ga-x58a-ud7 and Marvell sata problem

[snofte]

Ga-x58a-ud7 (2.0 version) has problem with the Marvell sata controller:
ordinary WD hdd with ntfs attached to marvell sata port, checked with chkdsk /r.
Returns "unspecified error" in stage 3.
Switching same hdd to gigabyte sata port, and chkdsk has no problems at all ().
What is is with this marvell controller ?
xp sp3 with latest updates.
sata controllers set up as IDE in bios (version FC).

[Fatman]

Don't use the marvel ports use the intel. Those Marvel drivers are buggy!

[snofte]

ok, thanks.  But I really wanted the 6gb speed on new hdd, which the marvell should provide according to specs.
So there is absolutly no chance for a fresh driver ?  Maybe someone could give a link for download if possible ?
regards
erik

[alex80386]

ok, thanks.  But I really wanted the 6gb speed on new hdd, which the marvell should provide according to specs.

I don't think anyone ever gets that speed from a marvel sata controller do they?
Take a look at this for example:
http://techgage.com/article/battle_of_the_sata_30_controllers/1

Like Fatman said... Don't use the marvel ports use the intel.
If you really insist you could take a look at this thread...
http://www.evga.com/forums/tm.aspx?m=515774&mpage=1

[snofte]

Thanks all for sound advise.  One last puzzle:
I am now booting and running from ONE hdd connected to the marvell port.
Just for testing it out.  Seems to function, apart from the chkdsk error as mentioned before.
But should I not see the marvel under control panel/system/hardware ?
Here only Intel ich10 sata controller is listed under "ide-ata controllers".
And no drivers from marvell can be found via "sysinternals/autoruns/drivers.
Can anyone explain ?
regards
erik

 

[snofte]

UPDATE !! 
WOW !!
Solved ?
Well, did the following:
Gigabyte download drivers for marvell states version 1036.
It is this one that does not show the marvell in control panel/system/ide
On the Asus site, downloiad, I found that asus also has a driver for Marvell 9128,
this one has version 1045 !!!
Downloaded this one, and installed.
And bingo:  It shows us in control panel/system/ide.
AND: Chkdsk /r now runs fine, no abort on the same hdd
where chkdsk (with marvell driver 1036 installed) aborted in stage 3 !!!
One funny thing:  During chkdsk /f (with driver 1045) it
had to do some repair in stage 3 security description.
Also cleaning unused indexes.  But it finished OK !!
This repair never happend when i connected to the intel-port.
So is this 1045  driver from asus the answer ?
Any suggestions ? 

[snofte]

New update :
Reply from gigabyte support re the marvell driver:
"I must say it looks a bit strange with the version number of the driver on our site, I have to double check that."
So maybe new driver will be available in the gigabyte download site which could improve things ?
 

[Dark Mantis]

Hi

As has always been the case since the motherboards were initially released the Marvell 9128 controller has always been a letdown. Right from the word go it soon became obvious that it couldn't manage what it was built to do - the SATA3, 6GB/s throughput. It even struggled with a single drive but there was no way it could cope with a RAID0 array with all the extra load that put on it.

There has been a lot of updates and new drivers etc released but none really helped and this wasn't just Gigabyte either that had the problem but every manufacturer that included this chip as the SATA3 controller on their boards.

Even with the latest drivers it could still be brought to it's knees very quickly by asking it to perform as it was supposed to do originally.

[snofte]

Update:
Yes, I understand now the marvell problem.  But maybe it can be made to be stable in ide mode (not raid)
with latest drivers.  I tried the gigabyte 1.0.0.1036 and tested with chkdsk.  Aborted with unspecified error.
Tried the 1.0.0.1045 from asus site.  I did not abort, but did some peculiar "fixes" before finish ok.
Tried the 1.2.0.1027 from  station-drivers ((http://www.station-drivers.com/page/marvell.htm).
It worked ok via chkdsk /f.  Have a discussion with gigabyte support on this, and here is their reply so far:
" Thanks for the information! I am a bit curious about the driver version you said you downloaded. Right now I don't have the links I looked at earlier, but I will recall that on Gigabytes site it was version 1.0.0.1036. Was the version you downloaded 1.0.0.1045? Link to where you found the driver maybe? If you zip the file, you may also send it to me. The reason I ask is that the files I saw earlier was like 1.2.0.1027 (or similar) and I got a bit unsure if it would work on this board so I did not suggest it to you."

So I will see how this turns out.

[snofte]

update:
My marvell controlled hdd has so far proved ok with the 1027 driver.  Ran meny tests.
One puzzle:
uninstalling the marvell driver, rebooting, and the hdd still seem to function !
Marvel controller does not show in device manager, but the hdd attached to this controller seem to work ok !!
Defaulting to another driver, or what ?
can anyone explain

[Dark Mantis]

I am sure that it is a case of it simply reverting to the default Windows drivers instead. Have you tried benchmarking them both to see the difference ?

[snofte]

Tried benchmarking via ATTO on standard wd sata hdd (WD3200AAJS)
1.  hdd in marvell slot via marvell controller with 1027 marvell drivers: 105-107 mb
2.  hdd in marvell slot, marvell drivers uninstalled/disabled.
     Did not help simply to uninstall in device manager,
     marvell drivers kept popping up again at reboot.
     So I had to completly uninstall marvell drivers via revo uninstaller.
     After boot, still with hdd in marvell slot, new hardware was detected,
     and new controller popped up in device manager :  std dual pci ide controller.
     atto on this one gave almost the same. 103-112 mb
3.  Switched hdd to intel slot. after boot, atto gave almost the same, 100-112 mb

So on std sata hdd`s, result is almost the same.

 

[snofte]

Reply from gigabyte team scandinavia:_
"I found the version 1.2.0.1027 so I have not yet verified it with Gigabyte. I will ask technicians if they can verify any of these versions.
I have had no problems with Marvell. However, nowdays I don't use it unless I have to. The reason for me is that I run with high overclock and just don't want too many drivers. (i don't use sound devices either)
At the moment, I see no reason for you to change driver again.
Thanks again for the information. I will let you know Gigabytes official opinion. Send me a reminder if I don't write you back durng this week. "

[snofte]

Final update:
From gigabyte support:
"Some newer Gigabyte motherboards have the 1.2.0.1027, so I say stick to this one. Let me know if you run into issues with it."
So it seems marvell 1.2.0.1027 is the one to use.
Have also noticed that bios FD9 has some new settings for the marvell, maybe worth looking into.