Dual Marvell 9230 Cards, Driver Updates + Slow Boot Issue
Kudos to
Concorde Rules and all the other contributors - this is an excellent thread and much appreciated, if only in hindsight.
Before reading this, I installed two PCIeX 2X cards with a total of eight SATA 6G ports, six internal SATA and two external eSATA, into a Mac Pro 4.1 Dual Xeon 2.26 w/16GB PC1033, with mostly satisfactory results.
Here is my experience.
The cards I chose were the Marvell 88SE9230-based Syba SI-PEX400057 (4 internal SATA ports) and Syba SI-PEX400058 (2 internal SATA + 2 external eSATA ports), which sold w/rebates etc. for under $30 ea. incl. S&H (NewEgg). The early 2009 Mac Pro itself, and the drives and cards were all chosen to maximize performance/$ potential. This rig is relatively cheap.
The cards host seven SATA 6G drives: internally there is a Samsung 840EVO bootcamp drive (minimum 206MB HSF partition + Win 7 SP1 Ultimate on the NTFS balance); a Crucial M4 64GB for OSX 10.8.5; a second M4 HFS scratch drive; a pair of 2TB Seagate Barracuda 7200RPM HFS drives in a RAID0 array, and finally 2TB HFS and 3TB NTFS external drives via the two eSATA ports (limited to 3G by the enclosures' USB bridges, which I plan to jumper to restore 6G).
The MP's four internal bays are populated with 2TB HFS drives in an OSX software RAID0 array. I'm using MacDrive Pro 9 for Win 7 for read/write access to the HSF drives.
Slughead, to answer your question in post# 72, MacDrive reads/writes to HSF drives including the Marvell RAID arrays so all drives are accessible to both OS's. Backup of OSX is to an external 1TB Time Machine drive over USB2. Win 7 takes its chances at present.
So this machine has a lot of storage, some of it quite fast. Planned use is video editing using Premier Pro CS6 in Win 7 and FCPX in OSX.
Physically, since both optical bays are filled (a DVD and BluRay burner respectively) the five internal drives are shoehorned into the PCIe bay where they benefit from its ventilation (the Barracudas do get warm). Power is from a tap off the ODD power harness, at the 2x2 connector where it plugs in the backplane board.
No photographs, please. Jobs would rise from the dead if he saw what I did to his baby.
The cards were purchased in early December 2013 and came flashed with BIOS v.1.0.0.1010, Firmware v.2.3.0.1031 and Boot Loader v.1.0.1.0002. Windows 64bit drivers are v.1.2.0.1037, and a web-based interface "Marvell Storage Utility" app, v.4.4.0.2013, provides (via an Apache instance) an interface to the controllers for setup and maintenance of the RAID arrays (and Marvell's "Hyperduo" (fusion disk) HD-SSD combos, which I don't use).
Slughead, this sounds the same as Highpoint's WebGui software, which I haven't seen. It too can email me error msgs. Uh, thanks, but no thanks.
The boot camp partition was set up on a drive mounted temporarily in the 2nd ODD position as usual. Subsequent changes in choice of drive were accomplished using CCC and Winclone 3.4. Booting was sensitive to slot choice: by experiment it seems Slot 4 for the boot camp drive's card and either lower numbered slot (I chose 2) for the other card was required by the Apple UEFI. Slot 3 is populated with a 4-port USB2 card, powered from the same ODD power harness tap as the drives.
I'd like to use a RAID0 array of the M4 drives for the boot camp drive and put OSX on the Samsung 840, but I haven't tried removing all other lowered-numbered drives to see if Boot Camp Assistant works on a Marvell RAID0 array. It seems unlikely given the limitations imposed by Apple.
Trying different drives for boot camp and using the Winclone "shrink partition" function may have caused an issue with Win 7's BCD, because drive id collision issues were experienced subsequently. I probably booted into Win 7 with both drives, host and clone, attached simultaneously. Live and learn. These were addressed with Mark "SysInternals" Russinovich's authoritative post on the subject:
http://blogs.technet.com/b/markrussinovich/archive/2011/11/08/3463572.aspx
So: currently there is an unacceptably long boot delay into WIN 7: 30 sec. from chime to black-screen (OS selection); 145 sec. before the Marvell BIOS screens appear briefly; another 145 sec. of blinking cursor before a Win 7 green start bar, and finally another 20-30 sec. before Win 7 login appears.
Once it's started it's fast, but that's the longest boot time I've ever seen. Paper tape-driven PDP8's were faster (showing my age).
A reinstall of Win 7 on the boot camp partition is probably required. Before doing that (Win 7 updates and Adobe installs take forever) I will try to update the drivers, BIOS and firmware from higher-numbered driver and firmware versions at:
http://www.station-drivers.com/index.php/listes-constructeurs/70-marvell/47-marvell
I hope this information is of use and/or interest. I'd like to learn of others' experience with (1) multiple Marvell 9230 cards; (2) slow boot issues with Boot Camp Win 64-bit OS on PCIe SATA 6G cards; (3) setup of Win 7 boot camp on Marvell 9230-based SSD RAID0 arrays.
