ASUS A7N8X + 2x 512 MB Crucial RAM = Instability. Suggestions?

[Palek]

Well, it seems my cry of joy in another thread was a wee bit premature. I still have random lock-up/reboot problems with my computer.

I just purchased a brand new A7N8X Deluxe (BIOS 1001.g) with two sticks of Crucial 512MB DDR 333 MHz RAM. The Processor is a 2400+, the power supply is an Enermax 350W model, the vidcard is an ATi Radeon 9500 Pro. Overall a pretty nice system with nothing but high quality parts. But high quality parts do not guarantee stability anymore these days, it seems.

I get random lockups under Windows 2000 SP3 when I have both sticks installed in slots 1 and 3. The message is usually something like:
DRIVER_IRQL_NOT_LESS_OR_EQUAL (I think this is the message I get, not sure now) followed by a driver, most recently "atapi.sys". The lockups are not triggered by any specific action, they will sometimes happen during a game, sometimes when I switch tasks, or when I start or shut down a program.

I run the memory at very mild speeds, in sync with the FSB at 133MHz and optimal (slow) timings.

I removed 1 stick and ran some burn-in tests in Sandra then a little torturing with 3DMark2001 SE. The machine worked fine. Then I swapped this module with the other one, did the tests again, and there were no problems this time, either. From this I concluded that I have two healthy RAM modules - although my conclusion could be incorrect, in which case I would appreciate it if someone pointed this out to me. Anyway, now I am suspecting that I have a flaky motherboard. Could this be the case? I will try to run the machine tonight with all the RAM on one channel (ie slots 1 and 2), but needless to say ideally I would like to make use of both channels. I know some people dispute the validity of two channels without the presence of an integrated graphics core, but I do not care about that. I want my ASUS motherboard to provide the advertised features at maximum stability.

As always, any input will be appreciated! Thanks!

 

[mechBgon]

First try this: update from the BIOS Of Doom (1001.G) to the 1002 beta that chizow has tested and found to be a keeper on his board: thread Set the RAM to Sync and not SPD.

I hate to say this, but as one Enermax 350W owner to another, you may need a bigger PSU. If yours is like mine, it has only 185W on the 3.3V + 5V lines combined (look at the label). That's 35W less than an Antec SL300 and almost 60W less than an Antec TruePower330.

You can test by reducing the electrical load as much as possible (pull non-essential cards, unplug non-essential drives and put in a low-powered 3D card if you have one). One DIMM can use as much as 15W, so it could be the straw that's breaking the camel's back. Or maybe I'm just being a pessimist Good luck with the troubleshooting!

 

[Palek]

Thanks, mechBgon, that was pretty fast!

Well, after reading a few other threads started by other disheartened A7N8X users, I suspected my BIOS might be the culprit. Interestingly enough, after trying some mild overclocking, the silly thing locked itself at a multiplier of 6.5, which is of course totally wrong, and the CPU still runs at its designated speed anyway, as confirmed by WCPUID.

As for my Enermax 350W not being powerful enough, all I can say is, Yikes!!! But I will try your suggestion and unplug a few things if the BIOS upgrade does not work. I am a little reluctant to install a beta BIOS, though... Would the 1001.e be better compared to the 1001.g?

 

[mechBgon]

Originally posted by: Palek
Thanks, mechBgon, that was pretty fast!

You notice I don't use the snail avatar

I don't actually have an A7N8X-Deluxe or even an nForce2 board, so I'd hate to lead you astray, but I follow the nForce2 threads closely and there haven't been any catastrophes with the 1002 beta bios that I've seen.

I installed a 1016 beta bios on my old Asus KT333-based board here last night and it's working fine too. Still trying to figure out how to use the multipliers in the BIOS, looks like I've got to unlock the CPU to play with those...

Anyway, here are the threads that come up on a full-text search for "A7N8X" & "1002" in the body, in the relevant forums, if you want to review before deciding:


Thread 1
Thread 2 (linked above)
Thread 3
Thread 4
Thread 5

 

[Palek]

I am having a terrible time with this motherboard so far. I installed the 1002.001 BIOS, then cleared the CMOS just to be safe. Then I went ahead and set my FSB to 166, the multiplier to 12 and the memory to 166 at Optimal timings. My CPU is not unlocked by the way. So, I save, then I save and exit, and some strange characters appear in the middle of my screen. I have to turn the machine off. Nothing happens after turning it on, so I reset the CMOS, go back to the BIOS, set everything properly, meaning FSB at 133MHz and multiplier at 7X/15X, memory at 166 MHz, and now I get "Memory failed test" even with a single stick. I swap them out, the memory test still fails.

So I boot up anyway, download the 1001.e BIOS and prepare for flashing again. I have no better luck with this BIOS either. Failed memory tests just about every time I boot up. I did reset the CMOS after flashing the new-old BIOS.

So I go back to 1002.001 again, reset CMOS, insert 1 RAM, low settings of 133 FSB/15x/133 RAM. This time it works. I load up Win2k, fire up 3DMark2001 for some testing. It fails even before it finishes the first run. It crashes back to desktop, then the machine becomes totally unresponsive. Power off and on again. I swap out the RAM, same thing happens with this one as well.

At one point I did manage to set both the FSB and the RAM to 166 MHz and start up with a 13x multiplier. This was the one and only time I managed to overclock the FSB and the CPU itself past 2400+. But even before Win2k loaded up completely it locked up with a nice "IRQL_NOT_LESS_OR_EQUAL" message.

So, basically my machine was at its best with the 1001.g BIOS and a single stick of RAM in slot one. Right now I have the RAM in the second slot, the FSB at 133, RAM at 133, multiplier at 15x, and I disconnected two case fans and my DVD-ROM. I still have problems, 3DMark2001 just crashed back to the desktop.

WHAT ON EARTH IS WRONG???

 

[Palek]

Update

I finally got everything to work without crashing immediately, and there are no boot-up memory test failures so far, even when using two modules in dual channel mode. I went back to the 1001.g BIOS which, oddly, was the only one that worked well for me. I checked the RAM modules, squeezed them in as hard as I could without cracking my motherboard in half, then even wiggled them a little left and right to get proper contact with the slots. I pressed down on the videocard, too.

First I booted the machine with the DVD and the CDRW unplugged and ran a loop of memory burn-in tests in Sandra, then ran 3DMark2001SE a few times. This time everything worked. Then I went ahead and plugged back the two drives, and ran the same tests again. No crashes so far... Even right now the machine is on, crunching UD data. When I go home in a few hours I will find out how it went...

Another interesting issue... I have an Alpha PAL8045 with a 4000 rpm fan and used Arctic Silver III (closely following the instructions on their website) as the thermal compound. The BIOS reports my CPU temperature at 40 C, and according to the ASUS Probe utility in Win2k, even at full load I only get a maximum temperature of 36 C. Isn't this a little low? Not that I am not happy about that... The temperature did not change much with the different BIOS versions, either.

Anyway, to sum things up, things are starting to work right, or so it seems right now. Hopefully you will not hear me whining again - not about this issue, at least.