Sabertooth X79 Disk Corruption Issues

[jimgoings]

I would appreciate some help troubleshooting a disk corruption issue that I've been battling for a couple of days.

Specs:
ASUS X79 Sabertooth (BIOS version 2.10.1208)
3930K Proc (revision C2)
Corsair H100 CPU Cooler
32 GB G.Skill Rapjaws (8x4GB)
Samsung 830 256 GB (x2) (january 2012 firmware)
WD 2TB Caviar Black
EVGA Nvidia GTX 580
Corsair Obsidian 800d Case
Corsair AX1200 PSU


I followed jaquith's excellent x79 post here: http://www.tomshardware.com/forum/303873-30-wont-resume-sleep-booted-raid#t2068879

1) I made a bootable USB stick with GPT UEFI.
2) Set SATA mode to RAID
3) Removed all cables from other drives (only one drive connected during install)
4) Installed Windows Professions w/SP1 (real ISO from Microsoft, not pirated) - used the latest Intel RST drivers
5) In all cases, Windows has installed without error
6) Once in Windows, I would get erratic behavior when interacting with the disk. Unzipping files might end abruptly, with only some files unzipped. Sometimes installation of drivers would complain about missing files. And about 50% of the time when installing windows, the first reboot will crash with a missing or corrupt system file error.
7) I narrowed down an easy way to identify the corruption by doing a "sfc /scannow" - in all cases it comes back with an error saying that some corrupt files could not be repaired. As a point of comparison, my old rig has zero errors after years of use.

Hardware Tests done:
- Memtest ran overnight (2 full passes of all 8 sticks) - zero errors
- Prime95 ran for 2 hours - zero errors (this occured on one the rare times the OS would load and I could install Prime without hitting a disk error - took a few tries)
- Hardware temps all look good (CPU idle is about 28c)

Other methods I've tried:
- same steps above using the other Samsung 830
- same steps above using the WD Caviar Black
- same steps above using AHCI instead of RAID
- used the non-Intel (black ports) SATA controller
- changed SATA cables
- changed USB sticks (OCZ 16GB and SanDisk 16GB)
- I also tried regular MBR flash drives

I called ASUS (not very helpful) and they advised that I RMA the board back to Newegg.

 

[jaquith]

Q - Which drives are corrupting? SSD or HDD or both.
Q - Are the SSD's in RAID 0 or 1 or separate?
Q - Did you install Intel Rapid Storage Technology enterprise driver software V3.0.0.3011 ; it's a 260MB+ file
http://usa.asus.com/Motherboards/Intel_Socket_2011/SABERTOOTH_X79/#download
/when I did the 'guide' ASUS had the prior 3.0.0.2003 on their site./
Q - What other changes to the BIOS besides SATA -> RAID?

GPT itself shouldn't cause a problem, the HDD's >2TB (2TB+) need the GPT and 2TB can run under MBR.

used the non-Intel (black ports) SATA controller

Both the Brown/Black are Intel X79 chipset.
Intel® X79 chipset :
2 x SATA 6Gb/s port(s), brown
4 x SATA 3Gb/s port(s), black
Support Raid 0, 1, 5, 10
Marvell® PCIe 9128 controller :
2 x SATA 6Gb/s port(s), gray

 

[jimgoings]

Q - Which drives are corrupting? SSD or HDD or both.

Both SSD's and the HDD exhibit the same behaviors.

Q - Are the SSD's in RAID 0 or 1 or separate?

Seperate. I don't plan on using a RAID at all.

Q - Did you install Intel Rapid Storage Technology enterprise driver software V3.0.0.3011 ; it's a 260MB+ file
http://usa.asus.com/Motherboards/Intel_Socket_2011/SABERTOOTH_X79/#download
/when I did the 'guide' ASUS had the prior 3.0.0.2003 on their site./

I used the latest ASUS one and I also used the latest direct from Intel. And to be clear, I've stopped loading the drivers after installing Windows as the OS is already corrupt at that point (as the system file scan shows). I have been using the F6 drivers though.

Q - What other changes to the BIOS besides SATA -> RAID?

I disabled the Marvel controller (don't plan on using it). Everything is default... no overclocking or other changes that I can think of.

GPT itself shouldn't cause a problem, the HDD's >2TB (2TB+) need the GPT and 2TB can run under MBR.

I tried with MBR too. Same issue.

Both the Brown/Black are Intel X79 chipset.

Right you are. I was just confused because of the reference in the BIOS to the ASM1061 Storage Controller.

 

[jaquith]

I recall reading some folks had issues with the (Jan 2012) firmware if (they) updated it them selves. Q - did you update the SSD Firmware?

If it were 'me' and as you know all too well installing the OS is pretty easy, I would perform a 'Secure Erase' and start from scratch; see - http://blog.ocztechnology.com/?p=367 As I recall Magician offers the same options. Pretty much anytime I re-install an OS on SSD's I Secure Erase and write 0's (zeros). Don't worry about writes, at this rate you'd be a old man before there's too many. Downloads - http://www.samsung.com/us/support/owners/product/MZ-7PC256D/AM

Re-install the OS then the Chipset driver first, test before connecting any drives. Use the Intel SATA3 port (brown) and nothing else.

If the problem persists then it might be the X79 Chipset; there was a very minor update from C0->C1 but it's mainly a virtualization fix. However, if it's bad there's no other 'fix' than to RMA the MOBO. If you purchased from e.g. NewEgg they offer 'Expedited RMA' -- essentially you buy a 2nd MOBO but they don't charge for overnight delivery, then you send in the 'bad' component and the CC is credited.

 

[smorizio]

I would pull out all the ram but one stick and see if the errors are still there.could be a memory issue with the mb where the ram is damging the data writen to the hard drive. HAd an issue like that with p35 mb and 1066 ram. the mb posted and ran memtest fine but would blue screen or currupt data on the hard drive. switch to slower 800 speed ram never had a problem.

 

[jimgoings]

I recall reading some folks had issues with the (Jan 2012) firmware if (they) updated it them selves. Q - did you update the SSD Firmware?

If it were 'me' and as you know all too well installing the OS is pretty easy, I would perform a 'Secure Erase' and start from scratch; see - http://blog.ocztechnology.com/?p=367 As I recall Magician offers the same options. Pretty much anytime I re-install an OS on SSD's I Secure Erase and write 0's (zeros). Don't worry about writes, at this rate you'd be a old man before there's too many. Downloads - http://www.samsung.com/us/support/owners/product/MZ-7PC256D/AM

No, the SSDs already had the latest firmware.

I don't think starting over will matter much since I get the same behavior on 2 SSDs and a regular HDD. I setup the RMA with newegg online already. I will call them tomorrow to get the expedited option.

Thanks so much for taking the time to help Jaquith. I really appreciate it.

 

[jimgoings]

I would pull out all the ram but one stick and see if the errors are still there.could be a memory issue with the mb where the ram is damging the data writen to the hard drive. HAd an issue like that with p35 mb and 1066 ram. the mb posted and ran memtest fine but would blue screen or currupt data on the hard drive. switch to slower 800 speed ram never had a problem.

Seems extremely unlikely given I did 2 passes of memtest and 2 hours of Prime95 without a single hiccup. All issues have been related to storage and while you're comment is certainly possible, it's pretty remote.

 

[jimgoings]

Best answer selected by jimgoings.

 

[jaquith]

It's very unusual to have (2) failed SSD's, and if it happens again my eyes would be at both the MOBO (X79 chipset) and/or PSU. On my Sabertooth X79 my HX850 was bad and killed x2 HDDs and a MOBO i.e. PITA. NewEgg is great!

Certainly, I'd question the Windows image ISO file. While I'm not a fanboy of downloading ISO's, these are supposed valid and good clearly you need your registration - http://www.w7forums.com/official-windows-7-sp1-iso-image-downloads-t12325.html In my guide it shows you how to use your store bought ISO or DVD.

Good Luck!

 

[jimgoings]

Certainly, I'd question the Windows image ISO file. While I'm not a fanboy of downloading ISO's, these are supposed valid and good clearly you need your registration - http://www.w7forums.com/official-windows-7-sp1-iso-image-downloads-t12325.html In my guide it shows you how to use your store bought ISO or DVD.

Indeed. I didn't mention it, but I did use two different ISO's AND two different USB keys.

I should have the new board tomorrow if it ships today. Newegg was awesome and issued me a refund as they no longer do replacements on Intel boards due to folks bending pins in the socket. I HATE the socket design as they are too fragile. I bent some pins on the first i7 board I built - not sure how. So now I'm super paranoid about the whole process.

Funny - I mentioned to the guy on the phone at Newegg that they should offer a server to install the procs into sockets for folks that buy both at the same time. He laughed and said that will never happen as they also think the socket design is horrible.

I'll post back here with the results of my new motherboard. Really appreciate the help Jaquith (and others!).

 

[jaquith]

On LGA 2011 I think 'U' then 'V' and then 'V' then 'U' to reinstall. It takes 30 sec or less to install the CPU. Frankly, in the day when the pins were on the CPU 'I' was a lot more nervous installing the CPU...

Yep, NewEgg is very cool!

 

[jimgoings]

New motherboard and I'm having the same exact problems.

- I am now downloading a fresh ISO from Microsoft.
- I am switching USB keys (again).
- Switching to one stick of RAM

I read elsewhere about a user with the same board and memory that had "corruption" issues at the default memory clock. He said that enabling X.M.P. solved it. That sounds fishy to me, but I'm willing to try anything at this point.

Do you think it's possible I have a bad proc? If so, why wouldn't memtest or Prime95 have indicated that?

 

[jimgoings]

BTW - should I open a new thread so I can properly award the correct answer? I wanted to give credit for the help in this thread which is why I marked that best answer.

 

[jimgoings]

ok, using the new parameters:

- new ISO
- new flash drive
- one stick of RAM
- using the rear USB port instead of front
- using ASUS Intel RST drivers (v3011)
- setting X.M.P. mode so memory switches to 1866 Mhz.

Installation of Windows went fine (as it always has), but running system file checker "sfc /scannow" yielded NO ERRORS for the first time ever!

I added the RAM back and am now installing Windows fresh again to see if I can pinpoint the issue (and eliminate the chance that the success was a fluke).

 

[jimgoings]

I believe I have narrowed it down to the X.M.P. setting.

I added all of the RAM modules back in and reinstalled windows with X.M.P. still enabled. No issues.

I then turned X.M.P. off (memory went back to 1600Mhz default) and Windows crashed on installation.

Turned X.M.P. back on and, again, no issues. I'm now proceeding with the rest of the driver installation and hoping for a smooth ride.

 

[jaquith]

First, very odd with the RAM 'typically' if anything it's the opposite.

I would tweak (2) Voltage settings: VCCSA -> 1.20v and CPUVTT -> 1.20v; it's best to have them @ the same voltage.

 

[jaquith]

BTW- I didn't see the other post, and I'm not 'here' for BA's I'm 'gold' x7 or x8+.

I hope that PITA and small change has you running fine. However, IMO test the crap out of the RAM with Memtest86+ and IMO overnight with NO errors.

RAM 'should' run without errors aka no corruption!

 

[jimgoings]

Just following up to close the loop:

Because I don't like to leave things to chance or "mysterious circumstances" I did several installations of Windows and confirmed that corruption occurred every time the memory was set to 1600 MHz. When set to 1866 MHz (via X.M.P. switch in BIOS), everything was fine. I figure this is just some issue with the memory settings on this particular part with this particular RAM.

I ran Memtest all day while I was at work (6 passes) and ithere were no errors.

I've been running great ever since. This rig is blazing fast and I gave my old i7 920 to my 9 year old son! Thanks again Jaquith.

 

[nikorr]

This topic has been closed by Nikorr