Jump to content

Memtest stable, but unstable with anything else


Recommended Posts

This is my first post to this forum so please be gentle with me ;)

 

I built the rig in my sig back in July-August, and have been struggling with it ever since. I'm all out of ideas, so I hope that somebody here can help. Please forgive the long post - I'm just trying to include any information that might be relevant.

 

After I built the rig, the first thing I did was to flash the 623-3 BIOS, since I have an X2. I then set about determining the max OC of the RAM. Using memtest86+ v1.60, I've successfully tested the RAM at 2-2-2-6 1T, dual channel, at 212 MHz (32+ passes).

 

I then installed Windows at (presumably) stable CPU/RAM settings (and only 1 stick of RAM) and the chipset drivers, and Ghosted the partition to a DVD just in case it got trashed. After a lot of trial-and-error, I got two instances of SP2004 to run at 305x9 for >28 hours, when one instance failed, so I assumed that 304x9 would be rock stable. Temps are OK: I've never ever seen the CPU temp higher than 45 deg C. There's a 120mm fan right over the Ballistix to avoid overheating.

 

The system seems to favour the 9 multiplier. With memtest, I've run the following settings without errors (minimum 32 passes):

200x11, 1:1, 2-2-2-6 1T, dual channel

304x9, 7:10 (210 MHz), 2-2-2-6 1T, dual channel

304x9, 3:4 (228 MHz), 2.5-2-2-6 1T, dual channel

304x9, 5:6 (248 MHz), 2.5-3-3-6 1T, dual channel

 

Just to avoid potential confusion: in this post memory timings are shown as Tcl-Trcd-Trp-Tras (the order in the DFI BIOS is Tcl-Trcd-Tras-Trp).

 

I've also successfully memtest'ed the Ballistix at stock settings on an Asus K8N-based system, individually and together (single channel, as the K8N is a 754 mobo).

 

But - and this is what's driving me crazy - I haven't been able to find any combination of frequencies, voltages and timings that are stable in Windows or Knoppix (live CD) - not even at stock or sub-stock - when running dual channel.

 

Running 2T and/or low RAM frequencies (1/2 divider) and/or low HTT and/or single channel tends to reduce instability, but not enough to have a completely stable, halfway-decent performing system.

 

I've had my share of POST failures, BIOS freeze/failures, lockups, BSODs etc. - for example, this morning, at stock settings, the system kept entering the nVidia RAID IDE ROM BIOS instead of the DFI NF4 BIOS (strange, because RAID is disabled in BIOS and currently only one HDD is connected) - so I've had the CMOS cleared several times.

 

Suspecting a dodgy BIOS, I've downloaded it again and reflashed it, but to no avail.

 

I've trawled the posts in these forums and although I've found many people having similar-sounding problems, the prescribed solutions don't seems to work in my case. I've read the guides, tried and played around with other people's Ballistix settings, and with RGone's settings, but nothing seems to help.

 

What makes it extra frustrating is that I can't seem to conclusively isolate the problem to the CPU, RAM or mobo. My prime suspect - and this is just a hunch - is the CPU's memory controller, but I doubt that I'd be able to justify an RMA (in my experience, Danish retailers usually require clear evidence that the product is defective before accepting an RMA).

 

Unfortunately, I don't have access to any other Socket 939 CPUs/mobos or dual-channel RAM (or >=480 Watt PSUs) to test with, so my options seem pretty limited.

 

I've measured the 5V and 12V voltages with a DMM, and they are stable: within a couple of percentage points of spec., and fluctuations of maybe +/- 0.01V from power-on to running the OS. I've haven't screwed up the courage yet to directly measure Vcore and Vdimm (this will require dismantling the whole system, to gain sufficient access to the mobo).

 

Any help would be very much appreciated - many thanks in advance.

Share this post


Link to post
Share on other sites

  • Replies 30
  • Created
  • Last Reply

Top Posters In This Topic

Hello,

 

There is a weird issue with the Seasonic power supplies, sometimes the pc wont boot.

Might be a good idea to borrow another PSU. (different brand).

 

Have you lowered the LDT ratio?

Share this post


Link to post
Share on other sites

Hello Sharp,

 

Well, with the 304 HTT I've mostly used a 3x LDT multiplier, but what with all the other variables I haven't made playing around with the LDT a priority. I do remember at one point I was down to maybe 780 (whereas 1000 is the ideal), but I didn't notice any particular improvement at the time. It's an idea, I'll do some more systematic testing with it this evening.

 

I have read that people have been having problems with the +5VSB on the Seasonic S12s (http://www.dfi-street.com/forum/showthread.php?t=17916). I can't rule out that part of my problems may be due to the S12, but it's a relatively rare occurence that my PC won't boot the BIOS (as opposed to booting the OS, which can be difficult) - up till now I've been able to deal with such situations by pressing the Insert key on startup or by clearing the CMOS.

 

In any case, if I can narrow down the field to one particular component, it'll be in for the chop.

Share this post


Link to post
Share on other sites

Clear the CMOS for 10 minutes.

 

Then try this without overclocking. (use stock settings).

 

GENIE BIOS Setting

>DRAM Configuration..................Press Enter = New Menu

FSB BUS Frequency....................200

DRAM Voltage Control.................2.7V

 

GENIE BIOS Setting >DRAM Configuration

DRAM Frequency Set (MHZ).............200 (DRAM/FSB:1/01)

Command per clock (CPC)..............Enabled

Cas latency (tCL)....................2

RAS to CAS delay (tRCD)..............3

Min RAS active time (tRAS)...........6

Row precharge time (tRP).............3

Row cycle time (tRC).................11

Row refresh cycle time(tRFC).........14

Row to Row delay (tRRD)..............3

Write recovery time (tWR)............2

Write to read delay (tWTR)...........1

Read to write delay (tRWT)...........1

Refresh period (tREF)................AUTO

Write CAS latency (tWCL).............1

DRAM Bank Interleave.................Enabled

 

DQS Skew Control.....................Decrease Skew

DQS Skew Value.......................0

DRAM Drive Strength..................AUTO

DRAM Data Drive Strength.............AUTO

Max Async Latency....................6ns

Dram Response........................Normal

Read Preamble Time...................5ns

Idle Cycle Limit.....................AUTO

Dynamic Counter......................Disabled

R/W Queue Bypass.....................AUTO

Bypass Max...........................AUTO

32 Byte Granularity..................Any value

Share this post


Link to post
Share on other sites

OK, the continued story:

 

I left the CMOS to clear for 10-15 minutes (having done all the power, battery and jumper stuff), but when I tried to start up again I had the previously-seen BIOS-booting problem again. This time I noted that the two of the diagnostics LEDs were on, indicating that the graphics adapter hadn't been detected (certainly, the LCD screen detected no video signal). For kicks, I powered down, removed the second stick of RAM (leaving one stick in slot 2), powered up and this time 3 LEDs were on, indicating RAM not detected. Odd.

 

I've been through this scenario before, but Sharp's comment about the PSU started me thinking, and I recalled that the Stacker comes with what you might call a jumpstart extension cable that allows two PSUs to be started at the same time. So I got that installed along with a second PSU - an old 300W, 20-pin Fortron - so that the Seasonic powered the Ultra-D and a single optical drive, and and the Fortron powered the fans, HDD, water pump etc.

 

This setup didn't alleviate the startup problem either, so I left the CMOS to clear again, this time all night (I needed some sleep anyhow). This morning, I still had the startup problem, but eventually defeated it, loaded optimised defaults, rebooted, and entered and saved the settings proposed by Sharp.

 

Unfortunately, these settings weren't stable either:

- The BIOS sometimes hangs in the first screen, just after displaying the "Main Processor" line (before displaying the memory size);

- Knoppix hangs while it's loading files, or shortly after the graphical desktop is started;

- Windows get stuck in the boot-selection screen, or in the copyright screen (with the moving bar thingy), or reboots from there; or BSODs (PAGE_FAULT_IN_NONPAGED_AREA)

 

(All this is still with only one stick of RAM.)

 

This is pretty much par for the course. Thinking back, it's my overall impression that I've had more startup problems (BIOS hanging in the first screen or not starting at all, with LEDs indicating that RAM or graphics haven't been detected) with the CPU at stock or near-stock settings than at the 304x9 CPU overclock (usually with RAM at 195 or 210 MHz, depending on the divider). Hang on a sec...

 

(...goes off and reboots the PC several times...)

 

Hm, hanging in the BIOS seems to be much more prevalent at these stock settings - say, one BIOS hang in 3-4 reboots - than at the 304x9 overclock that I've been using for much of the past several weeks, which would hang in the BIOS once for maybe every 50 reboots (give or take a couple of dozen either way), at least an an order of magnitude less often.

 

This doesn't necessarily have to have anything to do with the settings themselves: I shifted from OC'ed to stock settings because of the problems I was having - if the problems now seem worse, it could (imaginably) be because I have one or more components that are gradually degrading, rather than because the stock settings are bad as such.

 

I'll have to do some more experimenting to confirm or reject this hypothesis. Here's what I'll try:

 

a) Go back to single PSU (Seasonic)

B) Go back to dual channel (no big hopes for improvement here, but you never know...)

c) Try 304x9 CPU overclock, with RAM at 195 or 210 MHz, keeping the settings proposed by Sharp.

 

I'll report my findings here.

 

Sharp, thank you very much for your efforts so far. If you - or anyone else - have any more ideas, well, just keep 'em coming, they'll all be very much appreciated. Many thanks in advance.

Share this post


Link to post
Share on other sites

I only have probs like this when OC'ing ur SPD on the ballistix should be 2-2-2-8 @2.8 so lets star there 1st tho load optimised defaults, enter ur memory settings and change them to CPC enabled ur timming @ 2-2-2-8 and the rest@ default values, now set up

Cpu startup VID...................1.425 Volts

CPU VID ............................. 1.425 Volts

CPU VID Special Control...... 110%

LTD Voltage..........................1.4

ChipSet Voltage....................1.8

Dram Voltage........................2.9

 

this sort of set up is normal used for stubborn 2 sticks of any ram but may help you get stable, and then work backwards with the above as you dont in most cases need volts that high, and of course read as much as you can in threads with X2 users as it seems other have the fair share of troubles with these CPU, good luck!

Share this post


Link to post
Share on other sites

- Sharp:

Yes, I have tested them individually, but it's a while ago now. But I did a successful memtest om both sticks (dual channel) only a couple of days ago, and I would expect both sticks to be OK. Do you have any experience of RAM sticks passing memtest together, but failing individually? It wouldn't seem logical, but reading these forums it seems that logic sometimes has its limitations :) , at least at the level we puny mortals are able to apply it.

 

 

-CsA-TAZ:

The SPD isn't quite correct: recently, Crucial's webpage has stated the rating of the Ballistix as 2-2-2-6. I'm pretty sure I've testing the timings you've proposed somewhere along the way, but I'll give them a try anyhow.

 

Thanks to you both for your input.

Share this post


Link to post
Share on other sites

I would RMA the mobo RAM and CPU, one of thoughs is deffinately faulty, if i had to put my money down on one or the other, id say the mobo. but just rma all 3 to eliminate them, if the problem persists, its pretty safe to assume for some reason the board doesnt like the PSU.

Share this post


Link to post
Share on other sites

@ExRoadie

 

I couldn't agree with you more - I try not to assume anything. One stick is being memtested (at Sharp's suggested timings) as I write this; 18 passes so far, no errors (4 1/2 hours' testing). The other will go in tomorrow morning.

 

And the guide was where I started - please don't ask me to start from scratch! (Breaks down in tears)

 

@kuniva

 

For the reasons I've stated in my first post, RMA'ing will be my last resort. Without hard evidence that a particular component is actually defective, chances are the retailer will simply return it to me (and charge me for his time used to check it, and for packing & postage).

Share this post


Link to post
Share on other sites

@mc123

 

Just be aware that a single OS crash can corrupt files rendering the rig OS unstable while the hardware checks out OK.

 

I'd rather reinstall the OS once than deal with continued stability problems.

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×
×
  • Create New...