P67A-UD5-B3 Sleep issue

[C3D76]

Hi all,

Just got the board, I like it so far but am limited by an S3 sleep issue for my 24/7 OC.
So far, everything points to "CPU PLL Overvoltage" setting in BIOS not sticking on S3 resume.

For example, I have tested "stable" (1 hour Prime Blend test):
Stock.......
4.4Ghz: 44x Multi, PLL OverV disabled/Auto, other settings identical to x45 multi
4.5Ghz: 45x Multi, PLL OverV enabled, other settings identical to x44 multi
Temps are fine, and load voltage around 1.32v @4.5.

Sleep issue:
Stock = Sleep OK
44x with Pll OverV OFF = Sleep OK
44x with Pll OverV on = Sleep Problem
45x with Pll OverV on = Sleep Problem

So it goes to sleep fine, but on resume fans spin & there is no vga.
Then:
- 10 seconds later it reboots
- after reboot every 2 seconds it goes into an endless reboot cycle.
- I Power off PSU, then reset Bios.
- Enter Bios to set any of my OC's or Optimized defaults (even OC with OverV=on does not matter)
- Reboot, and system just resumes from sleep normally in a couple of seconds as if nothing happened (10 minutes later...)


Like I said, I think this is related to the "CPU PLL Overvoltage" setting in the BIOS not sticking on resume. The board will NOT boot above 44X Multi if this setting is not enabled, but S3 sleep only works if I disable it .


Anyone here experience similar problems? I tried tweaking voltages to be able to boot with the 45x Multi OverV= off, but no luck. Any solutions until they release a new BIOS?


Cheers

EDIT: Tested with BIOS F2 and F3C

[Lsdmeasap]

This is a known issue, CPU PLL is the cause you are correct.

You'll have to wait for future BIOS updates for this to be fixed.

And it is your CPU itself that limits what multi the board will boot with when PLL is enabled/disabled, not the board. Just FYI there.

[C3D76]

Hi LSD, long time man :)

Thanks for the info, I had no idea this was a known issue... still new to this board.

I'll just wait for a new BIOS for my 24/7 OC. This issue only affects sleep above 44x, but the board is otherwise completely stable and functional.

Yes, it makes sense that it would be the CPU since the board doesn't need to be OC'd with the CPU's unlocked multi.
I just had a doubt when I saw this GIGABYTE P67A UD5 + 2600k ....new platform, lot of fun - Page 2 - XtremeSystems Forums
And I thought "44x multi coincidence?"
I am also not used to being able to boot and stress test so easily at 4.4, and then have a wall where even booting is impossible at 4.5. I would have thought it could at least boot, and then crash in windows or during OS loading or something..

Anyway, all is good. 4.4Ghz with almost default settings is very acceptable :)

Cheers,

Ced

[Lsdmeasap]

Sup man!

No problem, and sorry if I sounded short or rude (Didn't meant to at all), I was in a hurry when I replied about all of that.

I can boot x48 without CPU PLL, some CPU's can't even boot at x44 without it, so all the CPU's are very different and some can even do x49 without it. So if yours stops at x44 without it, and you need S3 sleep, you'll have to stick with 4.4Ghz for now until they can fix that in the BIOS, if they can.

[C3D76]

Hey, no worries. I did not take it like that at all + your input is always appreciated.

See you around the forum :)

[Lsdmeasap]

Cool to know, thanks man!

Post back anytime, directly meaning don't be a stranger and post back often

[dilburt]

You didn't mention your RAM settings but this sounds exactly like an issue that appears to affect most if not all X58 boards. The issue will only come into play when using RAM that is overclocked in a configuration that requires some voltage setting above default. If your RAM is on AUTO, my guess is that your multiplier and PLL OverV settings are causing the auto RAM settings to bump up the DDR voltage.

The root cause for this issue is corruption of the STR data (Suspend To RAM) due to an under-voltage condition. This is due to an apparent bug in the X58 that causes the RAM voltage to be set to 1.5v when entering S3 sleep no matter what the prior DDR voltage was set to. When the system resumes, the voltage returns to the higher voltage but of course, the damage has been done and the STR contents have been corrupted.

This is simple to verify yourself if you have a volt meter. You will also need RAM designed for 1600 mhz or higher that requires something greater than 1.5v when overclocking. Most RAM including Corsair, G.Skill, etc. is like this. Configure your RAM for overclocking at 1600+ with a voltage setting around 1.65 volts. Connect your DVM to the DDR voltage test point and start the computer. Note the reading of 1.65v. Enter S3 sleep, the voltage will drop to 1.5v. Wake the computer, the voltage will return to 1.65v.

The contents are corrupted so your mileage may vary at this point. Almost everyone says that the CAS setting has changed upon resume but other corruption also exists. I have seen reports that ASUS has fixed this in BIOS for some boards but I don't know if this is true. It seems like there should be a workaround possible in BIOS but you would think many more systems would have been fixed by now if that was the case.

- D

[C3D76]

Hi D,

Thanks a lot for the suggestions.

In my case, I doubt it is the problem because RAM is rated 1866 @1.5v and that is what I have been running it at.
I did not optimize mem timings, I just let XMP do it's job for now (manual voltages and mem Multi though). So that could def be a problem, but so far @ 4.4Ghz for almost a week without restarting and S3 has been flawless (Stress tests + heavy use between Sleep, PLL OverV off)

I recently sold my GAX58 and S3 was hit or miss until F4/F5 Bios. I had OCZ 1600 7-7-7-24 @ 1.65v, and what you are describing could possibly have been the problem with early BIOSes.
I definitely did see strange things happening with the timings, they would show as 8-7-7 instead of 7-7-7. But I think I remember reading on this forum that the timings were actually not being changed, but just reported wrong in Everest/CPU-z/....

And I read my first post again, it is not that clear. It sounded like I was blaming the board for not booting @ 4.5 without Pll OverV, my problem was the board not waking up with that setting enabled.
I actually did not think of it, but I will test S3 with no OC at all, and Pll Overv = On. I'm curious as to what will happen... Will it wake up? I'll have to try later...

Ced

[wazza300]

My board was the same when it first came out but s3 was fixed a few bios updates later,with so much control being integrated into the cpu on sandybridge it could be down to actual cpu or related voltages or you just need to wait for current bios's to mature

[C3D76]

Hey wazza. Holy post count man, you've been busy :)

Stock with Pll OverV on = Sleep OK

I wasn't sure if it was going to work...
After reading a whole lot more on this, the issue is supposedly that with OverV on "Sleep/Hibernate will not operate properly due to Intel Specs"

When I have more time I will try and see at what multi and/or voltages sleep becomes a problem with OverV=on. (with my particular setup)

Ced