Blue screen woes with new build--need advice

OK, so I managed to install Windows 7 Ultimate 64 bit on my new rig that I put together over the weekend, but have gotten at least 4 BSODs over the course of one evening as I was trying to install drivers and such.

First question: when SATA mode in the BIOS is set to AHCI, is there anything special one has to do before one installs the OS to prep the HDD or something? Isn’t formatting it in NTFS good enough? Is there a pre-OS-install driver to install somehow?

Once I had the OS installed, I popped the ASUS cd in (the motherboard is a P8Z68-V-LE) and installed the Intel chipset driver, the mass storage driver, the Realtek Network driver and the HD Audio driver (although that last one reported failure). Then I figured it was best to download Windows Updates so I did that. My next plan was to install the nvidia drivers for the MSI GTX560Ti card I have installed, but 7zip reported that the checksum failed when trying to extract the drivers from the downloaded installer.

Meanwhile I’m getting BSODs all over the place. A couple of them reporting “page fault in non-page area” and another couple something about “memory management.” Anyway, I’m thinking my big mistake was not waiting to install the graphics card until after I’d installed its drivers. Since the Sandy Bridge CPU (i5 2500K) has its own basic graphics, I think I need to take the 560 Ti card out and make sure everything else works (and the BSODs stop) first.

Meanwhile I’m letting memtest86+ run overnight to make sure the RAM is fine.

Is there a best order to install drivers in once the OS is installed? Also has someone managed to create a slipstreamed disc of Windows 7 with at least SP1 incorporated into it yet?

I know, tl;dr-- any advice/insights would be appreciated.

Windows 7 supports AHCI natively, no need to do anything special. It’s the correct setting for a modern hard disk.

Why are you installing ANYTHING from the Asus CD that’s like 2,000 years old and was unearthed with the Dead Seas scrolls? Get Intel drivers directly from the Intel website, get Asus drivers from the Asus website, and get Nvidia drivers from the Nvidia website. Also check for an updated BIOS on the Asus website.

I don’t know about the interaction between the Sandy Bridge integrated graphics and an external card but generally I’d assume memory trouble when you get page faults. Running memtest is good. Did you make sure your BIOS uses safe default settings? Upclocking RAM is tricky business, even if the RAM allegedly supports the speed you’re trying to use.

Order of installation should make no difference, and certainly not cause BSODs.

OK, good point about the Asus CD. I figured some drivers were better than none till I got a chance to upgrade them. Also going to the ASUS web page for the board, it looks like there are three BIOS updates, including one that released just a few days ago, so the first thing to do will be to update to that, I suppose. Should I just go straight to the latest one?

Anyway, looks like memtest86+ isn’t finding any errors running since I last wrote–it’s completed 7 full passes with the RAM running at 1600 MHz (vs 1333). The RAM is on the approved list and it’s 1600 MHz Corsair DDR3.

I should add the data point here that I got the first BSOD last night while I was still installing Windows 7 (I was right about to enter the product key), so maybe dropping back to the safe default settings for now isn’t a bad idea (at least until I get the BIOS updated to the latest version). Still, wouldn’t memtest86+ show errors if the upclocked BIOS setting (I just clicked on the big button in the GUI, didn’t mess with individual timings BTW) were the source of the problem?

Check the update notes for the BIOS releases and see if they mention anything that might be relevant. If so you should update to the latest version. Any new version might introduce new bugs as well as old ones, but since you’re already experiencing inexplicable crashes you can only try and hope for the best.

I don’t know under which conditions memtest might fail to show a problem, but I would treat it as a tool that can only definitely show problems but not their absence. Definitely also try running at 1333 memory speed and see if the crashes happen again.

Try running Memtest86+ from a cold boot. I had a similar problem to yours back in January when I was building a new Windows 7 system, lots of seemingly random BSoD’s related to page faults, even during Windows 7 install. Turned out the Corsair RAM had a timing issue that only cropped up when the machine was coming off a cold boot. Memtest86+ showed no errors if the RAM was nice and warmed up, but when I ran Memtest86+ from a cold boot suddenly the chips were riddled with errors. Turned out to be bad timing settings on that particular version of the RAM, so Corsiar sent me a replacement set with a later revision number and everything worked fine.

I totally agree with Chris as well, never use the CD/DVD that comes with a mobo or video card. There are always later, more refined versions of the driver software available directly from the manufacturer’s website.

Thanks for the replies, Chris and S.M. I want to do all the things you guys are suggesting, but what’s the best order?

Try memtest from cold boot: how long does one have to wait for the RAM to be “cold”? Is an hour enough time?

Assuming that doesn’t show errors: Drop the RAM speed to 1333 would be the next step, I presume, or should I update the BIOS first?

I don’t know if you’ve done this already, but have you checked to see if in the bios, the correct timings and voltage for the memory are set?

Also, how many sticks of ram are you running with? It’s also recommended to just test one stick of memory at a time, and if possible in different slots to see if its possible the ram slot or the ram that could be broken.

I think what he mean’s by cold boot, is to shut-off the PC and then power it back on after a minute or so instead of using the Reset button or rebooting through the OS.

BTW, looking at the last three updates for bios all mention system stability and memory compatability, I’d at least update the bios to the latest version before running the tests again.

OK, this is super annoying. Turns out my RAM kit is NOT on the officially approved list. I got the following two lines confused–the top one–2 sticks of 4 GB, is for the much more expensive “Dominator” variety, and the bottom one is 4 sticks of 2GB, of the kind I got (XMS3), EXCEPT that I got the kit with 2 sticks of 4.

CORSAIR CMP8GX3M2A1600C9(XMP) 8GB(2 x 4GB) DS - - 9-9-9-24 1.65V
CORSAIR CMX8GX3M4A1600C9(XMP) 8GB(4 x 2GB) DS - - 9-9-9-24 1.65V

Man do I hate stuff like this. Why the hell wouldn’t ASUS test my stuff when it makes so much more sense to put 2 sticks in vs. 4 sticks? Is going with the officially approved RAM super-important? Also, I’ve read somewhere that the 1.65 voltage is higher than recommended for the Sandy Bridge processors.

Why do I never have these issues in building out a system? It’s a given that most of the machines I touch are mostly pre-built with minimal mucking around, for work, but there’s still quite a few unique builds I do that I part out completely and stuff just works. Go figure.

No, unless you’re pushing overclocking to the limits.

If you got the BSOD during the Windows 7 install, it’s BIOS upgrade time. If that fails, then 99% you have some bad hardware.

Updating the BIOS is my first priority. Just now, as an experiment, I rebooted the system (after it had been running memtest86+ since around 11:45 last night, 13 full passes at 1648 MHz with no errors) and set the RAM to run at 1333, then booted into Windows 7. I was only on there for about 10 minutes, but I managed to install avast antivirus free and register it, and no BSOD at all. So if the BIOS update doesn’t fix it so I can run that RAM at 1600, I’m going to try with some different RAM, already ordered. If that RAM works, I’m returning the stuff I have now to Amazon (still within the 30 days). If it doesn’t work, I have to suspect the motherboard itself (especially if I temporarily take the video card out of the equation entirely by removing it from the machine for now).

Update: I’m happy to report that updating the BIOS seems to have helped a lot, maybe solved the problem entirely. Once I had updated it, I ran Windows with the RAM running at 1333 MHz for an hour or so while I dinked around installing utilities and drivers and such, and after that I went back into the BIOS, enabled XMP (which boosted the RAM speed to 1600) and prayed while I started Windows back up. Ran it for another good hour or more while I installed more stuff and did a bit of web browsing. Nary a BSOD. Here’s hoping everything keeps going swimmingly. Can’t wait to play some Skyrim on this baby. Should I use IntelBurnTest first or something?

Congrats, that must have been some small bug in the memory controller that was fixed by the BIOS update. You could do a short stress test just to check that CPU cooling is working properly. Those tests don’t make your system any better, though; they just check for weak spots and age the hardware.

Thanks. One last concern I have and wonder if I can do anything about: since I had a BSOD during the installation of Windows 7, is there a way of checking the integrity of the installation (something akin to Steam’s “verify game cache”)?

This is why I loathe building my own box. :/ The mobo/CPU/RAM/BSOD troubleshooting cycle of death.

Not sure, but I believe the Windows setup program has a feature to verify the installation.

Thanks, Chris, I’ll check that out.

One other thing–this XMP option I described above was in the Advanced BIOS (vs. the EZ version). The first time I’d set the RAM to 1600 MHz I’d done it through the EZ mode’s big “optimal performance” button (the rightmost of the three). I don’t know if doing it one way vs. the other makes/made any difference or if it was the BIOS update that fixed things (probably the latter)?

“Optimal” performance usually means some (slight) overclocking and aggressive timing, so yeah, you should avoid that option until the system is set up and running stable. The Intel DX58SO motherboard I have slightly also overclocks the system by default. That seems to have become fashionable in recent years.

The actual real world difference between RAM running at 1333 and 1600 is so small I don’t understand why you’d even bother bumping to 1600 even with the updated BIOS when such a setting gave you BSODs in the past.

You probably won’t even get a single full extra frame per second from the higher memory timing, is an extra 0.5 fps worth the uncertainty in stability?