Problem with PCI slots

[Josh777]

Hello, I am having a problem with the PCI slots on my motherboard. I will explain:

 

I first found that when I run a stess test in Prime95, I sometimes got errors and sometimes didn't. I soon discovered that, when I was connected to the internet, Prime95 would give errors, but when I wasn't connected, Prime95 ran fine. (I have dial-up.) So I thought the problem was with my modem. I tired another modem, same problem.

 

At this point, I didn't know what was causing the problem. I posted my problem at HardOCP.com, and someone suggested that it might be a problem with my PCI slots. So, I figured out a way to test this:

 

I plugged a PCI USB 2.0 card into the other PCI slot. Then, I plugged my external CD drive into the card. I then tried to install two programs from that CD drive. The PC kept restarting during the installations. Once it even restarted right when Windows spun up the drive, before I even started the install.

 

To make sure it was a problem with the PCI slots, I unlugged the CD drive and plugged it into one of the ports on the motherboard itself. This time, I was able to install the programs with no problems. (Except one of the disks was scratched and wouldn't get past a certain point, but that has nothing to do with my problem.) I also tried putting an older 5.1 sound card in one of the slots. When I went to install the drivers for the card, the installer failed to detect the card.

 

At this point I knew there was a problem with the PCI slots. So, I started trying everything I could to fix it. I updated the BIOS to 6/23, that didn't work. I ran memtest, it went through a pass with only one error. I rescanned the address that had the error, and it passed this time. I checked the CPU temps with Speedfan and in the BIOS. They are at 38C idle and 42C load, very normal. I installed and reinstalled the chipset drivers. Nothing would solve the problem, however.

 

So, I was pretty sure I would have to RMA the board. But then I noticed something in Speedfan. My -12V rail was being reported at -9.40V and my -5V rail was being reported at -0.48V. Considering I have a quality PSU, I'm pretty sure these are just reading errors. But if they aren't, could they be causing the problem? Also, is there anything else I can try to fix this problem besides taking the whole system apart and RMA'ing the board?

 

BTW, I have not overclocked anything in the system. Everything has been at stock since I got it.

[Aurhinius]

have you got ALL the extra power connectors plugged in?

[red930]

Get a DMM and check your PSU rails.

[SADtherobot]

Well, the fact that you got any errors in memtest shows that your system isn't 100% (especially an error on the first pass, you should get 0 errors on an infinite number of passes). This would also suggest that your system wasn't 100% stable when installing Windows, so that may be causing some of your problems. You should tweak your BIOS settings a little until you have no errors in memtest after many hours, as well as no errors in Prime95 after 24 hours, then re-install the OS. To verify that your PCI slots are operating at the correct speed, use clockgen, if it reports that your PCI bus is running at 33.33MHz (or very close to that) then your slots should be fine.

[Josh777]

have you got ALL the extra power connectors plugged in?

 

Yes, I have all four power connectors plugged in.

 

 

Get a DMM and check your PSU rails.

 

Okay, I will do that. Which pins in the power connector carries the -12V and -5V rails, and do I test with the PC on or off?

 

 

Well, the fact that you got any errors in memtest shows that your system isn't 100% (especially an error on the first pass, you should get 0 errors on an infinite number of passes). This would also suggest that your system wasn't 100% stable when installing Windows, so that may be causing some of your problems. You should tweak your BIOS settings a little until you have no errors in memtest after many hours, as well as no errors in Prime95 after 24 hours, then re-install the OS. To verify that your PCI slots are operating at the correct speed, use clockgen, if it reports that your PCI bus is running at 33.33MHz (or very close to that) then your slots should be fine.

 

What exactly do you mean by "tweak the BIOS settings a litte"? Do you mean I need to underclock my system? I bought this board because it was rated the best overclocking board, so something is wrong if it is unstable at stock speeds.

 

BTW, my RAM is set a 2.5-3-3-8 I think and VDIMM is set at 2.6V, but the BIOS says the actual value is 2.53V. Maybe I need to turn on the "boost VDIMM .03V" option in the BIOS?

 

EDIT-Oh, and I have a few things to add to the first message. First off, I don't know if the PC restarting during the installations was due to instabillity or a problem with the PCI slots. Second, I just reinstalled Windows using "format" instead of "quick format". That solved a boot problem I was having, but not the Prime95 problem. Third, twice I've gotten a blue screen when starting Windows. The message only flickers for a second, then the PC restarts and Windows starts. The message wasn't displayed long enough for me to read it, but I could see a hexadecimal code. It reminded me of a "STOP" error, where the word "STOP" was dispalyed followed by an error code. Don't those have to do with the memory?

[SADtherobot]

What exactly do you mean by "tweak the BIOS settings a litte"?

 

I mean try adjusting your memory timings and voltage. Your memory voltage is too low, VX likes to use a LOT of voltage, try using 2.8V, then run memtest86 again, and believe me, there is no way you can put too much voltage into the VX with your RAM voltage jumper in its current position (the most you can put right now is 3.2V, which VX can handle without problems) so dont hesitate to tweak it. At stock clock (200MHz, DDR400) the VX is capable of running at 1.5-2-2-5 timings at a 1T command rate (just as a future reference), but I'd suggest you try to get it stable at stock timings before pushing it to really low latency.

[Josh777]

I thought that my memory was rated for 2-3-3-8 at 2.6V, but okay. I will try memtest again at 2.8V. I don't think that's causing my problem with the modem though. Oh, and I edited my last post before this one, you might want to read that.

 

 

Thanks everyone for all of the help.

[SADtherobot]

I thought that my memory was rated for 2-3-3-8 at 2.6V, but okay.

 

It is, but that's just the "guaranteed to run at these settings on most motherboards" rating, it is very basic and mostly for compatibility purposes (such as for boards that can't give a lot of voltage). On OCZ's website it says your RAM is rated to run at CL 2-2-2-8 at 3.2V (CAS-TRCD-TRP-TRAS). You bought "VX" which stands for Voltage eXtreme, that RAM is made to have a ton of voltage put through it in order to achieve some really nice low-latency, high bandwidth settings. OCZ rates that memory to handle up to 3.5V, so running it even at 3.2V is well within its tolerance. If you had problems when installing Windows or when booting Windows, your computer is definitely unstable and your installation is probably corrupt, I'd make sure everything is completely stable, then re-install the OS. Also, as I mentioned earlier, you can use Clockgen to check the speed at which your PCI slots are operating (it should be very close to 33.33MHz).

[Josh777]

Well, I just ran 2 more passes on memtest and 75% of a third pass with the VDIMM at 2.8V. There was an error in pass 2 at the same exact place it errored last time. It was pass 2, test 6, at the 485.3MB point. What does that mean? Is my system unstable with that address not working?

 

BTW, clockgen says my PCI clock is at 33.34. That means there couldn't be a hardware problem with the PCI slots, right?

[SADtherobot]

Well, your PCI slots should be working correctly, unless there is something wrong with the circuitry on the motherboard (not very likely). Most likely your Windows installation had some problems configuring itself, and when you were adding drivers for other devices, those most likely became corrupt also (which explains the weird behavior). Well it seems as though you will have to try different memory settings (like setting the timings manually, and trying a different tref), and perhaps a new BIOS. There are many guides around here that give good suggestions about how to configure your VX, and you might want to switch to a -2 BIOS, since they are supposed to be better for VX (you are probably running 6/23-3 right now, get 6/23-2 or if that doesnt work, perhaps 5/10-2fix would).

[Josh777]

Okay, I switched to the 6/23-2 BIOS and set the memory at 2-2-2-6 3.2V with the Tref at a number I heard was a "UTT/BH-5 sweet spot". It seems to be stable now, I ran 5 passes of memtest and they all passed with no errors. Another thing I did was, I got my fans spinning at 100%. They were connected to the "fan only" connectors on my PSU, which I did not know would control the fans based on temperature. I connected them to the regular molex connectors. The memory no longer burns my finger when I touch it, so I think heat might have had something to do with the errors I used to have.

 

I reinstalled Windows with the RAM at 2-2-2-6 3.2V. That didn't solve my problem with the modem, but I was able for the first time to install the NF4 drivers with no problems.

 

Okay, so my real question is this: Is there a test I can do to figure out if there is something wrong with my motherboard before I OC it and void the warranty? So far I have had no other problems with the PC, so I don't think having Prime95 errors when I'm on the internet is an issue to worry about. I don't, however, want to keep a motherboard that has something wrong with its's circuitry.

 

When I asked this question at HardOCP.com, I never really got an answer. I really hope that someone here can give me an answer do that I can go ahead and OC my system like I've wanted to since the day I built it.

 

EDIT-Oh yeah, and one more thing. I want to do folding@home on my PC. Since I get Prime95 errors while I'm on the internet, does that mean that data from Folding@home will also be corrupted while I'm on the internet?

[Josh777]

Last night I ran a Prime95 "blend" test for 12 hours. It passed, so my system appears to be stable. (Stable enough for me anyways. I don't know if it would have passed 24 hours, but I never run my PC 24/7 anyway.) I also installed some programs with my CD drive plugged in to the PCI card again, and this time there were no problems. So I don't think it's a problem with my PCI slots.

 

Still awaiting an answer to the questions in my post before this one.

 

Thanks everyone for all of the help so far.