Building/Upgrading a PC for TS2
KatEnigma:
Why not just run MemTest of the RAM?
Yeah, I was Not Amused that i had to replace the mobo, and that I hadn't really looked into why my game was running so poorly until after I couldn't exchange it. :P
Hegelian:
Sorry to hear about the BSODs. I am doubtful the hdd is the problem; it just doesn't seem like the kind of error a bad hdd would produce. Do you have startup problems, like the "no system disc error"? I have had only two drive failures over the years, both recent: The hdd in this laptop tested okay on the self-test (in BIOS) but had recurring problems with not being recognized as the system disc, or generating missing or corrupt Windows system files. Sometimes it would still boot, but toward the end it took forever to get past the Windows logo screen. But when it did boot into Windows, it ran with no problems as long as I did not shut down or let it go into standby.
During the holidays I had my old DOS/Win98 machine set up briefly for some maintenance—it has been mothballed since I moved here two years ago—and started getting a S.M.A.R.T. warning that one of its old Western Digital IDE drives is failing. I picked up a replacement for cheap on eBay (I didn't want a big new drive since the failing drive is a combination of FAT16 and FAT32), but have not installed it yet. The reason I mention this is because the drive itself shows no symptoms of having a problem. So I have one failing drive that tested okay, and a drive that reports imminent failure but is not symptomatic. :P
You didn't mention above the specifics of the RAM you bought. This seems a more likely culprit than the hdd. This may be stating the obvious, but have checked the BIOS to be sure the RAM is set to its specified settings? We had a problem here when I upgraded Reggikko's PC with a new motherboard and CPU that run with an 800MHz FSB without considering the RAM, and the PC2700 she had just wasn't stable at that speed. Some enthusiast motherboards tweak their default speeds for RAM and the CPU, so you may want to be sure the RAM timings in the BIOS are what the RAM manufacturer specifies. As you probably already know, you can test your RAM by running just one module, and then a pair, and then if you don't have problems, doing the same with the other pair (it is possible your motherboard is one that requires running one, two, or four modules, but not three). If you find one that appears to be bad, you can run one of the other modules in that socket to see if the problem is the module or the socket. MemTest is worth a try, as well.
Motherboard. Sadly, even the best motherboards have relatively high rates of defective units and failures, compared to most other computer parts and to virtually all other consumer products. I've had a couple go bad on me, one a highly-regarded ASUS P2B-S which developed a problem that we only diagnosed a few weeks before the warranty ran out (the vendor made the exchange—ASUS's customer support was abysmal); and the other was a Gigabyte board (also with the BX chipset iirc) that I got as a closeout from the old TC Computers and which was never really right, constantly returning corrupt Registry errors on boot (in Win98) and loading the previous good version, so that whenever I made any change to the system, I had to manually save the Registry several times in succession to be sure that on the next boot, the restored Registry would have my current configuration. Also, I had a friend whose new motherboard had a parallel port that was DOA.
So once you've eliminated the possibility of driver issues, I would think RAM and then motherboard as the next most likely culprits, followed by a defective video board, and then. . . . ???
jsalemi:
Yes, I got new RAM, and made sure it was on Gigabyte's list of supported types, so that's not the issue. I did find out the issue though, so far -- it does look to be that the new HD 3850 failed. I started having a problem with an unknown PCI device this morning, and after booting with nothing else in the system but the video card, still got it. I have a 600w power supply, so I know there's more than enough power to spare for it. So I put my old X1300 back in, and the problem went away.
Now that's not saying that the mobo is perfectly OK, but the problems did start a day or two after I installed the new video card. Since the system as a whole is new, I had to do a process of elimination to find out the problem, not believing it could be a brand new video card causing it. So it goes.
Anyway, I got a RMA from newegg, so it's going back tomorrow. So far the system has been stable with the old video card -- not a single BSOD since I swapped them. So cross fingers and hope it's only the card and nothing else. (It's an HIS version of the 3850, BTW.)
Hegelian:
This is no surprise, really. Video boards and motherboards are probably the least reliable components in a PC. The first Radeon X800GT I bought (a Sapphire model) failed within two days of installation; it's replacement is running fine in Reggikko's machine some three years later. EE friends tell me that if a PCB component is going to fail, it will most likely fail within 72 hours—not that they can't fail over time, like that P2B-S which ran fine for about seven months before it started burning out video boards (the AGP voltage on that model was out of spec and would fry boards that couldn't handle the difference), and then about 11 months out another known design flaw in the north bridge became symptomatic and I had to RMA it.
Incidentally, the north bridge problem only manifested itself under stress, and I was using Photoshop a lot at the time. It was actually a Photoshop engineer on Adobe's forums who informed me of the problem, which had progressed from Photoshop to Word to Netscape over time—the PC would simply lock up, requiring a cold reboot. Apparently, Photoshop stesses your hardware more than most applications, and it seems that once it "triggered" the problem, the effects gradually spread to less demanding applications. This was when I learned that lockups—as opposed to crashes—are always hardware-related (driver problems can be considered hardware problems).
jsalemi:
Yea, that sounds about right -- the video card failed within 36 hours after I installed it. The thing that told me it probably wasn't the mobo, CPU or power supply is that most of the BSODs would happen when pretty much nothing at all was going on on the machine -- maybe I was checking email or websurfing, but not really putting any stress on it. And to be sure, I turned off all the automatic overclocking the Gigabyte supports, bringing the BIOS down to bare basics, and still got the BSODs. And they happened randomly, with no particular one software cause I could isolate.
What drove me nuts is that it wasn't the same error message two times in a row -- sometimes it was the 'paged in unpaged memory' error, other times it was the 'irq not equal', and other times some obscure message that I had to search all over the net to find. In all cases, research said the errors were hardware problems, most likely memory, and since no particular driver ever showed up in the BSOD, it was a process of elimination. I swapped memory in and out, and changed the slots, and it still crashed. I even put a volt meter on the video card connector from the power supply to make sure it was providing the proper voltage. I didn't get a real clue until the 'unknown PCI device' started showing up in the device manager, and the only PCI device in the machine at the time was the video card. So it goes.
So for now it's back up and running, seems to be stable, so I'm crossing my fingers and hoping the failing video card didn't take anything else with it. :P
Navigation
[0] Message Index
[#] Next page
[*] Previous page