Gigabyte GA-Z68XP-UD3-iSSD RAID Issue

[beano311]

I have a brand new GA-Z68XP-UD3-iSSD (the one with the 20GB mSATA SSD module), two OCZ Vertex 3 120GB SATA3 SSDs, two WD Black 2TB SATA3 drives, and a LiteOn Bluray burner. Motherboard BIOS is updated to the latest version (F5) and the SSD firmware is also up to date (2.11 I think).

I'm trying to RAID1 (mirror) both sets of drives and then use the mSATA SSD to accelerate the 2TB RAID.

First I tried having the SSDs on the motherboard's Marvell RAID controller and the two HDDs on the Intel controller. That wont work because the BIOS doesn't like it when both RAID controllers are in RAID mode, there's some error when you try to leave the BIOS about "too many ROM operations could cause certain ROM operations to fail." I continued anyway, set up both RAIDs, one in each controller, but couldn't get the Windows 7 install disk to boot. I put the install files on a USB stick, got that to boot, but all of the drives said Windows couldn't be installed to them.

So, now I have the two SSDs on the two SATA3 ports on the Intel controller, the two HDDs on two SATA2 ports on the Intel controller, and the Bluray drive on the Marvell side. The Intel controller is set to RAID (XHD) mode and the Marvell is set to AHCI. That works, Windows installs, no problems at all. I get everything installed, drivers, etc. I go into the Intel Smart Response Technology manager (ISRT), enable acceleration (caching on the mSATA SSD) on the 2TB RAID and reboot. Everything works fine for an hour until the ISRT manager pops up a warning from the system tray saying one of the drives in the RAID has failed. I look and sure enough, the first SSD has failed. Windows is running fine and everything, so I figured it'll rebuild the drive and everything will be fine. Shortly after that while an application is installing, the PC shuts down (regular shutdown as if I hit Shutdown on the start menu, not a hard shutdown). So I think okay, Windows Update was running, maybe it forced a reboot because it said "Installing Updates..." on the shutdown screen. It reboots, but when it comes back, both SSDs are now reading as failed in the Intel RAID controller and Windows wont boot. I thought maybe something was up with the RAID controller handling three RAIDs, maybe it was having trouble.

Then I reinstalled Windows, did drivers and the Intel utility... the two 2TB HDDs come up as uninitialized, so I initialize them. That takes about 6 hours. This time I didn't enable acceleration. Sure enough, 20 minutes after the drive initializes and I format it, one of the SSDs fails (the same one). I tried rebooting right after it popped up, thinking maybe I could restore it from the Intel RAID controller... nope, both drives came up as failed.

Does anyone have any ideas as to why the RAID is failing?

[Aussie Allan]

    Welcome Beano

   I know the the Intel controller will handle 2 raids......the third I'm not to sure......re-read Manual

  2...you may have a crook SSD (number 1 as you described)... Can you Secure erase and test?

  3...Try taking XHD out of the loop for the time being ...... go to straight raid instead in BIOS.....XHD has caused problems with early BIOS versions on new platforms

  4... optical can give funny issues in ACHI again with early BIOS versions....Default back to IDE for the time being

  Of course after all tese canges a fres install will be required........Sorry!

  For my own personal...... can you confirm what Marvell controller is taking care of the 20GB  iSSD..... I suspect it is the 9130 Marvell controller chip (88SE9130)........or 28.....maybe....but doubt

  With anything high end and cutting edge......you just have to expect teething problems.....but when you crack it and it's up and running 

 Aussie Allan

[beano311]

I will try again with XHD disabled.

My first guess was that SSD may be bad, but why would that cause the whole RAID1 to fail entirely? Any suggestions as to how to test the drive?

Thanks!

[EyeDontKnow]

I am watching this topic, as I am about to order this board, and attempt a RAID 10 setup with 4 x  750gb drives -- along with a 120gb SSD drive for the OS and applications.
Plus one large outboard drive for complete backups (Acronis).

Looking for someone with a similar setup. (if you have any good  links, please post).

I won't hijack the thread...I'll post my own, as my build comes along.
EDIT: http://forum.giga-byte.co.uk/index.php/topic,6457.0.html

Randall, USA

[Dark Mantis]

Hi EyeDontKnow (or should it be YouDontKnow?)  and welcome to the forum.

I am sure you will get plenty of help here from members with your motherboard. I will keep an eye out for your thread when you start it.

[Aussie Allan]

I will try again with XHD disabled.

My first guess was that SSD may be bad, but why would that cause the whole RAID1 to fail entirely? Any suggestions as to how to test the drive?

Thanks!

Hard to say why they both failed till we do a bit of fault finding wit good components.....  I think all manufactures have there own generic test suite or you can use "HDDSCAN"..... the new version 3.3 will eat everything from IDE to SSD....even Raid!........have you been able to look on the board for the iSSD controller chip tag as yet for me?

  Aussie Allan

[beano311]

The Marvell chip is 88SE9172... I can't find that chip at all on Marvell's site and I can't actually see the chip on the board because it's covered by the heatsink over the Intel Z68 chip, it only lists it as an 88SE9172 in the manual. I also can't find anything on the Intel RAID controller other than it's a Z68. The iSSD is on the Intel controller, not the Marvell.

The way I have it set up now, I have the SSDs and two HDDs on the Intel controller set to RAID mode with the SSDs RAID1, the HDDs unRAIDed, and the Marvell controller set to IDE and with just the optical drive on it. I don't know if the SSD is bad now as it seems like it's the second drive that's failing now, before it was the first drive. It's also not completely breaking the machine as I'm able to restore the drive and Windows doesn't crash; however, it does still keep failing. I'm also now using the latest IRST drivers from Intel instead of Gigabyte's drivers.

I tried HDDScan, the drive names came up as random characters for the SSD and blank for the HDD with little to no information in any of the other spots. I ran a read and a verify test anyway and they completed without errors in a few minutes.

Edit: I ran the trial version of HD Tune Pro and did a bunch of bench marks and all the different tests... no problems there. Also, since I've changed it to only the SSD being RAIDed, I haven't had one of the disks fail. I just enabled RAID1 on the 2 HDDs now from the Intel RST in Windows and it's initializing now. Just realized that my RMA period for Newegg was up today, so hopefully if I find anything wrong it'll be covered by Gigabyte's 3 year warranty.

[Dark Mantis]

Hi

The Marvell controller and the Intel are both capable of running RAID.

What you are asking though regarding the use of the SSD to act as a cache and speed up the RAID0 is quite possible I would have thought if the SSD was faulty or sent corrupt data to the RAID it would indeed damage the RAID. The trouble is that this is all so new that not a great deal is known about it and not a lot of people have had experience of this setup.

[Aussie Allan]

He ain't whistling Dixie ..... I had similar problems setting up my X58 board with 4 drives on a PCI-E raid card, a Revodrive x2 and several other drives splashed around inside,.... something that might help you though is I had to do it one piece at a time with the revo first.......and once I added other raids under raid in BIOS......once stable switching to ACHI.......I suspect this freed up option rom space to effect the rest of the set up ... I had to go backwards and forwards a few times.....but got there in the end.

  Thanks for the heads up on the Marvell as well......entered into the data-base

  Aussie Allan

[beano311]

So, it looks like the RAID degradation issue is fixed, I guess turning them on slowly, one by one worked... however, now there's a freezing issue.

It's completely random, or at least it seems to be. Windows will freeze, but the cursor can still move for about 20 seconds, then that freezes too. When I try to shut it down by hitting the power button (just tapping it to evoke a normal, soft Windows shut down, not holding it to do a hard shut down), the HDD activity LED on the case blinks, like it's actually shutting down, but the picture is still frozen. Not sure what the issue is now, but I'm going to have a look at the Windows system event log.

Edit: I shouldn't have said anything... Disk 0 in the RAID1 SSD array failed. IRST won't "reset disk to normal". The freezing that I mentioned above happened again. First the cursor froze for 5 seconds, then came back and everything was fine. A minute later, I went to resize a window and the cursor changed to that normal double sided arrow, but it stayed like that after I moved it away. Then I went to click on a URL hyperlink and the cursor changed to the normal pointing finger. It was working even though it was an odd icon. Then it changed to a weird, artifacty cursor and shortly after that the whole PC froze. When I powered it back on, the first disk in the array came up failed and won't restore, just like it was doing a week ago.

Earlier, I ran it for a few hours through different benchmarking software to see if I could make it crash, 3DMark to work the CPU, memory, and video card, and PCMark which tests everything, including hard drives. It ran through everything okay, going through every PCMark test over and over for an hour and a half and was running fine for an hour and a half after that. The failing seems to be completely random and I've done HDD tests, so I'm not sure what to do...

Any further advice? Could it be something with the cache not syncing correctly on the drives causing it to fail when the PC freezes, should I have "write-back cache" enabled? Also, there are no errors in the system event log either. Thanks for the help.

[Aussie Allan]

 Now we have a system up,........ and have established the board will work with several drives and raid active..........we need to start ticking off a few things from the most likely..........first one is you guessed it.....Memtest..........one module at a time ....in your boards slot one..........10 passes before swapping for the next one ............this is the procedure to absolutely rule a module  good and not fail under most conditions

  Aussie Allan

[beano311]

I ran Memtest86 on the first stick of RAM and it made it through a pass successfully but when it finished the first pass, the screen turned to mostly gibberish, I attached a picture. It looks like it's still running, time keeps counting and the progress keeps advancing, but I thought that was weird. I tried a second stick and got the same gibberish. Any ideas why it's doing that?


EDIT: I looked up the issue on the Memtest86 forums and found that it is most likely cause by system memory/RAM being incorrectly mapped to video memory or vice versa. This seems like it's exactly what the issue is as I've notice video issues right before the PC crashes (as I wrote about in my last post). The post on the Memtest forums can be found here: http://forum.canardpc.com/threads/56838-%21u7U%21-weird-symbols-but-no-apparent-errors?p=4606208. If it turns out that is the issue, what is a good way to test it further and find out what exactly is causing the incorrect memory mapping?

Also, thanks for all your help Allan!

[Aussie Allan]

 "Huston" ! ...."We have a problem"...........Well it's actually a good result.......better then finding nothing towards fixing the root problem.

  In theory I think your detective work( link) is connected............What memory have you installed ? (Full description please) and was it bought as a matched set DUAL channel set/s

  Are you using DDR3-1 slot for memtest ( the white one furtherest from the CPU)

 Does it list the memory as compatible in the gigabyte website ( it may not be listed but still supported) try and Goggle your board and memory and read a few hits to confirm compatibility.......and come back with the information requested.

  Aussie Allan

[beano311]

I was using slot 1, tested multiple sticks with the same garbage text issue.

The memory that's in there is G.Skill Ripjaw series 16GB (4 x 4GB) quad channel kit, model number F3-14900CL9Q-16GBXL. Not much on that board/memory combination, however a few reviews on Newegg said that they used that kit with the same motherboard. I have the memory clock at 1600, which is what the BIOS defaults to, but the sticks are rated at 1866MHz (which is what the motherboard is supposed to be rated for too). I have been experiencing the same Windows crashing issue with the clock at both 1600 and 1866.

Gigabyte doesn't list that specific model as supported, but it does list other Gskill Ripjaw 1866 dual channel kits.

[Aussie Allan]

 Beano311

 Right .....I'll get DM to check over all the posts to confirm i missed nothing......could actually be a crook board except for one thing we did not check..... and thats the LGA pin bed.......you'll have to carefully remove CPU and cooler...........with a Macro photo or a good light and magnifying glass......look for any pins out of place........there pretty small.........might even look like a shadow where it shouldn't be!.........if the Macro (if you go that way) are good.....post them up.....we're  all pretty OK looking at good and bad LGA seats/beds..........post them at a good res so we can zoom in and have a good look........mean while I'll get the moderator to have a squizzie.

 Aussie Allan