0x00000124 BSOD

[keving98]

Hello,
I'm having some issues with a new build (actually I built it in Jan '12 but the customer is just getting around to testing it).
Here are the specs:
(2) Xeon HC X5690 Processors
EVGA Classified SR-2 5520 (56 bios)
Kingston (KHX1600C9D3K6/24GX) 24GB 1600MHZ DDR3 Non-ECC CL9 DIMM KIT6 (9-9-9-27; 1.65; From EVGA QVL)
(2) Kingston 240GB HyperX SSD SATA III - 1st for OS; 2nd for Pagefile/Swap
1.5TB Caviar Black SATA 6GB/s 7200 RPM 3.5" Hard Drive w/64MB Cache (Backups)
(1) PNY nVIDIA Quadro 6000 6GB GDDR5 ECC PCI-E X16
AX1200W Fully Modular ATX 80 PLUS Gold Power Supply
Logitech Wireless Desktop MK710
Intuos4 Medium Pen Tablet
(2) Antec Kuhler H2O 620 Liquid Cooling System
Windows 7 Pro SP1 64-Bit
Cinema 4D (with latest updates)

The problem I'm having is I'm getting 0x00000124 BSOD's while doing rendering with Cinema 4D. The file I'm trying to render takes about 4.5 hours to complete. It's never made it more than 45 minutes. It does this with multiple Cinema 4D files. The temperatures for the CPU's never hit 65c and the video card GPU's never hit 85c (Nvidia tells me 105c is max). The video card averages about 30% load and the all CPU cores run at 100%. The case hovers around 45c.

Here is a little of the minidump from BlueScreenView: 052312-12453-01.dmp 5/23/2012 4:10:18 PM 0x00000124 00000000`00000004 fffffa80`1309e038 00000000`00000000 00000000`00000000 hal.dll hal.dll+12a3b x64 ntoskrnl.exe+7cc40 C:\Windows\Minidump\052312-12453-01.dmp 24 15 7601 333,120

Here's some of the the things I've tried:
Updated the Nvidia drivers to latest Performance version
Tried a second Quadro 6000 with a different firmware version with the same result
Updated every motherboard, etc. driver I could find
Checked for Cinema 4D updates
I tried tweaking the bios to match the memory. I don't know if the settings I set in the bios were correct or not. The 1600 memory I have installed is running at 1333 and initially I let the bios auto configure because I had no interest OC'ing (I don't know why I didn't just buy 1333 memory to begin with). Everything I did produced an unstable system.
Tried Dummy OC (BSOD on boot).
Replaced the memory with six sticks (12GB; CMX6GX3M3A1333C9; 9-9-9-24 1.5v) Corsair XMS3 1333 memory (not on EVGA QVL) with same Cinema 4D result. Side-note: CPU-Z saw 12GB but Windows only saw 6GB (ran Cinema 4D anyway).
Ran Cinema 4D with the Logitech wireless desktop and Wacom tablet disconnected.

Forgot to mention: I have the Quadro in PCI-E Slot #5 (PCI-E x16/x8) to keep the card from covering the chipset fan. Don't know if having slot #1 - #4 empty makes any difference.

I've ran Memtest 3 times with (2) completely different sets of memory and can't get past 3.5 minutes. No errors are reported but it just becomes unresponsive. Both sets of memory were the Kingston (KHX1600C9D3K6/24GX) 24GB 1600MHZ DDR3 Non-ECC CL9 DIMM KIT6 (9-9-9-27; 1.65; From EVGA QVL). It's the only memory on the EVGA QVL available locally.

Any idea what might be going on?

Any help would be greatly appreciated.

Kevin G

 

[Convicted1]

BSOD 124 means you need more VCore Voltage 9 out of 10 times.

 

[redduc900]

... or VTT voltage, especially w/ 24GB of RAM installed.

 

[Convicted1]

... or VTT voltage, especially w/ 24GB of RAM installed.

Yeah... I had actually forgotten about that.

And in this case I'd say it's probably more likely to be VTT than VCore.

 

[EarthDog]

Nice rig.. def try raising vtt first and see if it responds with that. If not, try vcore.

 

[wingman99]

Did you check to see if the memory timing is set correctly in Bios?

 

[keving98]

You guys are the greatest!

Changing the VTT settings did the trick. I haven't ran Memtest yet but it finished the render for the first time.

Other interesting results of changing the VTT settings:
-- The time to complete the render went from 4.5 hours (estimated because it would never finish) to 6.25 hours (estimated and then confirmed). It was averaging 1 frame completion per 12 seconds before the change to 1 frame completion per 19 seconds after. I'm a little disappointed in the added time but to put it in perspective the customers next fastest workstation was estimated to take 3 days to do the render. They'll still love the machine.
-- The average GPU load changed from 30% to 8%.
-- The GPU average temperature dropped 26 degrees from 83 to 57.
-- The CPU average temperature remained about 60 degrees.

The only bios settings different from default are the VTT settings. However, once it ran the render for 2.5 hours I could tell it was working properly and I shut it down and changed the memory timings and memory voltage to 9-9-9-27 and 1.65v and started the render from the beginning.

One other question though:
I'd like to have the memory running at what it was spec'd at (1600). I know it won't make much of a difference in performance but it bugs me a little that it's under-clocked so to speak. I don't want to destabilize it in the process though. If it won't hurt anything, what other settings besides the timings and voltage do I need to change to get the memory running at 1600? I'm not too lazy to look up the information myself but every time I changed the settings in the bios to try and achieve 1600 it crashed. I don't know if I'm missing something or it's just not attainable for this configuration.

Thanks again for all your help.

Kevin G

 

[Janus67]

Should just be rated timings, RAM voltage, and possibly some more VTT tweaking.

 

[keving98]

Thanks Janus67. I'll give that a try.