Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

3112 SATA data drives for the Classic Mac OS?

Are you interested?

  • yes

    Votes: 19 100.0%

  • no

    Votes: 0 0.0%
  • Total voters
    19
DearthnVader

DearthnVader

Suspended
Original poster
Dec 17, 2015
2,207
6,393
Red Springs, NC
Who is interested in a Classic Mac OS driver for the Sil3112 Serial ATA card?

The point of the driver is to make data drives available to the Classic Mac OS on 3112 cards that do not have a large enough EEPROM to support the 'NDRV' in ROM.

The disks would not be bootable and be subject to the partition size and format limitations of the Classic Mac OS you chose to boot.
 
  • Like
Reactions: Ataman Honcho Honchev, mustagcoupe, G4fanboy and 1 other person
Just for the fun of it I tested it in my 9600 with Mac OS 7.6.1 and the attached SATA drives did mount( Where Have ALL My Files Gone! because I had HFS+ and JHFS+ disks ).
 
Last edited:
  • Like
Reactions: LightBulbFun
The SIM (SCSI Interface Module) complying with written mac OS 9 rules is doable.
The SIM complying also with unwritten rules is somewhat less easy.

Just the driver (not a SIM interfacing with the SCSI Manager 4.3 and not designed to boot) is a different story.

BTW: I am at "Beta 11" right now. Like before aiming at SCSI Manager 4.3-compliance and bootability.
One is clear: as far as performance, the SImage 3112 / 3512 is quite mediocre and the 3114 is straight awful.
 
Today's tests... Anyone can explain how these three 3112-s were packed into a single PCI slot of the G4?
;-)
SImage3112.jpg
ASPPicture.jpg
 
  • Like
Reactions: ojfd and LightBulbFun
DearthnVader said:
Who is interested in a Classic Mac OS driver for the Sil3112 Serial ATA card?

The point of the driver is to make data drives available to the Classic Mac OS on 3112 cards that do not have a large enough EEPROM to support the 'NDRV' in ROM.

The disks would not be bootable and be subject to the partition size and format limitations of the Classic Mac OS you chose to boot.
Click to expand...
You can download it. Also for 3114 with all four channels on. I posted the "Beta 4" and "Beta 5" few weeks ago.
There aren't earth-shaking changes between the "Beta 11" and "Beta 4" or "Beta 5".

The only thing is that I tried memory-space access for 3112 in "Beta 11", the same way as it (SHOULD be ALWAYS) done in 3114 (but looks like Wiebetech didn't do that). The performance turned out to be the same, so it is not a big deal. The 3114 can be accessed through I/O space tho - but than you have to treat the four SATA channels like two PATA channels with "Master-Slave" device pair.

I didn't try that on 3114 yet, the performance with four independent channels accessed via BAR-5 MMIO register is already bad enough. I don't think that a worse scenario would yield better results.

Besides, emulating master-slave on a 4-channel PCI card pretty much breaks my device numbering everywhere, so probably would need to significantly modify the Open Firmware and "X" numbering as well. No mood for that.

And, of course, would completely break the upcoming 3124 logic where you can have up to 15 (practically: 5) devices attached to the same channel FOR REAL.
 
Last edited:
joevt said:
What's a 8086:5038 ? Might be more clear if you did ioreg or lspci or pcitree.sh in Mac OS X.
Click to expand...

im guessing @Ataman Honcho Honchev knows exactly how they where packaged in here :)


Its supposed to be a fun comment to see if anyone else can figure it out :)

Ataman Honcho Honchev said:
Today's tests... Anyone can explain how these three 3112-s were packed into a single PCI slot of the G4?
Click to expand...

to which end im going to say that im guessing ya managed to get one of those Funky PCI-X RAID cards working? the ones that have 3 chips behind a PCI bridge?

if so very awesome! :)
 
  • Like
Reactions: Ataman Honcho Honchev
LightBulbFun said:
to which end im going to say that im guessing ya managed to get one of those Funky PCI-X RAID cards working? the ones that have 3 chips behind a PCI bridge?

if so very awesome! :)
Click to expand...
Yes, the one made by LSI. I have few more (made by Adaptec), they have 3512, not 3112.
Both are supported now.

The main problem: normal SATA cables won't fit, I can't close the case of G4.
I had to get these "L" shaped connectors. Which are more difficult here to get than a premium champagne for $12.
Controllers1.jpg
Controllers2.jpg
Controllers3.jpg
 
  • Like
Reactions: LightBulbFun
joevt said:
What's a 8086:5038 ? Might be more clear if you did ioreg or lspci or pcitree.sh in Mac OS X.
Click to expand...
That's the I/O unit of the Intel 80302. The bridge is the Intel i960.
The 80302 packs the bridge and the I/O unit.

It is very easy to kick the 80302 into transparent mode - than you see all the three 3112 (or in the case of Adaptec: 3512) SATA chips. If you have a SIM driver (for "9") or KEXT (for "X") which does not go belly-up with multiple SATA chips in the same machine, than you should be fine.

In "Beta 11" I (hopefully!) solved the problem that if the SIM is loaded as an extension the drives attached to the cards do not show up on the desktop because the SCSI (mass storage) driver stored on the drive does not get loaded.

My SIM now loads all the SCSI drivers which are on attached drives but the drives are "unaccounted" for.

The other problem (unsolved) is that I wold like to pack the SIM, the KEXT and the OF in the ROM of these cards.
There is 16Megabit ROM chip on these cards, just unsure how I can address it.
 
Last edited:
  • Like
Reactions: Amethyst1
What is really funky tho is a SPATA (or PSATA?) controller. Never worked on it before, now the time arrived.

The SImage 3132 is actually a PSATA / SPATA controller. But the time was short, no manufacturer enabled the "PATA" part of PSATA, so that PATA port was (and is) dead, unoccupied and disabled in the firmware.

These funky VIA6421A controllers have nearly unlimited availability, three two SATA ports and one PATA port, always alive.

The inexpensive VIA6421A (true, it has small ROM size of 64K - but that's a lesser problem) arrived today and there is no problem with turning on the G4 Digital Audio or QS. If there is no problem with the sleep, than VIA6421A could be a better choice than the entire Silicon Image 3112 - 3512 - 3114 family.

But in order to know that I have to make quickly a SIM for VIA6421A and see how the "sleep" does work. So there will be some entertainment for me in coming days.

By the way: these "monster" RAID adapters with six SATA ports from either LSI or Adaptec (see the pictures above) have no problem with sleep. Same as the SeriTek cards. The Intel I960 bridge barely has any effect on the performance.

True, neither these cards (nor the SeriTek family) was made "cheaply"...

Was thinking to name that SIM to something like "Spaten", that beer is quite popular in the area (even that I am not a great fan of beer).

But just in case and to be politically / "alcoholically" ("alcohol-politically"?) correct (who knows who could be offended? And why make an advertisement to a booze to begin with?) - decided to name it "Viaduct3". Abstract bridges are on the money here (not famous presidents). Therefore that looks to be politically correct thing and names like "Viaduct3" should be fine.
 
Last edited:
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back