Solved Near Daily BSOD Since Hardware Upgrade

[Whirlwind]

So for the longest time I was on my Bloomfield i7 920 processor and with a sale on newegg for the new Haswell i7 4770k I decided to take the plunge and upgrade.

I reformatted my SSD and reinstalled everything, however ever since, I have been getting BSODs almost every day. Sometimes it is the KMODE_EXCEPTION_NOT_HANDLED message, but mostly it is the IRQL_NOT_LESS_OR_EQUAL message. It has happened in all circumstances, be it active gaming, active web browsing, or an idle state with just a web browser left open (meaning I left the computer on, went downstairs to water plants, and came back to a rebooted machine 15 minutes later, with power savings features disabled).

Processor: i7 4770k
Mobo: Asus Z87-Pro
GFX: EVGA GTX 560 2gb
PSU: Corsair TX750

All the motherboard drivers are up to date, and the motherboard released a BIOS update which I installed, to still continue to see BSODs before and after the update.

Everything worked perfectly fine before the hardware upgrade, so hopefully the dump data can shine a light on the cause. Thanks ahead of time for any help.

 

[JustATest]

Hello,

I'm not going to be the guy that provides the ultimate soultion for you; however, your plight has raised some curiosity on my part. Anyway, I looked up your specs and I see that Intel lists the i7 920 processor as being compatible with DDR3-800/1066 memory and the new Processor: i7 4770k is compatible with DDR3-1366/1600 memory. So, given that you have not mentioned anything about a memory upgrade that was coincident with your CPU upgrade, and given that your BSODs could point to a memory management problem, how did you handle the memory requirements for the new processor. Specifically, did you just overclock the old memory? Did you buy new memory and just forgot to mention it? Did you just leave the old memory running at 800/1066? I suspect that trying to use the old memory could be causing an issue for your new processor.


EDIT:
I see your mobo starts support at 1333Mhz memory so I assume all of my previous assumption do not apply

 

[Whirlwind]

The memory was the same memory, Corsair XMS DDR3-1600. I never overclocked it on my old system and both there and on this one it shows up in the BIOS as 1366. The original kit was a 3x4gb, since Bloomfield was triple channel, and so I simply only put in 2 of the sticks in the proper channel slots.

 

[JustATest]

The memory was the same memory, Corsair XMS DDR3-1600. I never overclocked it on my old system and both there and on this one it shows up in the BIOS as 1366. The original kit was a 3x4gb, since Bloomfield was triple channel, and so I simply only put in 2 of the sticks in the proper channel slots.

Yes, see my edit in previous post. So, you used slots DIMM_A2 and slots DIMM_B2 I assume?

 

[JustATest]

Just a few more thoughts and then I'll be on my way and make room for the experts:

Make sure the memory you moved is securely seated. ( I worked the Crucial forums for quite a while and I can't tell you how many time that a reseating of the memory fixed a problem)
Try running on 1 stick at a time (in slot DIMM_A2) and see if that helps. If so, repeat for all 3 sticks individually. (This will test* to see if any memory chips have been damaged by static discharge while handling) *If some work and some don't.

 

[Whirlwind]

Yes, I used DIMM A2 and B2 as those are the slots that my mobo's manual suggested for only running 2 sticks.

As for proper seating, having had a problem in the past also remedied (not BSODs, but i forget what) through reseating, i've always made sure in following builds/case moves/etc you always make sure that the memory is fully seated, push tot he click and then a couple extra pushes just to be sure. But that is something that will be done regardless because I borrowed a pair of 1600 sticks (only 4gb total ick) from my friend who is going out of town for the weekend, so I've stuck those in and am testing now.

I had hoped that someone examining my dmp files could have confirmed it, but it was my primary suspicion that using pieces of a "triple channel kit" might be the cause, enen though just logically it feels like it shouldnt, since kits aren't really tuned for each other as much as just making sure proper quality is the same for all 3.

Mixed feelings, i'm hoping that it still BSODs because that means I won't have to spend money buying new memory, but I also hope that this is the problem so that I know for sure what it is.

 

[JustATest]

Yes, I used DIMM A2 and B2 as those are the slots that my mobo's manual suggested for only running 2 sticks.

As for proper seating, having had a problem in the past also remedied (not BSODs, but i forget what) through reseating, i've always made sure in following builds/case moves/etc you always make sure that the memory is fully seated, push tot he click and then a couple extra pushes just to be sure. But that is something that will be done regardless because I borrowed a pair of 1600 sticks (only 4gb total ick) from my friend who is going out of town for the weekend, so I've stuck those in and am testing now.

I had hoped that someone examining my dmp files could have confirmed it, but it was my primary suspicion that using pieces of a "triple channel kit" might be the cause, enen though just logically it feels like it shouldnt, since kits aren't really tuned for each other as much as just making sure proper quality is the same for all 3.

Mixed feelings, i'm hoping that it still BSODs because that means I won't have to spend money buying new memory, but I also hope that this is the problem so that I know for sure what it is.

Well, I think you took a good step towards trying to resolve your issue but you also could have tried your own RAM (one module at a time) as previously suggested. There is still a slight possibility that one of those modules is defective and the only way you'll find out is to try them one at a time. Good luck and let us know how you make out.

 

[Whirlwind]

Well, I think you took a good step towards trying to resolve your issue but you also could have tried your own RAM (one module at a time) as previously suggested. There is still a slight possibility that one of those modules is defective and the only way you'll find out is to try them one at a time. Good luck and let us know how you make out.

This is true and was the intended plan for after I return my friend's RAM in 4-5 days.

Since my BSOD problems are random in nature and not repeatable through actionable steps, it faster in ruling out RAM itself as the source by using this set first, since it is known to be stable in my friend's machine and as said is intended as a dual channel set instead of my own pieces. Thus if this proves to not BSOD in the 4-5 days i'll have it, that means I can "safely" assume that 1 or more of my sticks is bad (or they're just not liking the dual channel) and run individual chip tests and/or just buy new sticks, thus saving the time determining the problem that I would otherwise spend wondering and waiting still while running 1 stick at a time waiting for another BSOD (naturally testing each of the 3).

On top of that, it could simply be (like my assumptions) that because it was originally a 3 piece kit, perhaps it is them memory wanting to hand the next assignment off onto a stick not present, thus faulting in channel modes but not in solo. That statement isn't backed by actual knowledge of how memory works, merely a conjecture based on the possible ways/reasons for a kit to be different from solo sticks.

 

[JustATest]

I understand you logic and can't disagree with you plan based on the time (you think) it would take to determine if an individual module is defective. However, any of your three sticks should work perfectly in pairs. The only physical aspect that makes them more compatible in a tri-arrangement is the fact they came from the same manufacturing batch number and all three modules should have the same manufacturing date. These are sold and identified with these specifics because operational characteristics can change (very minimally) from manufacturing batch to batch so this would be your best chance for identical triplets; however, take one away, and you'll end up with identical twins, and your PC should never know that one has been orphaned. Good luck and let us know.

 

[Whirlwind]

That's exactly what I always thought, and why I didn't want to comment on it in my initial post since I didn't know if it was actually that way or not, and wanted my dmp files to get looked through without the "oh look he's broken up a channel kit, thats the problem case solved" response.

 

[JustATest]

That's exactly what I always thought, and why I didn't want to comment on it in my initial post since I didn't know if it was actually that way or not, and wanted my dmp files to get looked through without the "oh look he's broken up a channel kit, thats the problem case solved" response.

I think you are doing a great job of explaining your issue in understandable detail and I like your thinking ahead mindset as well.

 

[IML8]

For some people, Windows 8 does not play well with the new Haswell C states and have reported that disabling C states in the BIOS has fixed their problems.

I don't know if this would apply to your situation, but it might be worth a try.

 

[mook663]

I 2nd the C State issue. I built a new pc the other day with a 4770k using a Gigabyte Z87X-UD4H + vid card/ram from old pc and had nothing but random bsod issues after installing a fresh copy of Windows 8.

I had zipped up all the info/dumps to makes a post on this board for some help, but I read about the PSU/C state issues with haswell chips just before I did. Disabled C1e, C3, C6/C7 in the bios and I haven't had a single bsod since. Initially tried just disabling C6/C7, but still had the random bsods happening so I disabled the rest.

If you are still having this issue, I would highly suggest trying to disable those and see if it helps you out!

 

[Whirlwind]

That is good to know, especially with this not being a laptop, so the power saving c state function isn't really a giant deal. Its kinda weird loading up asus's ai suite to see the realtime clocks/voltage/temps and seeing any of the cores at 0.08v even if it is only for a few seconds at a time during low activity things like just web browsing.

So far 2 days no crash with the other memory.

Edit: So far day 3 no crash. Ran my first 4gb stick on my friend's system that i got the current ram from for 7 hrs on memtest, 0 errors.

In the process of running the 2nd stick i was using for its own 7 hr session, though I probably don't need to, as it has resulted in 2 errors out of 3 passes (currently halfway through pass 4), both in similar address locations (first 6 digits out of 11 digit label are the same), on the same test #, #8, and the same Err-Bits label.

Prob just give that stick a 3 hr run to totally make sure, and then pop the 3rd stick in. I'll then pop the non errored sticks (at least 1 so far) back in my system and give that a run.

 

[Whirlwind]

Well it hadn't crashed in the 4-5 days I had the friend's memory in my system, and my 3rd memory stick came back with 0 errors after its own 11 hr session (stick #2 finished out at 4 errors in 9 passes). Put the 2 sticks that had no errors (previously I had the errord stick in when i was getting BSODs) in my main system and am now hoping that I get no more BSODs.....I'll update if somethign happens, and if it doesn't in a week I guess i'll mark as solved.

 

[Whirlwind]

Well its been a week since I put the 2 non-error sticks in and havn't gotten a single blue screen. The only changes to the system were that 1 stick and an update to Asus AI Suite that came out while I was memtesting the sticks. Since I didn't get errors with the other memory before the Suite update, I'm guessing that my problems were completely caused by bad sectors of memory and that I likely had that bad stick in the 1st slot, so it was getting accessed and filled more readily.

Thanks for the initial help to those that provided replies.