MARVELL SATA3 problem

[Dark Mantis]

Remember Microsoft Vista? I beta tested that for well over 24 months,  before they released a proper working version renamed Windows 7. I didn't get any thanks and was still charged full price for this wonderful replacement OS.

Of course, Gigabyte isn't as big as Microsoft so maybe they will be a little more understanding......... 

Zoom! There goes another one(flying pig of course)

[etwolf]

Just keeping this thread current.
 

[Dark Mantis]

Remember Microsoft Vista? I beta tested that for well over 24 months,  before they released a proper working version renamed Windows 7. I didn't get any thanks and was still charged full price for this wonderful replacement OS.

Of course, Gigabyte isn't as big as Microsoft so maybe they will be a little more understanding.........  

Zoom! There goes another one(flying pig of course)

Actually I have heard that they now have a whole squadron of them. Maybe it was easier to make flying pigs than fix the SATA3 problem

[absic]

OK I'm really ticked off about this now. I know, I'm running an AMD platform and this is in the Intel section but on my board I have the Marvell 9128 chip so I'm staying here until I get a resolution to this issue.

I got hold of my current motherboard, the GA-790XTA-UD4 primarily because of the SATA3 capabilities. I invested in 2 Western Digital 640Gb SATA 6/Gbs hard-drives so that I could run RAID 0.

I got up early and, feeling experimental I thought I would try and resolve this issue. I reconfigured my PC and have just spent over 8 hours trying to get a RAID 0 Array to run on the SATA3 Marvell Chip without success.

I have now tried in IDE Mode.
I managed to install the OS in only 9 Minutes 30 secs.
The PC decided to hang on 2nd start-up.
Rebooted, everything seemed fine installed Mobo drivers, Rebooted, OK.
Installed Graphic Card drivers, rebooted NO RAID Array found in BIOS. Shutdown completely, left 45 seconds rebooted OK.
Installed Sound Card Drivers, Reboot OK.
Checked Windows Experience Index found Hard-drive rating to still be 5.9.
Downloaded and Installed Crystal Disk Mark to test read/write speeds. System locked up after 1 read pass.
Reboot, RAID Array missing in BIOS.
Shutdown, leave for 45 seconds, restart, RAID Array back.
Tried to re-run Crystal Disk again only for the system to lose the RAID Array again and freeze up.
Went thorugh the process of shutting down & Re-starting, got back into Windows after 3rd attempt, opened Google Web browser, system locked up.
Rebooted, RAID Array missing in BIOS.

Tried re-installing OS in AHCI mode. Same problems as for IDE Mode

Disabled all the other SATA ports on my Motherboard. Tried again same situation in both IDE and AHCI Modes

Loaded Optimized Defaults in BIOS, in case it didn't like me running my RAM at 1600 Mhz. Repeated above steps, same thing happened.

I tried setting the SATA3 Firmware Selection to On-Chip and AUTO both gave the same results.

GAVE UP TRYING TO GET IT TO WORK AT THIS POINT! And yes, I did install the correct Marvell SATA drivers as needed.

Now, I do not consider myself a newbie, although I'll admit, I do not have a lot of experience with RAID. I have been logical and practical in my approach to getting the Marvell SATA3 controller, set-up as RAID 0, in my system to work and have been unable to do so. I do not think that I am at fault, unless anyone can tell me otherwise, so I have to conclude that there is a problem with the Marvell Chip when it comes to RAID.

In IDE Mode and running the two Hard-drives there are no problems and the system is stable.

Would someone from Gigabyte like to please tell me where I am going wrong or, better still, would you like to come and configure my system for me and make it work properly in RAID 0 on the Marvell SATA 3 Chip?

Current System Specs:
Chassis: NZXT Tempest Evo
Mobo: GA-790XTA-UD4 (BIOS F4a Beta)
CPU: AMD Phenom II x6 1090T BE 125W
Cooler: Noctua NH-D14 Air cooler
RAM: Corsair Dominator 2 x CMD4GX3M2B1600C8 (8 Gig)
GPU: Sapphire Radeon ATI 4850x2 (1 Gig)
Monitors: 1 x LG L1715S & 2 x LG L194WT
Sound Card: M-Audio Delta 1010
HDD's: 2 X WD 640 Gig Caviar Black SATA3, 2 x Hitachi Deskstar 1TB, 1 x Seagate 1TB, 1 x Hitachi 500 Gig
DVD/CD: Liteon
Graphics Tablet: Wacom PTZ3
Keyboard: Microsoft Digital Media Pro
OS: Windows 7 Ultimate 64 bit (Retail)

[onemilimeter]

Did Tomshardware or other popular reviewers report the problem? Or, they're paid to keep quiet too???

[absic]

The trouble with the popular reviewers is that they don't live in the real world. They get their kit from the manufacturer's and of course they don't want to be too critical in case the supplies dry up.

If I were in a lab, with plenty of resources I could probably figure out a solution but I'm not fortunate enough to be in that position. In all other respects, my mobo is solid and I don't know if this warrants an RMA or, if I am really stupid and just can't set it up properly.

If there is a problem with the Marvell 9128 Chip, Gigabyte should be honest and tell us, offer to replace our boards and put it down to experience. (Take a look at Toyota for what to do if you get it wrong!)

If they believe it is a relatively isolated problem with the odd Marvell 9128 Chip, here or there, tell us that, so we can RMA our boards and get what we paid for rather than leaving us in limbo.

[onemilimeter]

Has anyone have the emails of Gigabyte's top guns?

I've never tried the MARVELL SATA3 and RAID-0 (but I'm sure I will be using them in future) but if this problem is real (which I believe it's real because there are few similar reports) then it's a very serious problem. If it's not called "cheating", what else we can call it... Look at this advertisement:



Many people bought the board because of the "333". It's advertised by Gigabyte, so Gigabyte should take the responsibility although the main culprit is the Marvell 9128 chip (Marvell Chip 88SE9128). I remember someone said before that the speed of RAID-0 using the Marvell SATA3 controller is in fact slower than that using SATA2 (ICH10 controller). May be Gigabyte should correct the advertisement to be as follow:

[F5BJR]

*

Marvell Chip 88SE9128 is very good with my HighPoint RR622 card but limited to 500 mo/s because use PCI-e X1 !!

Marvell Chip 88SE9128  is very good with my HighPoint RR640 card but limited to 2Go/s because use PCI-e X4 !!

With HighPoint cards i have absolutely no problems  

*

Problem is not with GigaBYte hardware but with the firmware build by Marvell

I have 2 * IOISATA add-on cards with the same chip / firmware and exactly the same limitations / problems

http://www.ioisata.com/products/Host-Adapter/S3-PCIE1XG201.htm

Pierre

[Dark Mantis]

I would have to agree with your findings absic as when I first set up my board I too bought a pair of Western Digital Black SATA3 hard drives to use  on the Marvell SATA3 ports I got the RAID0 to work but as I had a SSD I was booting from that. I didn't try and load my OS onto the RAID device. It seemed to work ok for data only but to be fair I didn't use the drive that much.

I am just formatting them again to RAID0 and will see how I get on. At the moment I have them on the Intel ICH10 controller which is only SATA2 but I will try swapping them over to the Marvell SATA3 tomorrow and install the OS on them. I will report back when I have news.

Problem is not with GigaBYte hardware but with the firmware build by Marvell

Pierre if the problem is with the firmware of the Marvell chip why can't they just update it? They have had long enough and the problem has been known about.

[F5BJR]

*
May be a story of big money or license  ?

Maybe HighPoint bought another type of license and development kits to make their own firmware
that are neither GIGABYTE / ASUS / MSI / IOISATA / ASROCK ...

*

I have read the short form documentation for Marvell chip 88SE91xx series :

** RAID 5/10 is also available as an option when the 88SE91xx is connected with a SATA port multiplier **

I can mount RAID 5 / 10 with the RR622 / RR640 HighPoint cards with my 2 * ports multiplier enclosures ( 4 * disks in each)  
I can also use with 2 * ports multipliers ( RR622 ) or 4 with RR640 and not only one with Marvell firmware
Why not with Marvell firmware ?
So it is not my boxes that are not compatible

http://www.marvell.com/products/storage/storage_system_solutions/sata_controllers_pc_consumer/6_gbs_sata_raid_controller_88se91xx_product_brief.pdf

Question is why , who is the liar ?

Pierre

*********************************************************************************************************************************

* I have also an old problem ( > 1 year ) with another Marvell chip but 88SE6145
( Asus Motherboard P5E-WS-PRO and also with this chipset in an add-on card IOISATA  )

* time out with RAID mounted disks

not solved at this time !!

I have contacted ASUS , IOISATA , i have finnaly receive in April a new firmware and always the same time out problem
Problem exist also with this chip / firmware  and INTEL motherboard D975XBX2

[Dark Mantis]

Data Throughput Tests

To start with I must explain my testing procedure. I first tried just a data throughput of the different controllers with the drives in my computer. I would like to make clear that this is as near real world tests as is possible. The usual programs were running in the background and I haven’t made any changes or run any cleaning programs etc prior to the tests.  I have even added some data to the RAID0 drive array as it was newly formatted.

The testing was carried out with the help of Crystal Diskmark 3.0

The first series of tests were just as I normally run my computer, booting from the Corsair Performance SSD drive attached to the Marvell SATA3 port and with a SATA2 data drive for storage and a pair of the Western Digital Black SATA3 cache 640 GB drives that are configured as RAID0 both connected to the SATA2 Intel ports. I am using Windows 7 64bit operating system with 6 GB of RAM.
For the second series of tests I simply swapped the SSD drive to the Intel controller and moved the RAID0 drives to the Marvell controller. Everything else stayed exactly the same.

I want to emphasize that this is just testing throughput, nothing else.

 

  

Booting tests

My next series of tests were to try and install Windows 7 64bit onto the RAID0 array as a boot drive connected to the Marvell controller ports.

I have to admit that I failed miserably! I tried every way that I know of to install the RAID but nothing seemed to work properly. Every time that I tried to run setup for the OS it would either lock up or restart. The RAID array kept disappearing from the install dialogue box and only a reboot would get it back.  These are exactly the same problems that absic ran into when he tried the same thing on his AMD setup.

In conclusion I would have to say that the Marvell chip works quite well with a RAID0 array(data striping) as long as it is only for data transference. If you want to boot from it best of luck! It certainly doesn’t live up to the hype!

In view of these tests I would like to ask Gigabyte to comment on what action they intend to take to rectify this problem

[absic]

In conclusion I would have to say that the Marvell chip works quite well with a RAID0 array(data striping) as long as it is only for data transference

Maybe for you DM but I had the problem of the RAID array being dropped/lost so I wouldn't trust it to handle or store important data and the User's Manual clearly shows how to install the OS onto a RAID Array using the Marvell Chip, so I believe it should be able to do that. It should also be able to deliver the speeds quoted in the advertising but it is actually slower than my SATA2

I have run my own speed tests using Crystal Disk and I had some different scores to yours. Rather than fill the thread with pictures, if you are interested you can find them here:
http://img84.imageshack.us/i/speed001.jpg/ This image shows the speeds from my C: drive which has the OS installed on to it in IDE mode
http://img830.imageshack.us/i/speed002.jpg/ This image shows the speeds of my D: drive which, I had left empty after my attempts to install the RAID Array.
http://img197.imageshack.us/i/speed003.jpg/ This shows the speed from one of my SATA2 Drives attached to the AMD SB750 Chipset. As you can see this is actually faster than the SATA3

The following three images just give general information regarding the hard-drives tested.
http://img807.imageshack.us/i/info001.jpg/
http://img84.imageshack.us/i/info002.jpg/
http://img843.imageshack.us/i/info003.jpg/

I think a response form Gigabyte would now be appropriate as both Dark Mantis and I have issues with the Marvell 9128 chip on different platforms and we have tried to conduct similar tests which appear to have produced the same results.

[Dark Mantis]

Maybe for you DM but I had the problem of the RAID array being dropped/lost so I wouldn't trust it to handle or store important data and the User's Manual clearly shows how to install the OS onto a RAID Array using the Marvell Chip, so I believe it should be able to do that.

You wouldn't get any argument from me on that score absic. I never mentioned anything about it being reliable.
I tried  following the user manual as well after failing several times to be able to install the OS but without any more success.

http://img197.imageshack.us/i/speed003.jpg/ This shows the speed from one of my SATA2 Drives attached to the AMD SB750 Chipset. As you can see this is actually faster than the SATA3

That shows a substantial deficit on behalf of the SATA3.

[onemilimeter]

Has anyone relayed the problem to GGTS? What's GGTS response?

[Dark Mantis]

I believe that absic has PMed runn3R about it. We shall have to see what, if anything, they decide to do about it. My feeling is they will just try and ignore it and hope it goes away the same as the noise on the motherboards. I think if that is the case they will find that is the last thing that will happen.