Data corruption (CRCs) during install from CD's

L31mr0d · 13-07-2006 03:04PM #1

Ok i've had this recurring problem since I built my new PC. Whenever I install from a CD or a CD image to my computer, the installation always has corruption issues. Some CD/Images allow me to skip the file others just crash out. Heres the weird part, the actually files on the CD aren't corrupted, as I've tried installing them on my other computer and they install fine. Also, when I retry to install the CD, it doesn't always find the same file that is corrupted, it will randomly at some point during installation give me a CRC error. Some times after a few attempts I can get the CD to install all the way through without it spitting out an error.

I've dealt with PC problems all my life and I can usually troubleshoot them without assistance but this one has me stumped as sometimes it works and sometimes it doesn't. I've narrowed down that it isn't my CD drive as images are causing the same issues. My thoughts are that it could be a problem with my HDD (presently 2xWD in a RAID 0), a RAM issue or a problem with my NIC as I get issues also if I download a rar archive and try to extract it.

Has anyone encountered this problem, also is there a way of testing my HDD/RAM/NIC? I've tried stress tests on the RAM but its passed all of them.

EDIT:
I have an Asus A8N-SLi mobo (nf4), 1GB TwinMOS pc3200, 2 x WD Caviar (RAID 0)

SwampThing · 13-07-2006 03:18PM

Well, if you're having this problem with images from CD as well as downloaded files, I'd say you've ruled out your NIc as an issue.

If the PC is otherwise stable and runs appliations/games etc. error-free for long periods, then I'd rule out your RAM as well.

So, your RAID - can you give a little more detail - raid controller, disk spec's etc.

L31mr0d · 13-07-2006 03:40PM

Its a Hardware RAID 0, setup with the mobos onboard RAID controller in slots 1 and 2. It reads as "healthy" in the BIOS, and other than this niggling CRC problems, everything runs fine.

My AMD 3700+ CPU is OC'd from 2.2 to 2.8Ghz, my RAM is Dual Channel at 2.5-3-3-8 @ 1T command rate

I've a feeling it probably is the RAID thats at fault, but how do I test it?

SwampThing · 13-07-2006 04:05PM

I think you're right about the source of the problem.

Can you pop a stand-alone drive - even an IDE - in there and see if you can work with it?

This may or may not prove anything - if the OS uses the RAID for caching temp files etc., you may still get errors.
You could try this maybe with a RAR file and extract it to the same disk - you can configure where RAR stores temp file - again, save these to the stand-alone disk.
The drastic thing to do is clone your OS onto a stand-alone disk and remove/rebuild the array. You could try cloning the OS partition anyway - see if you encounter any read errors. Try a defrag in the meantime.

8T8 · 13-07-2006 07:32PM

It's probably not the RAM but give Memtest+ a blast anyway just to be sure.

Try the 2T command rate setting in the RAM timings.

If you can change your IDE/SATA cables in case one of them is defective.

Asus + nForce4 SATA = problems is a common affair seeing as you are using RAID then you already have NVIDIA's SATA drivers installed so they cant be removed.

What I would suggest is you update the bios to the latest version and be sure to clear the CMOS afterwords & then setup the bios settings again.

Install the latest chipset/IDE/SATA 6.86 drivers from NVIDIA at this link. (when prompted for the NAM say no) A corruption issue to do with the nForce 4 SATA and ethernet systems was fixed in newer drivers.

Disable NCQ support (if present) on your hard drives the option to do so can be found in the driver properties of the NVIDIA SATA controller.

Try forcing SATA-150 mode if SATA-300 is supported by the HD's.

If using the NVIDIA ethernet connection and the NAM (Network Access manager/Firewall) remove the NAM even if using the latest drivers.

hopeful · 13-07-2006 08:42PM

L31mr0d wrote:

My AMD 3700+ CPU is OC'd from 2.2 to 2.8Ghz

I've a feeling it probably is the RAID thats at fault, but how do I test it?

I'm going to guess that it's the OC thats the issue. RAID controllers really do not like being run too fast and can corrupt data very easily.

My 2c

L31mr0d · 13-07-2006 11:51PM

hopeful wrote:

I'm going to guess that it's the OC thats the issue. RAID controllers really do not like being run too fast and can corrupt data very easily.

My 2c

I don't think its the clock, but then again I can't really pin down where the problem lies. I was getting corruption since I built the system, before any OC'ing. I looked into the NIC for NF4 boards and they are crap, the first thing I did was uninstall NAM and turned off TCI/IP Offloading (causes corruption aswell) which stopped Steam constantly disconnecting and any large images or .exe downloads being corrupted, I just installed the latest nvidia drivers today but I couldn't actually see anywhere if the latest ones fix the problems with Nvidia NF4 NICs?

I didn't install the Nvidia SW IDE Drivers as they are ALSO known to cause corruption. Plus coruption caused by either the RAM or the CPU should show up under stress testing with Prime95 or OCCT right?

Apart from that, I was wondering actually if my FSB(HTT) could be adversely affecting the RAID. At present its at 255Mhz, up from 200Mhz (DDR400). Is there anyway of actually testing this though, to see if my OC's is ALSO causing corruption. I was under the assumption that if an OC was going to cause corruption, it would cause it all at once, and not just CD/Images. (BTW I have no problems installing from large .exe's)

8T8 · 14-07-2006 09:10AM

The fix for TCP/IP offloading first appeared in the previous 6.66 release I believe, the 6.86 release contains some newer fixes for some other stuff.

Because you used NVIDIA's RAID setup you do have NVIDIA's IDE drivers installed and they cannot be removed without dismantling the RAID array.

What I would do is install the NVIDIA storage drivers from the 6.86 release and it should update the drivers for your SATA/SATA RAID and PATA(IDE) controllers, after that roll back the PATA controllers driver to standard dual channel PCI IDE controller but leave the NVIDIA storage driver in place for the SATA controllers.

That way you gain the fixes in NVIDIA's SATA driver they have made but also avoid the compatibility issues with CD/DVDRW drives with the PATA controller.

Don't forget to try disabling NCQ support as well if your hard drives support it.

L31mr0d · 14-07-2006 11:11AM

My HDD doesn't support NCQ or TCQ. So it's disabled by default under the drives in the Nvidia NF4 SATA Controller. I did find that under disk drives in my Nvidia Stripe that tagged queueing wasn't disabled. Afaik this is TCQ. I'm not quite sure what exactly disabling this feature does as I can't get a lot of info on it.

Is there a way of testing the HDD now. Like a program that would move a large file around the disk until its corrupted?

8T8 · 14-07-2006 12:11PM

Not sure of a specific program to do that most people just copy lots of data around manually and see if it gets corrupted. Use an MD5 or SFV checker to calculate the checksum of a known intact file, move it around a bit and see if the checksum changes due to I/O corruption.

TCQ = Tagged Command Queueing really the same thing as NCQ I cant recall whether it actually helped on Western Digital's drives I'm leaning towards I don't think it did but you can disable it alright to see if anything improves.

Also check the event viewer in control panel/administrative tools under the system section and look for anything called delayed write error or anything flagged to do with the hard drives.

Edit: And do test without over clocking your CPU just to rule out that is the source.

L31mr0d · 14-07-2006 04:57PM

Ok i'm going to try that, do I just move the file around the disk or do I need to move it off the harddrive then back again. I could copy it across the network, but I don't want to use the NIC for anything just to be sure that it still isn't causing some of the corruption.

Also, from reading around. SATA ports 1&2 are unlocked and can be affected by an increased FSB above 240 (mines at 250 :eek:). So i tried changing the SATA connectors to ports 3&4 (which are locked) and then I changed the RAID settings to boot from them, but it wouldn't boot. It gave me an error that it couldn't find the file to start the OS, it was either missing or corrupt. So, in a cold sweat, I put the connectors back to 1&2 and it booted fine. Is there a way of changing my RAID over different connectors.

Also, do you have any info on what the benefits of the new SATA drivers are? The nvidia site is vary vague. Also how do I seperately uninstall the PATA drivers without uninstalling the SATA ones?

8T8 · 14-07-2006 06:42PM

NVIDIA are usually very secretive about what is fixed in their chipset drivers and never really gives away too much information especially on the hot topic of I/O corruption for instance when they silently fixed the TCP/IP issue, people only found out because someone from NVIDIA told the Tech Report.

Don't have any info of the status of overclocking and the SATA ports I would have thought some sort of lock that was in place to stop that kind of thing I'm kinda surprised if there isn't at least a bios option. I don't believe you can just switch your RAID array over to the 3 & 4 ports like that at least not without a lot of pain and suffering, check the NV RAID manual in case they offer some sort of "easy" RAID migration feature.

After you install the NVIDIA storage driver it will install the driver for SATA & PATA controllers. You will have an nForce 4 PATA controller in your device manager all you have to do is bring up the properties and click on roll back driver then select "Dual Channel PCI IDE Controller" from the list, wait and then reboot when prompted.

Data corruption (CRCs) during install from CD's

Comments