Who is interested in a fix for "Marvell" 6G SATA

[F5BJR]

*
I have requiert an account ( NDA ) a long time ... > 1 year

At this day no response !!

Pierre

[mpetroul]

I am not sure I understand the question. What is wrong with it that we should pay to fix?

[Aussie Allan]

 What Planet have you been living on ?........only joking!    the original Marvell Controller chip 88SE9128 was shipped across multiple platforms and brands, ... it was marketed as the new 6GB/S connectivity called SATAIII.........Unfortunately, unless you were a Marvell Engineer, running Marvell equipment and sitting on a Marvell chair, no one could get them to work at anything near the speeds that were advertised, ....Luckily it was backwards compatible so SATAII to sub SATAII  speeds were the best people could get out of them.

  Z68 boards are now shipping with the new 88SE9182 and initial report are ......"No more Problems"

 Aussie Allan

Update! forgot to answer the second part!..... there is a movement looking into the feasability of swapping out the problem chip with something that will operate at the full SATAIII spec........A lot of people purchased SATAIII drives and such that really haven't used them yet at anything faster then SATAII..........Hope this helps.

[Dark Mantis]

It is surprising the number of owners who still don't realise the problem they have with their boards. I suppose if they don't have any SATA3 drives then they would be none the wiser.

[Aussie Allan]

 Depressing really, if they only new what's available to them, .... so many people read the outside of the box, plug it in and assume it's running as advertised........ how can we get more of the posters to hit the  vote key?...........would be good if the poster was blocked  until casting a vote if there's an active voting window attached to the subject post.......mmmmmmmmmmmmmmm!

 Aussie Allan

[Dark Mantis]

Not possible as far as I know Allan. I know what you are saying though as a lot of people do seem to have an aversion to clicking in the polls. I really don't know why as it couldn't be simpler but there we go.

[Aussie Allan]

 As we say in Australia......."BUGGER"

  Aussie Allan

[mpetroul]

Youre actually right, I just got this board, a Z68-UD7 and I dont have any SATA III drives yet. I have been updating from my X58 system a piece at a time and I thought the biggest bang for my buck was my video cards first then the board processor and memory. I thought the marvel issue that I had been reading about had to do with the Sandy bridge bug and that the board I bought; all the B3's were immune to it.

 

[Dark Mantis]

Hi

If you have a Z68 board then you are not affected by this Marvell problem as you already have the updated 9182 chip installed.

The Marvell problem has nothing to do with the Sandy Bridge issue. It was all to do with the Marvell 9128 that was fitted to the X58 and P55 series boards.

[Wonderwrench]

Besides the fact that the Marvell 9128 controller is a dog and does not have the bandwidth to support true SATA III speeds it is also limited by the 1x PCI-E 2.0 connection. On paper a single PCI-E 2.0 lane can handle 500 MB/s. After overhead low to mid 400 MB/s I'd guess. So even if it were possible to install a 9182 chip on X58 boards you would be limited by the PCI-E 2.0 1x connection.  Marvell 9128 spec sheet. http://www.marvell.com/products/storage/storage_system_solutions/sata_controllers_pc_consumer/6_gbs_sata_raid_controller_88se91xx_product_brief.pdf

I think the Marvell 9182 controller uses 2 PCI-E 2.0 lanes so 1000 MB/s less overhead. Can't find the specs at Marvell. From the bench testing I have seen on the net is limited to around 600-650 MB/s sequential R/W's . Tested using two Vertex 3 120's in RAID 0. Tests using Vertex 3 drives on the SandyBridge  native SATA  III controller or an LSI card RAID card get results very close to 1000 megs per second sequential R/W's. So from what I see the new Marvell 9182 controller is much better than the 9128 but it is still a bottle neck. May work very well with a single drive though I have not seen any bench results yet.

Bill

[Aussie Allan]

 Hay there Bill.......Nice of you to jump in to one of my pet projects.............."Can't find the specs at Marvell"......What an understatement.

  There vetting procedure seems harsh,there web site bullet proof, but I'll keep trying!  All the imformation Ive been told is there, it's just that it's behind locked doors unless your corporate account or a high end developer.......agreed the 9182 would be better and a step forward.
   you seem to have a grasp of what's required , so do you have a better idea/ recommendation from the Marvell line-up besides the 9130...........And there's a question that's gone unanswered till now.

                                                                            NEW BIG QUESTION PEOPLE!

  If the Marvell chip was swapped out and replaced with say a 88SE9182 chip (same die size,some 76pin count) would it fire up or require BIOS code to be modified/changed ?.........Answer this and Get the booby prize!

  And Bill ,... yes I understand it's still a bottle neck,,,,,But a hell of a lot bigger

 Aussie Allan

[Wonderwrench]

 Hay there Bill.......Nice of you to jump in to one of my pet projects.............."Can't find the specs at Marvell"......What an understatement.

  There vetting procedure seems harsh,there web site bullet proof, but I'll keep trying!  All the imformation Ive been told is there, it's just that it's behind locked doors unless your corporate account or a high end developer.......agreed the 9182 would be better and a step forward.
   you seem to have a grasp of what's required , so do you have a better idea/ recommendation from the Marvell line-up besides the 9130...........And there's a question that's gone unanswered till now.

                                                                            NEW BIG QUESTION PEOPLE!

  If the Marvell chip was swapped out and replaced with say a 88SE9182 chip (same die size,some 76pin count) would it fire up or require BIOS code to be modified/changed ?.........Answer this and Get the booby prize!

  And Bill ,... yes I understand it's still a bottle neck,,,,,But a hell of a lot bigger

 Aussie Allan

I really can't say if it would work or not. I, we just do not have enough info available to even make a reasonable guess IMO.

Bill

[Aussie Allan]

 Well to get this going forward, might be time just jump in and pop a new chip on a board and see what happens.....R&D or Trial by fire?.....bit of both!

  Anybody have some X58 boards we can cut-up going cheap......?

 Aussie Allan

[Dark Mantis]

Hi Allan

I would think the least Gigabyte could do was to supply a motherboard to test on. Try contacting runn3R and see if he can help out.

[Aussie Allan]

 I was thinking this not 2 minutes ago.............saves me an email to you!...........But I still have not had a response from the last one to runn you looked at as well?

 Any suggestions?......I'm cold calling Marvell today down in Buckinghamshire to see if I can pull some teeth (chips) ...wish me luck!....AA