drive, MemTest86 can be done to do that if configured from Linux.

I'd be very reluctant to update the BIOS except with a 100% stable system. G.Skill memory isn't the most trustworthy, and neither is anything with heatsinks on it because almost always it's made with no-name or overclocked chips. The safest choices are Samsung or no-heatsink Crucial.
 
I have seen HCImemtest report errors where memtest86+ didn't report anything.
 
larrymoencurly said:
I've never seen HCImemtest report an error with RAM that failed MemTest86 or Gold Memory 5.07. I've also had no luck with MemTest86+ or newer versions of Gold Memory with any memory that passed the slow boot-up RAM test.

JustinSane:

Why can't you run any memory diagnostic except the one included with Windows? MemTest86+ can run from a USB flash drive, MemTest86 can be done to do that if configured from Linux.

I'd be very reluctant to update the BIOS except with a 100% stable system. G.Skill memory isn't the most trustworthy, and neither is anything with heatsinks on it because almost always it's made with no-name or overclocked chips. The safest choices are Samsung or no-heatsink Crucial.
Click to expand...

Yea a better memory test (memtest86 or HCI) will have to be done. It just sucks now that one of the sticks wont even let the computer boot. This happened to the last pair of RAM we used as well. So I'm wondering if the board is killing the sticks somehow or if we really are that unlucky.
 
JustinSane said:
Yea a better memory test (memtest86 or HCI) will have to be done. It just sucks now that one of the sticks wont even let the computer boot. This happened to the last pair of RAM we used as well. So I'm wondering if the board is killing the sticks somehow or if we really are that unlucky.
Click to expand...
I think the G.Skill is simply defective, which isn't surprising for heatsinked RAM. Is there any way you can borrow a Samsung module to rule out memory problems? Because it's always made with prime chips that aren't overclocked.
 
Shouldn't we get some sort of POST with no memory in the board at all? Took all RAM out and it just sits at no signal on the monitor.

No POST whatsoever now. No RAM, 3 different RAM sticks in ALL slots on the board. It powers on, fans spin, powers off, endless cycle. Reset CMOS, nothing. Time to RMA the board?

Board is boxed up. RMA time. Any new board recommendations? :(
 
Last edited:
Shouldn't we get some sort of POST with no memory in the board at all? Took all RAM out and it just sits at no signal on the monitor.
Click to expand...

No. You will not generally get any kind of post with no ram. You may get beeps but no post.
 
drescherjm said:
No. You will not generally get any kind of post with no ram. You may get beeps but no post.
Click to expand...

Oh weird. In any case it stopped booting completely no matter what RAM was used and no matter what slot. So RMA time for this board. Has anyone else had that happen where a board worked for a week then stopped working like that?
 
Have you tried just using one stick at a time to see if anything changes?

Try one stick in all slots, then the other stick in all slots. This should help you narrow down 1 bad stick or 1 bad slot.
 
Shadohh said:
Have you tried just using one stick at a time to see if anything changes?

Try one stick in all slots, then the other stick in all slots. This should help you narrow down 1 bad stick or 1 bad slot.
Click to expand...

Yup tried 3 different sticks in all slots. No POST at all after CMOS clears and everything. :(
 
Last resort: unplug AC power, remove CMOS battery, short CMOS jumper, short together the terminals of the empty battery holder (NOT the terminals of the battery!). Try powering up with the battery removed because that sometimes forces turn-on, even if you don't push the front power button. OTOH some mobos won't turn on unless a battery is installed.
 
either a bad i5 or it got shorted out right away. or else it's the RAM itself. you can try single channel as well.
 
You must log in or register to reply here.
Back
Top