Author Topic: any other chipsets other than sil3112 able to boot OS 9? is sil3124 possible?  (Read 16851 times)

Offline DieHard

  • Global Moderator
  • Platinum Member
  • *****
  • Posts: 2366
Quote
it makes most sense to have some type of download archive that collects all smaller utilities like this; all versions; to be posted here on downloads board; like macos9lives system utils collection etc; but the admins are too busy :D

It may seem that way, but quick forum searches are usually are better, if we start making separated downloads for stuff included in the OS; we are going to get even more convoluted.

macStuff

  • Guest
drive setup is a pretty important thing; especially when alot of people are trying to properly format their SSD's
pretty mcuh anyone using os9 in 2019 is probably installing an SSD

anyway good thing i have my own site to post things on; right? :)

Offline refinery

  • Gold Member
  • *****
  • Posts: 383
well i think the safest bet is really just tell people if they're gonna use SSDs with OS9 that they need to format them with a minimum of OSX 10.4... That has been my method and its worked for me 100%... All my desktop OS9 machines are using dual SSDs now. I run retrospect and they start up once a week and duplicate themselves to the second drive since im paranoid about them going bad since there's no TRIM enable for OS9.
got my mind on my scsi and my scsi on my mind

Offline DieHard

  • Global Moderator
  • Platinum Member
  • *****
  • Posts: 2366
well i think the safest bet is really just tell people if they're gonna use SSDs with OS9 that they need to format them with a minimum of OSX 10.4... That has been my method and its worked for me 100%... All my desktop OS9 machines are using dual SSDs now. I run retrospect and they start up once a week and duplicate themselves to the second drive since im paranoid about them going bad since there's no TRIM enable for OS9.

Six years since my 1st OWC SSD and not one issue :)

Offline IIO

  • Platinum Member
  • *****
  • Posts: 4439
  • just a number
starting with SSDs here this week.
ordered a bunch of evo 840, one will go in the audio G4.
which means that i have to connect one of the HDs to the 33 bus for now ... humh.
insert arbitrary signature here

Offline refinery

  • Gold Member
  • *****
  • Posts: 383
apparently the adaptec 1205sa uses a 3112 chipset?
i found this forum post where a guy talks about modifying Silicon Image drivers to include the adaptec PCI device IDs and it works:
https://www.techspot.com/community/topics/adaptecs-1205sa-users-heres-what-i-sent-to-adaptec.27250/

these cards pop up on ebay a lot, there's one right now as low as $9.
https://www.ebay.com/itm/Adaptec-ASH-1205SA-FAB-Host-Controller-SATA-Controller-Card-Fully-Tested/153516582938?epid=1304118779&hash=item23be4d281a:g:nx8AAOSwN5Fc9XPp
got my mind on my scsi and my scsi on my mind

Offline SDG

  • Enthusiast Member
  • ***
  • Posts: 58
  • New Member
apparently the adaptec 1205sa uses a 3112 chipset?

these cards pop up on ebay a lot, there's one right now as low as $9.
https://www.ebay.com/itm/Adaptec-ASH-1205SA-FAB-Host-Controller-SATA-Controller-Card-Fully-Tested/153516582938?epid=1304118779&hash=item23be4d281a:g:nx8AAOSwN5Fc9XPp

Looking at that card, the main problem I can see is that there is a differently shaped bios chip. I'm assuming it's the long narrow one alongside the processor chip. I can't see how it is possible to get one of the three required eeprom chips on it to flash it with the SeriTek firmware that can boot OS9.  That is assuming the design of the card is electrically compatible with the SeriTek firmware as it looks somewhat different from most I have seen.

Offline (S)ATAman

  • Veteran Member
  • ****
  • Posts: 161
  • New Member
All FT cards and all Sonnet cards sourced from either ACard or FT do boot MacOS 9.x.
Under two conditions:

1) The card has to have PCI or PCI-X interface. Obviously, the cards with PCIe do not boot MacOS 9.x.
If someone can hack the 2005 G5 to boot MacOS 9.x that would not help either because these PCIe cards do not have Open Firmware bootstrap code.
2) The drive in question has to have a generic MacOS 9.x SCSI mass storage driver partition. Without that partition the drive will be seen by Disk Utility only, but won't show up. That partition can be installed by the usual Disk Utility and it's siblings (FWB, ATTO, LaCie, etc). If not, there are some system extensions which are a generic SCSI mass storage driver in disguise - these will usually work, too. YMMV.

The only notable exception from the rule above are the PCI-X cards from Sonnet based on Marvell 88SX.... chipset.
These aren't bootable on either macOS X or MacOS 9.x

And some interesting addition: the originator of probably fastest first generation PCI-X SATA was Vitesse, not Intel. Even more - there was as I remember an other company who engineered these chips, than Vitesse acquired it, than Intel did partner with Vitesse. They made SATA based on some existing technology which had nothing to do with SATA and delayed. At the end Sonnet was p***ed off and turned to Marvell. But they wanted the Vitesse originally. That was sad.

Ultimately the Marvell turned to be the best choice before AHCI swallowed everything.

AHCI is really a good standard - what made G5 and all pre-Intel Macs difficult to expand was the fact that the best that-time AHCI controller (Marvell 88SE923x) had only two PCIe lanes and the architecture of 2005 G5 made it impossible to use it there.

I am guilty that I did not try that Marvell controller in an earlier G5 or G4. But since it is a PCIe controller, I would need a "bridge". The cost of proper 64-bit wide bridging would be prohibitive.

This is why there is no AHCI controller on G4.

For G5 there is a single choice, based on Marvell 9128. Unfortunately the performance of 9128 is very disappointing.
For good or bad, I did not try others like ASMedia because only Marvell AHCI controller had FIS-based switching for Port Multipliers. ASMedia stuck to command-based switching. But... in the fact, the Marvell 9128 is barely better, I can't address within the same PMP more, than two drives the same time. Which makes the FIS-based switching on 9128 rather a theoretical benefit.

I wish there would be a 4-channel PCIe SATA AHCI chip, but there isn't. The on-board SATA controller of 2005 G5-s is very-very dated.

Offline (S)ATAman

  • Veteran Member
  • ****
  • Posts: 161
  • New Member
Update: now I became unsure about the SImage 3124. I am 100% sure it is bootable, but don't remember (sorry, seniority moment) does it feature the proper SIM driver for MacOS 9.x or not. Hold on.

Update/2: Oooops, never did it. So don't look for Mac OS 9.x SIM inside of SImage 3124 ROM because it's not there.
Seniority moment. Only Open Firmware + PPC macOS-X driver is there, next to the BIOS. Apologies. But at least it's bootable, even from port multiplier (on PowerPC machines). 100% sure never did EFI because I am not that smart (yet). Maybe later... who knows. Do not held your breath back.

With SIP disabled the x64 binaries should work even in Catalina, provided it's a 3124 with a PCIe bridge. I did see some - but there is no single reason to use 3124 on PCIe machines if there is Marvell 88SX7042 (or even Marvell 88SE923x) and these cards aren't expensive.

Answering the original question:

1) 7174 does have SIM for sure.

2) Obviously, I do have 3114 SIM. It was never released because the performance was not really what I did dream about.

3) Unfortunately there is no 3124 SIM and never was made. I wasn't remembering correctly, it was long ago.

4) 3124 is a huge pain in the neck to program and I am not sure, the number of the second-hand cards and the demand would justify the efforts. The chip has also a fatal flaw. That flaw is that under certain circumstances (I don't remember exactly) it will confuse drives within the port multiplier. It will think that the drive in, say, bay number two is the one in the bay number five or v.v.
I do not recall, SImage would ever fix it, they tried to cover that bug with a very thick paint.

Under such circumstances 7174 is probably the best performing SATA card for MacOS 9.x - but unfortunately it's supply is very limited to say the least.
But hey, if the demand is there, 3124 SIM can be made. But it is not an easy task.
« Last Edit: December 08, 2019, 05:17:54 PM by (S)ATAman »

Offline DieHard

  • Global Moderator
  • Platinum Member
  • *****
  • Posts: 2366
Thank you (S)ATAman,

that was a great read !

Those "history" lessons are always insightful :)

Offline (S)ATAman

  • Veteran Member
  • ****
  • Posts: 161
  • New Member
By the way: just noticed the "Steelwine" being mentioned. It was made by Silicon Image and it has their port multiplier inside, intent to be sold by LaCie.
It had some limited RAID capabilities - but if you look inside, there is no backplane. Not many were sold for various reasons.

Besides the lack of the backplane the power supply wasn't very good. I still have two of these, the box was thrown away and I tried to recycle the power supply to test some SATA drives with a 9300-class SAS controller. No good, the SATA-III drives dropped out and protested. Changed the power supply: everything was fine.

Stay away.


Now please take a look at this:

https://www.ebay.com/itm/Promise-VTrak-J610f-16x-SATA-SATA-JBOD-Enclosure-mit-Fibre-channel/254448833373

Than buy this:

https://www.ebay.com/itm/LSI-Logic-LSI7204EP-2-port-4Gbps-FC-PCI-E-Controller-Card-Apple-Card-XSAN/303194611515

For G4 and such, maybe this:

https://www.ebay.com/itm/LSI-LSI7202XP-Apple-Dual-Channel-Fibre-PCI-X-DP-Mac-Xserve-W-Fiber-Cables/332631100211


I have really trouble to justify Silicon Image 3124 with grave bug inside of Port Multiplier firmware if there is something like above - in abundance.