GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

[qwertylesh]

Hi guys,
I have trouble with a system posting after the cards have been set to run in a sli mode
full specs:

Mobo: Gigabyte X58A-UD3R (Rev 2) Bios: second latest beta (FB11)
Processor: i7 930 (first stepping)
Ram: 3x 2GB Kingston value ram 1333 CL9 (KVR1333D3N9K3/6G)
Graphics: 2x Gigabyte GTX 465 (GV-N465UD-1GI)
HDD: 2 x 1TB WD Black 7200rpm
PSU: OCZ Fatal1ty 700w


Before I get to the template, I just want to say that I am not all that fussed weather the memory is at 1066 or 1333,
provided that it can pass the ram stress testing and remain stable.

I think the problem lies with the pci-e bus, the symptoms vary but I will describe them here:
Sometimes at Verifying DMI Pool data, the screen is black with a small blinking DOS style cursor but peripherals and all else is hard locked up.

sometimes during windows booting (switching from boot splash to GUI mode) the screen will go solid black and hard lock up -> this was the symptom at optomised defaults with minor general bios changes.

now though, after my attempted tweaking,
the boot up issues are far less frequent and the GUI switch lockup shows a light blue screen,
and takes a lot more soft restarts and testing to expose the issue, but it is still there

if the problem needs to be reiterated better, please let me know.

Here is the template which has the least issues,
I know raising the IOH core helped, also keeping Quick boot disabled seemed to help,
also the small bump of pci-e volts and locking the pci-e freq helped,
but something needs further bumping since it still is playing up. :(


template:

Advanced CPU Features:
CPU Clock Ratio ................................ [21x]
Intel(R) Turbo Boost Tech .................. [Enabled]
CPU Cores Enabled ............................ [All]
CPU Multi Threading .......................... [Enabled]
CPU Enhanced Halt (C1E) ................... [Enabled]
C3/C6/C7 State Support .................... [Disabled]
CPU Thermal Monitor ......................... [Enabled]
CPU EIST Function ............................ [Enabled]
Virtualization Technology ................... [Disabled]
Vmware or Virtual PC Programs
Bi-Directional PROCHOT ..................... [Enabled]

Uncore & QPI Features:
QPI Link Speed ..............................[Auto] - 4.8Ghz(x36)
Uncore Frequency ..........................[Auto] - 2667Mhz
Isonchronous Frequency ..................[Enabled]

Standard Clock Control:
Base Clock (BCLK) Control ................ [Auto]
BCLK Frequency (MHz) .....................133
PCI Express Frequency (MHz) ........... [100]<Manual 100.

Advanced Clock Control:
CPU Clock Drive ..............................[800mV]
PCI Express Clock Drive ................... [900mV]
CPU Clock Skew ............................. [0ps]
IOH Clock Skew ............................. [0ps]

Advanced DRAM Features:
Performance Enhance ...................... [Standard]
System Memory Multiplier (SPD) ........ [AUTO] 1066
DRAM Timing Selectable (SPD) .......... [Quick]

Channel A, (B + C = quick -> set by A)

Channel A Timing Settings:

##Channel A Standard Timing Control##
CAS Latency Time ......................9<manually set
tRCD .......................................9<manually set
tRP .........................................9<manually set
tRAS .......................................24<manually set

##Channel A Advanced Timing Control##
tRC ........................................27 <auto
tRRD .......................................4<auto
tWTR ......................................4<auto
tWR ........................................8<auto
tWTP ....................................... 19<auto
tWL ........................................ 7<auto
tRFC .......................................80< <manually set, auto would be 60
tRTP .......................................4<auto
tFAW ......................................16<auto
Command Rate (CMD) ................2<manually set, 1 would be auto, but I think the SPD is wrong.

##Channel A Misc Timing Control##
Round Trip Latency ...................47<auto

B2B CAS Delay ..........................-<auto


Advanced Voltage Control:

CPU
Load Line Calibration ................. Auto
CPU Vcore ...............................1.29375V <Auto (PC Health reports:
QPI/VTT Voltage 1.150v ............1.175 <auto
CPU PLL 1.800v .......................1.8<auto

MCH/ICH
PCIE 1.500v ...........................1.540<manually set, auto would be 1.50
QPI PLL 1.100v .......................1.100<auto
IOH Core 1.100v .....................1.160<manually set, auto would be 1.100
ICH I/O 1.500v .......................1.500<auto
ICH Core 1.1v ........................1.100<auto

DRAM
DRAM Voltage 1.500v ..............1.580<manually set, auto would be 1.500
DRAM Termination 0.750v ........[AUTO]
Ch-A Data VRef. 0.750v ..........[AUTO]
Ch-B Data VRef. 0.750v...........[AUTO]
Ch-C Data VRef. 0.750v ..........[AUTO]
Ch-A Address VRef. 0.750v.......[AUTO]
Ch-B Address VRef. 0.750v.......[AUTO]
Ch-C Address VRef. 0.750v.......[AUTO]


So yeah, the above profile proves to be quite and improvement but It doesn't solve the occasional post issues,
it is more reliable as above then using failsafe defaults without setting anything extra.

I'm quite certain the issue is because the cards are running in SLI, and the cards can handle a good stressing at the desktop,
as can the cpu and ram, it just seems to give me lots of greif getting it to always post reliably.

[wazza300]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

i would manually set qpi/vtt to around 1.2000-1.25000v,try with the latest official bios too,not a beta,maybe that bios is buggy

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

Thanks for that Wazza,
I'll see what the QPI/VTT is being set to on Auto with easytune, and I'll also flash back to FA, see if it helps.
Though I'm not confident FA bios will help, all the hardware is new, but I have no faith in the OCZ psu and think if its going to be a specific part it may be that.

I am just hoping it is something that can be worked around and not some sort of incompatibility which would leave me unable to get it working properly.

[qwertylesh]

update

ok, after spending a couple of hours of thorough symptom replication, I have determined the following.
Happens with or without sli enabled.
happens on each card individually
happens with a different power supply unit.

So I can safely say this is no longer a SLI issue, and having ruled out the graphics and power supply as causes of the fault, I am now of the belief that the problem lies with the motherboard or the ram.

Which is odd though, since when this system was first assembled, it was thoroughly prime blended and ram stressed to ensure stability.

I will be trying another triple channel kit on the system, and if the problem still occurs, I'll have to try another disk and another processor and if the problem sitll occurs that leaves the motherboard.

All system temps are fine, especially when only one of the 465s is only installed (no card sandwiching heat)

I am hoping its just a bad kit of ram and not the mobo, I won't know until I have ruled out the other components.

I'll keep yous apprised on the fault and solution (when found).

Also, I did flash back to FA, I haven't checked the QPI/VTT voltages out yet, but since it happens on plain optimized defaults with 1333 ram, I am really thinking its a hardware problem moreso then something which could be tweaked.

[qwertylesh]

Re: update

OK, so it turns out that the system had a faulty triple channel kit of ram.
this was causing the black or light blue coloured hard lockup screen when trying to start the Win7 GUI.

With a new kit and a single graphics card or both (out of SLI) the systems flawless.

Unfortunately, when enabling SLI, the system still has intermittent 0x0000116 BSODS :(

After some tweaking Ive managed to vastly subside the issue but changing these parts of the template:

QPI/VTT raised qto 1.235
PCI Frequency changed to 101Mhz
IOH Core raised to 1.20v
DRAM voltage raised to 1.62v
DRAM Timings locked to 9-9-9-24-2T (stock rated), DRAM Multi raised to 10 (@1333Mhz stock rated)
Uncore raised to 2.8Ghz (x25?)
Please let me know if any of these alterations are dangerous for permanent usage, Ive never had to adjust the pci frequency beforel, but it seemed to have allowed the system to run stable thus far, 5 full shutdowns, 10 soft restarts, no more bsods. :)

[Lsdmeasap]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

Nice to see you got most of this sorted out!

If you are using x10 memory multi you may be best keeping uncore to 20-22. Your PCIE is fine, you can use 101-103 or so without ever having issues

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

Actually, I still have this system and the problems are back to square one :(

I changed the ram kit, that stopped the light blue hard lockup when Win7 tries to start the GUI.

then I found one of the GTX465's had a DVI port which wouldnt output to analogue, so I changed that 465 with a new one.

But it just did the light blue screen hard lockup on me after I enabled SLI.

I was able to make it do this without enabling SLI before I changed the graphics card.

So now, I have 2 kits of 6gb DDR3, 3 GTX 465's, one ex58a-ud3r, and this persistant issue is still happening.

Maybe it needs 103Mhz PCI-e? All the cards and the motherboard are gigabyte, could it be configuration or maybe an issue with my board or could the cards need a bios update? I really think that I am not going to get this system to operate right.

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

What gets me the most is this isnt what I would consider a really high spec system, I've dealt with 960's and 1866Mhz ram in the past without any issues like this. Maybe the problem is because the board is Revision 2 and the bios isn't mature.
Im starting to suspect the board and power supply since ive tested the hell out of the ram and graphics (not to mention tried replacements) and pretty much gotten nowhere.
Ive just turned off Sata3/USB3/esata/PATA and will see if that makes any difference. :/

[Lsdmeasap]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

You can try 103 PCIE but not sure it's going to help, sounds to me like there is some other issue.

Have you started a ticket with Gigabyte support, just wondering if maybe they could send you a BIOS update for the cards or maybe a unpublished one for the board

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

Haven't contacted Gigabyte directly yet.

I have only one 465 and one piece of the ram in the system (first white slot of course), I just cleared the cmos and loaded optomised defaults.

I have setup a batch to restart the system every 5 seconds and put an abort shutdown shortcut on the desktop. I shouldn't done this days ago, its really sped up exposing the symptom when adding/removing hardware.

The 3 parts left not ruled out through testing yet are the mobo, disks, cpu, it must be one of these.

At this stage I am thinking its either due to the disks being WD Blacks (the firmware on these disks has given me greif in other x58 gigabyte boards once before)

or a bad board, I don't see how this problem could be due to the processor. If i do find that I cannot expose the problem with 1 piece of ram, I will remove and re install the processor and work my way back up to 3 pieces. If that doesn't work (not optomistic it will) then I'll setup another Win7 OS on a Seagate 1.5tb I have sitting here and ee if i can expose the symptom. This still doesnt solely rule out the processor, but aside from .......
^ tl;dr.
Its still stuffed with 1 GFX and 1 piece of ram.
I strongly feel its the motherboard, I dont think bios is going to make any difference, I had this problem on the beta and have it on FA, have it with 3 cards, 2 ram kits.

I'll try one last thing, running another OS on a Seagate. If I can still replicate the symptom, it leaves cpu and mobo.

[Lsdmeasap]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

I have a WD black 1TB without issues, but it is a storage drive.

Is there a BIOS update for the GFX Cards?

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

When I looked on the site a few days ago, there was no bios update for the cards.

I also came to the conclusion that its not the disk, the Seagate sata drive exhibited the same symptom.

I have since changed the mobo with another one and have setup the os to restart 5 seconds after starting, So I should know in an hour or two if the problem persists with another motherboard. :)

--

I should mention, for the new mobo, Im testing it with one card, one piece of ram, clearing cmos before powering on, then loading optomised defaults, saving, rebooting, then going in and setting the ahci and checking the voltages out.

After 5 restarts, the symptom persisted with the different motherboards.
The only remaining parts for the system that have not beein swapped with different parts to test is the CPU and the case fans/buttons/usb.

I've now unplugged the usb/power/fans so really the only untested hardware is the processor, and I can't believe that this 930 is at fault, if the system has the problem again since unplugging the front i/o, then aside from trying another CPU all that im left to believe is that there is a hardware incompatibility with these parts and it cannot be resolved by me immediately. :/

oh man, if it turns out to be the casefans or something, I'll be a bit cheesed lol.

Looks like i'll have to give my vendor a call and arrange a different brand of hardware if changing the i/o and cpu doesn't fix it.

[Lsdmeasap]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

How is this "Restart in 5 seconds" helping you diagnose anything?

Have you tested with an ATI Card? You should :)

[qwertylesh]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

the hard lockup restart seems to specifically occur around the point where the windows log on is loading the user account. This is why I'm using reboots (os commanded) to test the problem, and the batch has just made my job a lot more efficient.

its basically does a shutdown /r /t 5 in the startup folder, and I added a shutdown -a shortcut on the desktop, this way, each time I change hardware, power it back on, set the bios (if needed) and let it boot up, I know it will either hard lockup if the problem still occurs or it will just cycle endlessly if the problem is resolved.

update!
With no front i/o or usb or case fans attached, 1 piece of ram, 1 gtx 465, new mobo, new ram, new graphics card, psu ruled out, chassis ruled out, hdds ruled out, os ruled out, problem persists.

just wow, I never expected this to be so hard to find the culprit, I'll install an ATi card and allow it to cycle for a while, everything has been changed aside from the processor thus far, only some parts are still of the same brand like graphics.

If the ati card does the same problem, then i'll have no choice but to try another processor and Ill need to give gigabyte a ring.

[Lsdmeasap]

Re: GA-EX58A-UD3R and 2 x Gigabyte GTX 465 SLI posting issue

Wow, hope you can figure it out as it sounds like you have about tried it all!