F3-12800CL7D-2GBRM - BioStar T5 XE CFX-SLI

[speedyme]

My config:

BioStar T5 XE CFX-SLI motherboard; latest BIOS 10-28-09
4 sticks F3-12800CL7D-2GBRM
Intel i7-860

I have been experiencing random BSOD using the system. When it crashes, I can restart into BIOS embedded Memtest 2.11 and get consistent errors (usually error bits 00000040). Cold boot clears errors -- and it will run MemTest for hours without error -- I have let it run for as much as 9 hours. Error happens in active use of system, and also when idle.

Problem started with initial build/software load. At that time BIOS was set to defaults -- not overclocked. I have tried several adjustments since then, currently have memory set to 1600 - 7-8-7-24-2T - 1.63v (using settings from SPD XMP-1 and suggested to others on this forum). No other overclocking.

Thoughts? I appreciate your help!

[speedyme]

Update on symptoms

As I noted, if rebooted into memtest without cold boot it displays consistent errors in memtest -- error bits 00000040 -- in the address ranges 25xx - 26xx MB and 46xx-47xx MB. Do the address ranges and Error bits and the fact that a cold boot clears the errors help to isolate the problem to the Motherboard, memory or whatever?

Thanks for any help you can provide.

[Stevinchy]

I would try upping the voltage to 1.65. I have the exact same kit and they run fine for me and the only difference between our settings is the voltage.

[speedyme]

Tried 1.65 v. Still getting BSOD.

[GSKILL TECH]

Post up some BIOS screen shots and we can see if anything is set incorrectly.

Thank you
GSKILL SUPPORT

[speedyme]

How do I get BIOS screenshots?

[speedyme]

BIOS Screenshoots

http://www.flickr.com/photos/speedy_...7622956013603/

I appreciate your help.

[speedyme]

Anybody see anything set incorrectly? if not, suggestions?

[speedyme]

Added Memtest results

I have added an image to Flickr that demonstrates that the system will run indefinitely in MemTest cleanly -- if it starts that way.

Does this give you all any clues as to what is going on here?

Appreciate your help.

[speedyme]

Cold boot problem

I did some more testing. If I start the machine after it has been off for several hours, and boot directly into MemTest, I get the errors immediately. If I restart (w/o power off) the error persists. If I shut down (power off) and immediately power on, the error is gone -- and will stay gone as long as the machine is powered up -- I have run Memtest for 36 hours w/o error.

Does that explain anything?

[Tradesman]

Have you tried increasing the CPU/MC voltage a bit - might also want to see if you can borrow/try a different PSU, may be getting a power fluctuation on startup

[speedyme]

May be cured with BIOS setting change; thoughts?

I do not have access to another PSU -- without buying one.

I don't know how to "increase the CPU MC voltage a bit" -- I have settings in the BIOS for CPU Vcore (currently 1.126); CPU VTT (currently 1.153); CPU PLL (currently 1.85); and PCH voltage (currently 1.115) plus the DRAM voltage.

However, working on the suggestion that the problem is a power-related transient at cold boot, I disabled a setting in the BIOS that manages the VRM phases based on CPU load. Now all 4 + 2 phases are always 'ON' -- and, at least in preliminary testing, the problem seems to be gone.

Since heat is the enemy of stability and long life, what do y'all think about the thermal impact of this setting change? I suspect it is relatively insignificant.

Thanks for your help.

[speedyme]

No Joy

Well, the change in VRM phase mgmt. did not solve the problem.

I did some more reading and learned that CPU VTT is the setting you referenced as CPU/MC voltage. Increasing it to 1.21v from 1.153 did not help either.

I guess I can order up another PSU, if that is what you think would help. However, does it seem reasonable that a PSU transient would always generate the same error in the same ranges of addresses?

Appreciate your thoughts!

[speedyme]

Problem persists, new mobo

Fry's even-swapped the mobo with me, and I reordered the sticks of RAM in the slots. Same error, same memory address ranges. Absolutely repeatable on several hours cold boot.

Using increased VTT as suggested previously.

Does it make any sense that this error could be so specifically repeatable (Same error, same memory address ranges) if it was a PSU transient?

What now?

[speedyme]

2 sticks = No Error; 4 sticks = Error

Works with only 2 sticks. Need 4 since this machine exists to host VMs.