My new PC is cursed!

[Mize]

I put my money on Motherboard - sorry

Either you're experiencing BIOS bug connected with CPU VRM regulation or simply VRM's are dying.

Still, changing PSU seems to be the easiest option to try first! Good luck!

The problem with this theory is that a dying VRM would crap out under load, not just under idle. As for a BIOS bug, I'm on the latest (and last) BIOS for this motherboard (Asua P67 Sabertooth)...I could downgrade as there is a procedure for this. That procedure does carry a non-zero chance of bricking.

 

[Albuquerque]

Had you seen my ninja-edit from earlier about the Intel RST drivers?

 

[Mize]

Thanks Abq, I missed your ninja edit.
I know I have the latest RST drivers, but I've not tried disabling AHCI...shouldn't really make any difference for me as I'm not running and RAID, just a couple SSDs and a couple HDs.

 

[Mize]

No dice on using stock clocks homerdog...it idle rebooted after approx 20 hrs...AHCI next

 

[Npl]

Just a curious question... why is your Pc running 20hs on idle?

 

[Mize]

Just a curious question... why is your Pc running 20hs on idle?

To figure out what is causing idle reboots. There's no other "accelerated" idle test that I'm aware of. Kinda like running Prime95 + Furmark for 24 hours as a stability test...only without Prime95 and Furmark.

Trust me, when you get one of these in the middle of something it's very frustrating. Eliminating the idle reboots is very important for me. I have a laptop I can use so that I don't lose work so much, but I vastly prefer my larger keyboard and monitors.

 

[Npl]

Well, I meant I would probably never experience the bug since I shutdown the Pc after a couple hours at most. It sounded like your issue only appears after 20+ hours, but as I understand now it can happen anytime.

 

[Mize]

Well, I meant I would probably never experience the bug since I shutdown the Pc after a couple hours at most. It sounded like your issue only appears after 20+ hours, but as I understand now it can happen anytime.

It's intermittent. Once it happens the time until the next occurrence is typically 24-48 hours but could be as little as 3 or 4. Also, I do often leave my computer running for hours at a time as the boot times aren't stellar and I'm impatient.

 

[Mize]

Well it's not AHCI...reboot this AM with SATA on IDE.

Bumped the offset a bit more (which shouldn't be the problem since it idle reboots at stock)...need to dig around for another PSU to swap that out...grrr.

 

[Lightman]

The problem with this theory is that a dying VRM would crap out under load, not just under idle. As for a BIOS bug, I'm on the latest (and last) BIOS for this motherboard (Asua P67 Sabertooth)...I could downgrade as there is a procedure for this. That procedure does carry a non-zero chance of bricking.

Not in all cases. It all depends how the VRM's are programmed to behave under little to no load. Is board shutting some of them off to save on wasted power and be more green? Is VRM frequency changed?
For instance on my MSI board I can turn off Intel recommended dynamic VRM regulation (cutting phases) and enforce old school 'all VRM's active at all time' behaviour. Low power jumps up by about 10W but all phases are powered constantly.

 

[Lightman]

The problem with this theory is that a dying VRM would crap out under load, not just under idle. As for a BIOS bug, I'm on the latest (and last) BIOS for this motherboard (Asua P67 Sabertooth)...I could downgrade as there is a procedure for this. That procedure does carry a non-zero chance of bricking.

Not in all cases. It all depends how the VRM's are programmed to behave under little to no load. Is board shutting some of them off to save on wasted power and be more green? Is VRM frequency changed?
For instance on my MSI board I can turn off Intel recommended dynamic VRM regulation (cutting phases) and enforce old school 'all VRM's active at all time' behaviour. Low power jumps up by about 10W but all phases are powered constantly.

 

[Mize]

Not in all cases. It all depends how the VRM's are programmed to behave under little to no load. Is board shutting some of them off to save on wasted power and be more green? Is VRM frequency changed?
For instance on my MSI board I can turn off Intel recommended dynamic VRM regulation (cutting phases) and enforce old school 'all VRM's active at all time' behaviour. Low power jumps up by about 10W but all phases are powered constantly.

How do you do this?
I've disabled C1E now as well which has severely limited my down-clocking rate as the machine idles and bumped voltages, but I still get some odd low voltage reporting.

 

[Lightman]

How do you do this?
I've disabled C1E now as well which has severely limited my down-clocking rate as the machine idles and bumped voltages, but I still get some odd low voltage reporting.

On my MSI there is an option in BIOS to change VRM behaviour. On top of that I have measuring points for most important voltages directly from board next to ATX connector as well as indicator LED's for each phase so I know which phase is active and for how long (usually blinks very rapidly and drops to 1 phase when idle).

 

[Mize]

On my MSI there is an option in BIOS to change VRM behaviour. On top of that I have measuring points for most important voltages directly from board next to ATX connector as well as indicator LED's for each phase so I know which phase is active and for how long (usually blinks very rapidly and drops to 1 phase when idle).

Sounds more like a Z87 board than a P67...?

For VRM settings I have Phase control (Standard, optimized, extreme or manual), Duty control (T.probe thermal or c.probe current) and then I have VRM frequency (auto or fixed frequency)...

 

[Mize]

To summarize:
Idle reboots are NOT caused by:
1. inadequate Vcore *setting* (though it might be randomly dropping)
2. AHCI (Intel RST drivers)
3. C3, C6 or C1E states
4. VRM phase control
5. heat
6. Nvidia "36hr" reboot bug (old driver set did this)
7. Load line calibration

So now I ask you to vote here for my next action:
1. replace (as in buy a new) PSU
2. replace (as in buy a new) Motherboard
3. chuck it and go Ivy Bridge since you're going to have to sink that much time anyway
4. chuck it and go Haswell since you're going to have to sink that much time anyway

I'm not overly keen on Ivy or Haswell as they run hotter (unless you delid) and don't offer many benefits other than a small bump in speed and a small drop in power consumption.

decisions decisions.

 

[Grall]

We get to decide what to do with your money...? Ideas, ideas!

Haswell is going to get replaced in another six-eight months or so, you can probably hold out until then can't you? I'd feel a bit miffed myself having to put on the big spending-pants right in the middle of a product cycle.

You really don't have access to another power supply to borrow/scavenge for a week or so? Your kids, wife, mates etc all have Dell and HP PCs, notebooks, macs, or what? I'd suggest making a genuine effort not having to splurge for another PSU purely on a guess/hunch, unless money isn't a concern at all of course (in which case, good on you!) If switching PSU doesn't make any difference and you just can't live with the crashes, consider disabling enough/all power saving to stop the misbehaviour and shutting off/sleeping your PC when you're not using it, and then replace innards when a new generation of components appear on the market.

 

[Mize]

We get to decide what to do with your money...? Ideas, ideas!

Haswell is going to get replaced in another six-eight months or so, you can probably hold out until then can't you? I'd feel a bit miffed myself having to put on the big spending-pants right in the middle of a product cycle.

You really don't have access to another power supply to borrow/scavenge for a week or so? Your kids, wife, mates etc all have Dell and HP PCs, notebooks, macs, or what? I'd suggest making a genuine effort not having to splurge for another PSU purely on a guess/hunch, unless money isn't a concern at all of course (in which case, good on you!) If switching PSU doesn't make any difference and you just can't live with the crashes, consider disabling enough/all power saving to stop the misbehaviour and shutting off/sleeping your PC when you're not using it, and then replace innards when a new generation of components appear on the market.

I have an older Thermaltake Toughpower 850W but I think I replaced it because it was acting funny after I went to 2x GTX580s and now I have 2xGTX680s. Maybe it wasn't. I dont' remember. Bummer is that the modular cables are different than my current PSU.

The other thing is whichever PSU I borrow has to support the exact hardware or it's not a good comparison...

I like the idea of waiting for another gen though. I guess I'll try the old Thermaltake and see what happens...

 

[Albuquerque]

I know for Haswell you'll probably need another PSU anyway. Some PSU's have serious problems because the idle power draw is so little that they either don't regulate correctly or simply shut off.

Given the "unclean AC input" history you've told us about, I'd target the PSU first. Get one that's Haswell compatible (yup, they exist) and go from there.

 

[homerdog]

Yes you could try a PSU that uses DC to DC conversion to power the 3.3 and 5V rails. That way there is always a significant load on the +12V rail. That prevents the PSU's undercurrent protection from kicking in at idle.

Any PSU that uses DC to DC conversion to power the secondary rails should be "Haswell Compatible". You can even search for PSUs based on Haswell compatibly on Newegg. I know you don't have a Haswell CPU but it is worth a try if you don't mind plopping down for a new PSU

 

[Lightman]

I would go Haswell just for added AVX2 stuff. I'm holding on to my i5 2500k purely because Ivy didn't get any cheaper over time, even 2nd hand and Haswell wants new motherboard for too little gain it offers. Saying that, if I had a buyer for my board + CPU then I would go S1150 or eve S2011.

BTW my board is MSI Z68A-GD65 G3.