Bad memory or bad board? Need opinions

[sngbrdb]

Hi all, thanks in advance! Have a GA-X38-DQ6, received it from Gigabyte as an RMA replacement for a board that went bad on me (same model). Also have two 1GB sticks of Patriot pc2-8500 memory, which the board auto-detects at 5-7-7-21 (they're rated at 5-5-5-9).

The system ran fine for a few weeks, but I've suddenly started having app crashes and blue-screens.

Running memtest on each stick individually shows no errors, even after several passes. Running it with both sticks usually shows an error on the first pass, sometimes not until the second. I have a fan on the memory, but it still runs pretty hot (rated at 2.3V).

Monitoring system voltages, they all seem within range.

Side effects:

• my graphics card was at one point idling at 90C, which I'm assuming happened because the catalyst drivers crashed... haven't seen that but the one time
• Windows has become corrupted and chkdsk runs sometimes on reboot... assuming memory issues caused file corruption
• I occasionally get a message from the BIOS saying there's a checksum error, and the BIOS is being recovered

I'd call it a memory issue, but the fact that both sticks need to be in to fail and the message about the BIOS checksum make me think it could be a motherboard issue.

What are your opinions: memory or board?

Thanks!

[Chike]

Loaded BIOS oprimized defaults to begin with? If not please do then make necessary changes.

Have you run memory test in one stick in slot #3?
Which memtest have you used?

Please fill your forum signature with your system specs BIOS vers. and link to ram specs or at least model number.
Also post your BIOS settings using the template in this post.

Is it possible to boot windows with one memory stick and post a screen caprure of MemSet main and spd(click spd) windows?
MemSet can be found in Latest Overclocking Programs, System Info, Benchmarking, & Stability Tools

[sngbrdb]

Specs:
Robust Graphics Booster........[Auto]
CPU Clock Ratio ...............[8]
Fine Clock Ratio ..............[?]
CPU Host Clock Control ........[Disabled]
CPU Host Frequency (MHz) ......[1066?]
PCI Express Frequency (Mhz) ...[Auto]
C.I.A. 2 ......................[Disabled]
System Memory Multiplier (SPD).[Auto]
DRAM Timing Selectable SPD ....[Auto]

Performance Enhance = [Standard]

CAS Latency Time...........[5]
Dram RAS# to CAS# Delay....[7]
Dram RAS# Precharge Delay..[7]
Precharge Delay (tRAS).....[21]

ACT to ACT Delay (tRRD)....[4]
Rank Write to READ Delay...[4]
Write to Precharge Delay...[8]
Refresh to ACT Delay.......[56]
Read to Precharge Delay....[4]
Static tRead Value.........[6]
Static tRead Phase Adjust..[Auto]

CPU/PCIEX Clock Driving Control [800mV]
CPU Clock Skew Control [normal]
GMCH Clock Skew Control [Normal]

System Voltage Control [Auto]
DDR2 OverVoltage Control [?]
PCI-E OverVoltage Control [?]
FSB OverVoltage Control [?]
(G)MCH OverVoltage Control [?]
CPU Voltage Control [Auto]

Limit CPUID Max. to 3.....................: [Disabled]
No-Execute Memory Protect............: [Enabled]
CPU Enhanced Halt (C1E)................: [Disabled]
C2/C2E State Support....................: [Disabled]
x C4/C4E State Support..................: [Disabled]
CPU Thermal Monitor 2(TM2) ...........: [Enabled]
CPU EIST Function.........................: [Disabled]
Virtualization Technology................: [Enabled]

Integrated Peripherals
Legacy USB Storage Detect [Enabled]


---------------------
I'm hesitant to load BIOS optimized defaults, just because I have some RAID drives configured, and I've had some problems with losing those permanently when I reset my settings. But I started by resetting the BIOS to default settings and making my changes from there, and my CPU and memory related settings are set to 'Auto'.

I've tested each stick in each of the memory slots... they work fine as a single stick in any slot, but put together they fail in both a 1-3 and a 2-4 combination. I'm using memtest86+ v4.00

Here's a pic of the MemSet settings... as far as I can tell the motherboard is picking the EPP#2 profile, but is using more conservative settings than the profile calls for, for some reason.

Thanks again!

[Chike]

Is this correct: Performance Enhance = [Standard]?
And I am sure CPU Host Frequency is not 1066.
And System Voltage Control, are the [?] auto/manual?

Can you check BIOS PC Health Status for the applied memory voltage?

[Conners]

Robust Graphics Booster........[Auto] <-- Set to Fast
CPU Clock Ratio ...............[8]
Fine Clock Ratio ..............[?]
CPU Host Clock Control ........[Disabled] <-- Enabled
CPU Host Frequency (MHz) ......[1066?] <--this is your FSB of 333 going by your sig.
PCI Express Frequency (Mhz) ...[Auto] <--100
C.I.A. 2 ......................[Disabled]
System Memory Multiplier (SPD).[Auto] --|<-- What choices do you have
DRAM Timing Selectable SPD ....[Auto]---|

Performance Enhance = [Standard]

Set your memory at their rated at 5-5-5-9 timings and voltage.

[sngbrdb]

Chike:
Performance Enhance is set to standard, the lowest setting.

The CPU Host Frequency is greyed and reads 100... it's greyed because the CPU Host Clock Control is set to Disabled, which I believe lets the board pick the frequency automatically... assuming that's the 333 I'm seeing in the 333x8.

All of the System Voltage Controls that I put [?] are greyed because the control is set to [Auto]... it isn't clear to me if the values that are there are the real values the system is picking.

The applied memory voltages:
CPU - 1.284
DDR18V - 2.096

The 2.096 surprised me... the memory is rated as 2.1-2.3, so is the board setting this too low? The DDR2 OverVoltage is on Auto, so is the 2.096 the value before any OverVoltage is applied or after?


Conners:
What will setting the Robust Graphics Booster to Fast do for the system, and why change the PCI Express frequency?

The choices for System Memory Multiplier are a combination of 200, 266, 333, and 400, with what I'm assuming are multiplier values such as 2.0, 2.4, 3.2, etc. There are maybe 15 combinations I can pick from, but 400 is the highest, there is no 533. I'm thinking if I wanted to play with manual settings it might be better to pick 333 with either a 3.0 or 3.2 multiplier?

Wouldn't reducing the timings to 5-5-5-9 cause more problems than the current 5-7-7-21 it's at now?


Thanks to both of you, I really appreciate the help!

.

[Chike]

Originally Posted by sngbrdb

Chike:
Performance Enhance is set to standard, the lowest setting.

The CPU Host Frequency is greyed and reads 100... it's greyed because the CPU Host Clock Control is set to Disabled, which I believe lets the board pick the frequency automatically... assuming that's the 333 I'm seeing in the 333x8.

Seen this penomena before, something is wrong there, the default for disabled should read 333.
Try clearing CMOS, as suggested in the second part of this post, make the long 20 minutes version, if that doesn't work am overnight clear might do.
Wait for someone that knows more than me about RAID to advise you how to go about it when clearing CMOS.

Originally Posted by sngbrdb

The applied memory voltages:
CPU - 1.284
DDR18V - 2.096

The 2.096 surprised me... the memory is rated as 2.1-2.3, so is the board setting this too low? The DDR2 OverVoltage is on Auto, so is the 2.096 the value before any OverVoltage is applied or after?

In the EPP the voltage is 2.1v so that's what the BIOS set it for, with a little drop which can be expected. For DDR-800 it should be more than enough.

[Lsdmeasap]

2.1 is the programmed amount, so that should be fine. What you see of 2.2-2.3 on the sticker or specs is only the max allowed under warranty, not what you will have to use at all (Just what you can)

I posted suggested methods for clearing your CMOS or flashing/loading optimized with RAID in the Qflash guide, please read this if you are unsure >>
BIOS Flashing - A "How To ~ Qflash Guide"

[sngbrdb]

Cleared the CMOS, and my memory errors seem to have gone away... 3 passes through MemTest, no problems. I will keep an eye on it, but remain cautiously optimistic.

Clearing the CMOS resulted in the following changes:

• CPU Host Frequency now displays 333, not 100, though it's still set automatically
• Voltage Control was switched from 'Auto' to 'Manual', and all sub-settings (e.g. DDR2 OverVoltage) were set to 'Normal'
• DDR18V Voltage displayed in PC Health now reads 2.256 instead of 2.096

All other settings remained the same, still running at 533, 5-7-7-21.

The BIOS does seem more sluggish now, and MemTest took nearly twice as long per pass? I will take slow and working over crashing any day, but that seems to point a finger at incorrect BIOS settings, despite the numbers being reported. Would love to know what it was.

Thanks all for the help... any opinions on the sluggish response? I'd love to manually set the memory timings to what their profiles rate them for, though maybe I shouldn't mess them if it's working.


Thanks again all, really appreciate it!!


For reference, the clearing process:

1. Powered down and unplugged everything, including usb devices and network cable
2. Removed the battery and jumpered the pins
3. Ran errands for a couple of hours
4. Put everything back together and powered up
5. The BIOS hung the first time powering up... said it was searching for the BIOS image on the hard drive?? At this point the RAID array was not configured, so I don't know where it was searching or what it was trying to do.
6. Forced a reboot. Got stuck in a reboot cycle.
7. Used the power-button soft-off trick to break the cycle. Booted up into the BIOS
8. Loaded optimized defaults, configured the RAID drives, rebooted
9. Back into the BIOS to set the Boot Priorities (RAID drive doesn't show up until you save those changes).
10. Rebooted and started Windows
11. Windows reported a corrupted drive, ran chkdsk, and repaired some errors!! Related to the BIOS trying to find an image when the drives weren't set to RAID? Scared the crap out of me, whatever it was
    
12. Windows booted and all seems to be well... can't see any problems from the chkdsk fixes, so hopefully all is well.

.

[Chike]

Performance Enhance set standard or turbo?
I realy wonder why FRAM Voltage shows 2.256v as the EPP shows 2.1v.
Could you run MaxxMEM and post a screen capture of the results?(MaxxMEM has a built in feature for that) and another screen capture for MemSet main window (Alt+PrintScreen captures the front window only)