GA-X58A*-UD3R/UD5/UD7... MARVELL NEEDS to be TOLD !!!

[scjet]

 Hi everyone:

 Lets do our part in letting Gigabyte know that we're still waiting for them to get on Marvell's ___, and fix their SATA3 controller/driver issues for these great MB's...

Give 'em heck here: http://www.gigabyte.com.tw/Support/ServiceCenter.aspx


 I already sent them mine a few times:
_____________________________________________________________________________________-
From :  Rick [xxxxxxx@xxxl.com ]
Sent :  3/24/2010 03:57
Question :  The (Marvell)SATA3 ports in ANY Raid configuration do not work well at all ?!
Intels' onboard ICH10R (sata2) beats Marvell's onboard 9128 SATA3 controller. And why is this !? This is very bad on Gigabytes part.

Please read here:
http://www.hardocp.com/article/2010/02/20/sata_6gbs_on_your_new_motherboard/2
and here:
http://forum.giga-byte.co.uk/index.php/topic,1587.0.html
-----------------------------------------

Can you(Gigabyte) please FIX this "Marvell/Gigabyte driver issue" ASAP !

thankyou,

Rick.

-------------------------------------------------------------------------------
Model Name : GA-X58A-UD3R(rev. 1.0)
--------------------------
M/B Rev : 1.0
BIOS Ver : F5
Serial No. :
Purchase Dealer :
-------------------------------------------------------------------------------
VGA Brand : Nvidia      Model : FX1700
CPU Brand : Intel      Model :       Speed :
Operation System : Win 7 64-bit      SP :
Memory Brand :       Type :
Memory Size :       Speed :
Power Supply : W
_______________________________________________________________________

[Gigalee]

I agree...in fact, I did it last night!

I also reported the issue straight to Marvell (see link below).

http://www.marvell.com/company/contact.html

[scjet]

bump !

[scjet]

-----------------------------------------------
Well, I finally got a reply from Gigabyte regarding SATA3 speed/performance:
Now, if any of you Gigayte guru's out there can explain this to us ALL, it would be greatly appreciated.

Here it is:
"ANSWER: Intel ICH10r controller is built into the actual Southbridge chip where as Jmicron is a separate chip on the board which is connected via PCI-Express lane to the Southbridge, this will have effect in overall performance.
Both JMicron and ICH10r are SATA II 3GB specification but it does not mean that they will have the same performance, for example if you compare Corsair and Mushskin both memory vendor provide product which are2200MHz spec, performance will be different on it as well"
--------------------------------------
 
So, ummm what has the Jmicron controller got to do with Marvell9128 Sata3 controller.? I understand that they are "shared" via pci-e bus, BUT, I am NOT using any of the Jmicron, USB3,  or GSATA2 ports at all. ? Just the ich10R and (Marvell)Sata3 ports.
 Does their answer actually mean that this so-called "BANDWIDTH sharing/robbing" is STATICALLY-linked? (meaning that so much bandwidth is taken period !?), as opposed to "dynamic", where it would be taken only if used or needed ?

 anyway, sry for askin' 'bout this , but they were polite enough to answer above, so I'm just letting everyone know.

thx again.
Rick.

[skiigor]

-----------------------------------------------
Well, I finally got a reply from Gigabyte regarding SATA3 speed/performance:
Now, if any of you Gigayte guru's out there can explain this to us ALL, it would be greatly appreciated.

Here it is:
"ANSWER: Intel ICH10r controller is built into the actual Southbridge chip where as Jmicron is a separate chip on the board which is connected via PCI-Express lane to the Southbridge, this will have effect in overall performance.
Both JMicron and ICH10r are SATA II 3GB specification but it does not mean that they will have the same performance, for example if you compare Corsair and Mushskin both memory vendor provide product which are2200MHz spec, performance will be different on it as well"
--------------------------------------
 
So, ummm what has the Jmicron controller got to do with Marvell9128 Sata3 controller.? I understand that they are "shared" via pci-e bus, BUT, I am NOT using any of the Jmicron, USB3,  or GSATA2 ports at all. ? Just the ich10R and (Marvell)Sata3 ports.
 Does their answer actually mean that this so-called "BANDWIDTH sharing/robbing" is STATICALLY-linked? (meaning that so much bandwidth is taken period !?), as opposed to "dynamic", where it would be taken only if used or needed ?

 anyway, sry for askin' 'bout this , but they were polite enough to answer above, so I'm just letting everyone know.

thx again.
Rick.

This reply - a complete disappointment
I think man from support does not even understand the question.

I think you have to write another letter to support and explain (explain for support  ) that the Marvell9128 Sata3 controller is not connected to the Southbridge PCI-Express 1.0 lane like Jmicron. Marvell9128 Sata3 controller is  connected  to  Northbridge  PCI-Express 2.0 lane.
This is controller of the new SATA3 generation and it definitely have to surpass the speed of all controllers of the previous SATA2 generation

Ask them to make their own speed test of RAID0 disk array connected to the Marvell9128

[skiigor]

I found a few original GIGABYTE slides in the GA-X58A-UD7 presentation

http://forum.giga-byte.co.uk/index.php/topic,818.0.html

[skiigor]

But no RAID-0 tests  and no SSD tests (HDD - it is bottleneck and old technology today)

[scjet]

My Questions:

Question 1./ The (Marvell)SATA3 ports in ANY Raid configuration do not work well at all ?!
I was only able to achieve approximately Write->200 MB/s, and Read->100 MB/s !!!!!
Intels' onboard ICH10R (sata2) beats Marvell's onboard 9128 SATA3 controller. And why is this !? This is very bad on Gigabytes part.
Question 2:/ The Marvell 9128 is the controller of the new SATA3 generation and it definitely have to surpass the speed of all controllers of the previous SATA2 generation.! Why has Marvell not updated their drivers in this regard ?.
Question 3:/ Did GIGABYtE tech support "speed test" a RAID0 disk array configured(via BIOS configurator)
and connected to the Marvell 9128 ports ? and if so, please post me your specific speed test results for Sata3 "Raid 0"?
Question 4:/ Did Gigabyte also "test" these Marvel 9128 Sata3 controller ports using a "Raid 0" with SSD (Solid-State-Drives) ?
------------------------------------------

Giga's Answer :  Dear Customer

Question 1 Our team is currently investigating the matter and are checking on this matter internally and with Marvell

Question 2 Our team is currently in contact with Marvell regarding this matter

Question 3 Find in attachment

Question 4 Find in attachment

please set it as AHCI mode which will have better performance then IDE mode

http://forum.giga-byte.co.uk/index.php/topic,1587.0.html

[scjet]

 Well, I tried attaching their (2) under 100k jpegs, but all I got was "The upload folder is full. Please try a smaller file and/or contact an administrator." whatever.

 Basically, the pics would've just shown that  ICH10R(sata2 controller) performed generally better than Marvell's 9128 (sata3) contoller, even with SSD's.

 Howvever, in ahci-mode and using sofware Raid only, performance of the HD's connected to the SATA3 contoller right now, is atleast as good as ICH10R. -generally speaking right now.

 ok well, until Marvell "junk" semiconductors fix their controller/driver issues ? this horse is beaten enough, I'd say.

[skiigor]

Please, upload this attachment jpegs to the http://imageshack.us/
This is very interesting ...

[littletiger1213]

I just tested my RAID 0 and compare with other people's result online. It seems not far from mean although it is not necessarily in the top tier group.



RAID 0 2 x WD1500HLFS WD VelociRaptor150 GB, SATA 3 Gb/s, 16 MB Cache, 10,000 RPM

Controller: Marvell on Gigabyte GA-X58A-UD7

-----------------------------------------------------------------------
CrystalDiskMark 3.0 x64 (C) 2007-2010 hiyohiyo
                           Crystal Dew World : http://crystalmark.info/
-----------------------------------------------------------------------
* MB/s = 1,000,000 byte/s [SATA/300 = 300,000,000 byte/s]

           Sequential Read :   217.007 MB/s
          Sequential Write :   229.515 MB/s
         Random Read 512KB :    60.115 MB/s
        Random Write 512KB :    99.928 MB/s
    Random Read 4KB (QD=1) :     0.842 MB/s [   205.5 IOPS]
   Random Write 4KB (QD=1) :     2.946 MB/s [   719.2 IOPS]
   Random Read 4KB (QD=32) :     3.897 MB/s [   951.4 IOPS]
  Random Write 4KB (QD=32) :     4.305 MB/s [  1051.0 IOPS]

  Test : 500 MB [C: 9.9% (27.5/278.9 GB)] (x5)
  Date : 2010/04/04 0:36:35
    OS : Windows 7  [6.1 Build 7600] (x64)

[scjet]

Please, upload this attachment jpegs to the http://imageshack.us/
This is very interesting ...

-------------------------------------

Hey all,
 The below (2) jpeg attachments (as per Gigabytes' testing of their Marvell 9128 ports) that were uploaded to the above url, are the ones from Gigabytes sata3 reply to me, as per this thread.
 What I found atleast "honest" on Gigabytes part, was that they compared their results alongside the ICH10R sata2 ports as well....

sata3_HDD test:
http://img694.imageshack.us/img694/9986/sata3hdd.jpg

sata3_SSD test:
http://img215.imageshack.us/img215/2843/sata3ssd.jpg

-let me know if it don't work ?
thx,
Rick.

[Gigalee]

Well after much testing and system rebuilding I have to report that my X58A-UD7 is going back to the supplier for replacement.

It seems that the Marvell SATA III controller on the board is actually faulty......hence why I can't get the system to boot / run stably when the Operating System Drives are attached to it!

I've now got to strip-down my computer to remove the faulty Mobo......what a pain!

[scjet]

Well after much testing and system rebuilding I have to report that my X58A-UD7 is going back to the supplier for replacement.

It seems that the Marvell SATA III controller on the board is actually faulty......hence why I can't get the system to boot / run stably when the Operating System Drives are attached to it!

I've now got to strip-down my computer to remove the faulty Mobo......what a pain!

...sry to hear that Gigalee.
 the Marvell Sata3's have obvious driver issues, I guess it doesn't surprise me that some brd's have sata3 hardware issues as well.
 but good-luck in future fer sure, and let us know how it goes?

Rick.
 

[Gigalee]

Well, I finally got around to building my new machine (again) using the replacement motherboard...

...and it would appear that the Marvell SATA 6Gb/s Controller doesn't work properly on this one either!