Z87X-OC problems: 4 DIMMs unstable, GPU crashes...

[MBit]

Hello,

I assembled a new system a few days ago and ran into several problems, which I suspect to be mainboard related.


The rig:
Board: Gigabyte Z87X-OC 1.0
CPU: Intel i7-4770K
Memory: G.Skill TridentX F3-1600C7Q-32GTX (4x8GB, 1600 MHz, 7-8-8-24)
Graphics: 2x Nvidia GTX660Ti (SLI)
Storage: OCZ RevoDrive 3, WD Black hard disk, DVD-RW drive
PSU: BeQuiet Dark Power Pro 750W
Cooling: Air only
Misc: PCI TV card, 5,25" hard disk bay, eSATA bracket
OC: no overclocking
OS: Win7 64-bit

Bios:
Board shipped with F3, I flashed F4 through @BIOS, then realized F5 was available and flashed that.

The problems:

1)
I can't get the CPU to clock down when it is idle.
It always runs at it's turbo multipliers (x39, x38, or x37 depending on load).
Despite default BIOS settings (I used "load default settings" with all BIOS versions) I've never seen it operate at the default x35 (3500 MHz) or below.
I use CoreTemp, RealTemp and CPU-Z for frequency, temp, and load monitoring.
I also enabled all power saving features in the "Advanced CPU Core settings" page in BIOS Setup to no avail.
In a review of the board (http://www.techpowerup.com/reviews/Gigabyte/Z87X-OC/6.html) the reviewer wrote that some of these boards had 'custom Turbo profiles that boost performance at "stock" settings'.
The default settings on my board however did not leave the impression of being a 'turbo' profile.
If it helps, I can upload BIOS screenshots.
Edit: Oh, nevermind, I forgot that Windows energy saving settings play a role here, too. Coming from a pre-energysaving world  

2)
The system doesn't run stable with all 4 DIMMs installed.
Win7 crashes at random times, sometimes during boot, with random blue screens.
Memtest86 shows a lot of errors. I ran the test at default 1333MHz, with the XMP profile (1600MHz) and even at 1066 MHz.
Using any two DIMMs in any slots at any of these frequencies runs stable.
This issue seems to be widespread among Z87 boards:
http://forum.giga-byte.co.uk/index.php/topic,12563.0
http://forum.giga-byte.co.uk/index.php/topic,12635.0
http://www.overclock.net/t/1397947/bsods-maybe-caused-by-ram-and-other-problems
I'm running fine for now with 2 DIMMs installed, but that can't be the solution, right?  

3)
Sometimes during use of the graphic cards (Gaming) or afterwards (Folding@home GPU) the graphics driver crashes and in most cases gets restarted by Windows. I am using version 320.49 (latest as of now).
After reset, I'm stuck at low GPU frequencies until reboot. Overheating is not an issue.
Similar issues are reported in this thread:
http://forum.giga-byte.co.uk/index.php/topic,12833.0
It's actually this thread that made me consider the board as a cause.

4)
Does anyone know a program that can control the board's fan connectors, including the 3-pin ones, depending on multiple temperatures (cpu, gpu, hard disks) just like SpeedFan? Unfortunately the current version of SpeedFan doesn't work well with the board.


Maybe someone here can help me or has even encountered the same problems.

[bigfatron]

I've been having the '4 sticks bad, 2 sticks good' problem too.  After working through the problem i'm currently getting some joy with disabling the C-state power management options. On the D3HP BIOS at least thats disabling options for C1E, C3 & C6/C7. I have left the thermal safety and EIST options on Auto.  Its been rock solid with no BSOD incidents in a few hours of running with those settings.  So you might want to give that a try?

[MBit]

I'll definitely try it!
Expect some Memtest results in a few hours 

[MBit]

No luck here.
I've disabled those three C* options, set the mem clock to 1066 MHz and ran Memtest86 for a few hours. In test 7 and 8 the first errors showed up.
With higher clocks, they show up earlier, e.g. in test 4 or 5.
This is pretty much the same picture as I got with the C* options enabled.

Edit: bigfatron, I highly suggest running Memtest aswell. I wouldn't rely on the absence of bluescreens.

[ex58]

For 4x8GB...(32GB total) you must increase some voltages in BIOS......
Do not use XMP,set RAM manually.

Good thread to read:
http://www.overclock.net/t/1401976/the-gigabyte-z87-haswell-overclocking-oc-guide
http://www.overclock.net/t/1397684/z87x-oc-z87x-oc-force-owners-discussion-fan-club

Good luck.

[MBit]

ex58, can you tell me which voltages would need increasing?
In The Gigabyte Z87/Haswell Overclocking(OC) Guide several voltages are mentioned to support memory overclocking.

However, I'm not even overclocking the memory. Memtest showed errors on 1066MHz, 1333MHz and (XMP) 1600 MHz.
Contrary to what you said, the guide also states: "The first thing you should do is to load XMP on your module".
And: "Haswell’s IMC is more powerful than Ivy Bridge’s, has more timings that can be changed, handles higher density DIMMs better, and handles 4 DIMMs better."
... well then Ivy Bridge must have been awful at 4DIMM stability...

[ex58]

Heh,I am more practical,then theory said.
First of all, your memory is not on OC "Memory Support List" (so XMP is not supported).
Second,your memory latency is very low (7-8-8-24)......try to relax eg.8-9-9-24.
Third,try to increase DRAM V to 1.52-1.54V.
Also helps to increase Voltage a bit + (from Auto-Default) as follows:
-CPU System Agent V...(very helpfull)
-CPU I/O Analog V
-CPU I/O Digital V

Deafault voltages you can see in BIOS.

Edit:
As you are from Germany,why didn't ask JZ for advice.

[Svarog]

The GPU issue you got happens with older Drivers aswell. For now the only solution is to use F4.

It seems F5 doesn't like some NVIDIA cards.

[Svarog]

Been told today that in one of the F5 betas something with the PCI-E slot speeds had been changed.

Probably explains our GPU issues.

[MBit]

Using F4 for now, seems to work stable, had some hours of SLI gaming and F@H.
I also had a voltage setup that ran a full pass of Memtest86 without errors on 32 GB / 1333MhZ / JEDEC timings (CL9 etc.) but I lost it. Now when I try to set those voltage settings from memory again, all I get is errors. I basically did what ex58 said, set  DRAM V=1.54V, and I think SysAgentV = CPU I/O A = CPU I/O D = +0.1V but it's not stable anymore 

Anyway, the board is supposed to run the  hardware components without issues at stock settings and that's what I just wrote in a ticket to Gigabyte, maybe I get some helpful reply.

Thanks for your suggestions so far, they provided a lot of insight and help.

[Svarog]

Using F4 for now, seems to work stable, had some hours of SLI gaming and F@H.
I also had a voltage setup that ran a full pass of Memtest86 without errors on 32 GB / 1333MhZ / JEDEC timings (CL9 etc.) but I lost it. Now when I try to set those voltage settings from memory again, all I get is errors. I basically did what ex58 said, set  DRAM V=1.54V, and I think SysAgentV = CPU I/O A = CPU I/O D = +0.1V but it's not stable anymore 

Anyway, the board is supposed to run the  hardware components without issues at stock settings and that's what I just wrote in a ticket to Gigabyte, maybe I get some helpful reply.

Thanks for your suggestions so far, they provided a lot of insight and help.

The only thing that's broken for me with F4 is the Renesas USB 2.0/3.0 ports. With F4 they cause some USB devices to disappear, this doesn't happen in F5 but that doesn't work with my Videocard >_>

I'm quite bummed coming from my perfectly working X58A-UD7.

[bigfatron]

No luck here.
I've disabled those three C* options, set the mem clock to 1066 MHz and ran Memtest86 for a few hours. In test 7 and 8 the first errors showed up.
With higher clocks, they show up earlier, e.g. in test 4 or 5.
This is pretty much the same picture as I got with the C* options enabled.

Edit: bigfatron, I highly suggest running Memtest aswell. I wouldn't rely on the absence of bluescreens.

The system has been rock solid for a number of days with the C-state options off. In fact i'd even gone as far as overclocking up to 4.2GHz over the weekend without mishap.  Despite getting a system that now doesn't fall over they've asked me to try a few other settings with C-state re-enabled (presumably to see if the C-state stuff isn't the root cause but just one possible cure) which i'll report back on if its successful.

[MBit]

Update:

I'm using Bios F6 now, no problems with the graphics cards anymore, but memory errors persist, just like with F4.

The Gigabyte support team assembled a system similar to mine, but didn't find any errors.
They also mentioned the C-state options, which isn't helpful for me.
They told me to change the PSU, which I did, but it turned out not to be the cause.
They told me to test other memory, which isn't available to me at the moment.

I also removed all peripherals (pci/e cards, drives..) but they were'nt the cause either.
Yesterday, after installing F6, I successfully ran a full pass of Memtest86 with all 4 DIMMs @ XMP (1600 mhz 7-8-8-24). Then let the system run overnight under high load (f@h cpu and gpu). In the morning: BSOD. Ran Memtest again: Instantly got errors again. So it just went unstable overnight ... Does anyone besides me find that strange?
Running out of solutions soon ...
I'll wait for a new answer from support.

[madLyfe]

i am having the same issues with my Z87 UD4H board.. im running 2 of these http://www.newegg.com/Product/Product.aspx?Item=N82E16820233180 cosair kits: CMZ8GX3M2A1600C9R is the model #.. i have (2) 2x4gb kits so all 4 dimms are filled.. memory is fine in memtest and im guessing its the same issue that everyone else is having.. has anything new come from gigabyte in terms of a fix? or atleast a quick fix? im trying the whole disable the cstate power options and will reply back when done..

[bigfatron]

i am having the same issues with my Z87 UD4H board.. im running 2 of these http://www.newegg.com/Product/Product.aspx?Item=N82E16820233180 cosair kits: CMZ8GX3M2A1600C9R is the model #.. i have (2) 2x4gb kits so all 4 dimms are filled.. memory is fine in memtest and im guessing its the same issue that everyone else is having.. has anything new come from gigabyte in terms of a fix? or atleast a quick fix? im trying the whole disable the cstate power options and will reply back when done..

They were regularly feeding me updated beta BIOS files and asking me to try things but the C State disabling was the only thing that worked reliably. In the end I managed to get 2x8GB at a knock-down price so stuck that in and put the 4x4GB (which is fine, just not with this board) on eBay and that should cover most/all my cost.  RMAing the board would've incurred cost and alot of hassle so this worked out as the best solution unfortunately.  At the end of the day i'd had the ticket open with support for over a month.