Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
What is the difference between the firmware and the bios?
As far as i get it, the bios is the file which is somehow comparable to the rom of a video card. It holds the fCode making the card being recognized by open firmware as bootable. Even with a non fCode- or maybe no bios at all, the card works in Macs, but will not boot.

The two files are flashed seperately, each with it's own command.
 
Last edited:
  • Like
Reactions: DearthnVader
The settings on the card are stored in non-volatile memory (separate from the firmware and BIOS). You can configure the card in one machine and take it out and put it into another and the setting will not be lost.
Thanx for the explanation!

And great find! So now - finally - we have a bootable option for the G5 which is faster than it's own SATA 1. Something like this has not been available before as, even marketed as "SATA 2", to my experience the Firmtek-Seritek never exceeded r/w speeds of ~ 150 MB/s.

So great progress for the old machine! Looking forward to flash and test tomorrow.
 
Last edited:
So i report back with kind of partial success:

Configuring the card with lsiutil, DOS version, worked fine as described. I get the promised speeds of over 270 MB/s read/write with bigger data blocks now, which is just twice of what the G5's onboard SATA offers.

Sidenote: The DOS-version of the utility needs DOS4GW.exe to run, just like the ATI flashers.

But, even with both of the files in the earlier posted archive successfully flashed, the card refuses to boot in my case. It well shows up in System Profiler with "Firmware: 1.26.0.0 Fcode 1.00.49", but with a different name than in your case: It's "SASWT4I" instead of (the correct) "SAS3041E". So anything else having to be configured with the utility or am i missing something else?
 
Last edited:
  • Like
Reactions: Amethyst1
So i report back with kind of partial success:

Configuring the card with lsiutil, DOS version, worked fine as described. I get the promised speeds of over 270 MB/s read/write with bigger data blocks now, which is just twice of what the G5's onboard SATA offers.

Sidenote: The DOS-version of the utility needs DOS4GW.exe to run, just like the ATI flashers.

But, even with both of the files in the earlier posted archive successfully flashed, the card refuses to boot in my case. It well shows up in System Profiler with "Firmware: 1.26.0.0 Fcode 1.00.49", but with a different name than in your case: It's "SASWT4I" instead of (the correct) "SAS3041E". So anything else having to be configured with the utility or am i missing something else?
It’s possible there is a resistor strapping or some other difference with your card. I will try to post detailed photos of my card this evening.
 
  • Like
Reactions: flyproductions
So i report back with kind of partial success:

Configuring the card with lsiutil, DOS version, worked fine as described. I get the promised speeds of over 270 MB/s read/write with bigger data blocks now, which is just twice of what the G5's onboard SATA offers.

Sidenote: The DOS-version of the utility needs DOS4GW.exe to run, just like the ATI flashers.

But, even with both of the files in the earlier posted archive successfully flashed, the card refuses to boot in my case. It well shows up in System Profiler with "Firmware: 1.26.0.0 Fcode 1.00.49", but with a different name than in your case: It's "SASWT4I" instead of (the correct) "SAS3041E". So anything else having to be configured with the utility or am i missing something else?
are you able to post a system profiler screen shot of how it shows up in PCI devices exactly?

and also then in Open firmware dev / ls of your device tree with the card plugged in and a drive attached :)
 
  • Like
Reactions: Amethyst1
Success complete now! 😃

Got it running at full speed and booting now!

Had to erase the boot services area of the flash with

Code:
sasflash -o -e 5

Also had to extract the Fcode part of the bios-file posted, beginning at offset 0xB200 and ending at offset 0x12400, save this in a seperate file (see attached) and flash only this to the card. The complete bios-file also has an x86 bios image in front of the Fcode part, which seems to have blocked the Fcode from beeing "seen" by OF in the Mac in my case.

But now everything works perfectly fine!
 

Attachments

  • sas1064e_b3_fcode.rom.zip
    14.3 KB · Views: 126
Success complete now! 😃

Got it running at full speed and booting now!

Had to erase the boot services area of the flash with

Code:
sasflash -o -e 5

Also had to extract the Fcode part of the bios-file posted, beginning at offset 0xB200 and ending at offset 0x12400, save this in a seperate file (see attached) and flash only this to the card. The complete bios-file also has an x86 bios image in front of the Fcode part, which seems to have blocked the Fcode from beeing "seen" by OF in the Mac in my case.

But now everything works perfectly fine!
So now cards like this one maybe flashable?

Not my listings:



Also I wonder if the can get any of the PCI-X SAS LSI cards working:


And I wonder if they would work as 64Bit PCI???
 
  • Like
Reactions: LightBulbFun
So now cards like this one maybe flashable?
...or in some cases, as for @Tratkazir_the_1st and @randomdamage, just work without any flashing at all if the Fcode-bios is allready present!

I think in my case the problem was the card beeing INTEL-branded, which makes it appear as "SASWT4I", having a slightly different firmware and no Fcode image on it out of the box. I returned to the firmware it came with. And now, with the Fcode-only bios image, it works just fine.

But the real trick, giving it any advantage over the G5's stock SATA, was to set the minimum link speed to 3.0 Gbps via lsiutil. So thanks for finding and posting this again!
 
Last edited:
  • Like
Reactions: Amethyst1
Yes, this is not Photoshop! 😁

sas.jpg
 
...or in some cases, as for @Tratkazir_the_1st and @randomdamage, just work without any flashing at all if the Fcode-bios is allready present!

I think in my case the problem was the card beeing INTEL-branded, which makes it appear as "SASWT4I", having a slightly different firmware and no Fcode image on it out of the box. I returned to the firmware it came with. And now, with the Fcode-only bios image, it works just fine.

But the real trick, giving it any advantage over the G5's stock SATA, was to set the minimum link speed to 3.0 Gbps via lsiutil. So thanks for finding and posting this again!

Here is a link to the specific card I bought:
https://www.ebay.com/itm/204155881774

Have fun! 😀
 
Can you please explain what exactly we need to do?
As you want to go the way i did, using DOS, you need a bootable floppy just as for video card flashing. Copy both of the executable files in the archive attached to it. As lsiutil is free to use, i hope posting it is ok. Boot a PC with the SAS-card in it of the floppy. At the prompt launch the utility by just typing "lsiutil" and enter.

Then exactly follow the steps described in the nice step-by-step-guide, @randomdamage posted in #95.

When done successfully, your card should show similar speeds to the ones you see in my screenshot.
(BTW, which 2GB RAM sticks do you use?)
GEIL DDR2. Not my auction. Sure they are 6400, but these are backwards compatible.
 

Attachments

  • lsi_utility.zip
    401.4 KB · Views: 85
Last edited:
  • Like
Reactions: Amethyst1
So just came across some smaller problem:
Booting with the SAS-controller works perfectly fine by holding the option key. But by setting the boot-volume with the control panel does not. Seems like the the card / volume attached to it takes a bit too long to appear at boottime. So always another disk directly attached to one of the Quad's SATAs takes over.

So question is, is there any way to set up a delay, maybe in NVRAMRC, the Mac has to "wait" for the boot volume beeing set to appear before switching to another one?

Edit: Problem seems to have "magically" disappeared. Quad booted six times in a row with the "right" disk now. In the end it seemed to have been no more than a bad cable connection, as a bit of wiggeling / un- and replugging seems to have solved it.

So for now, nothing more to add than two pics of unleashed G5 storage power! 😝

storage_cards.jpg


In the upper slot the SAS-controller, 270 MB/s, booting!
In the lower the M.2 AHCI-SSD doing 800 MB/s.

...and, from the back, a bit of poor man's RGB too!

interfaces.jpg
 
Last edited:
Here is a link to the specific card I bought:
https://www.ebay.com/itm/204155881774

Have fun! 😀
So I went ahead and ordered this card along with:


I'm going to test it in a Beige G3/ PM QS/ PM MDD/ PM G5 1.6Ghz PCI in the PCI slots and compare it to the SIL3112 to see if it offers better performance.



That is assuming it will work in some or any of these Macs.

The PCI to PCI-E card will be on the slow boat from China so it's going to be early March before I have any results.

Quick question that may have already been covered, does this card work with Tiger or it requires Leopard?
 
So a little follow up, as i cheered a bit to early:

As one of the more physical weaknesses of these cards turned out them having absolutely zero mechanical support for the cable connection of any kind on their ports. So if one does not want to end up in an very annoying endless dissappearing-reappearing-loop of the startup disk, the best bet should be to only use premium cables and quality SSDs! ☹️

I just ordered some of these and am actually mirroring the boot drive to a SAMSUNG 850 evo to exclude the cheapo-SanDisk drive as the culprit. Will report back!

Edit: Most likely wasn't the cabeling but this SanDisk-piece of crap which seems to not play too nice with the SAS-card. After migrating to the SAMSUNG i have yet to come across any booting-problems or even complete dissapearances of the disk.
 
Last edited:
  • Like
Reactions: Amethyst1
It requires Tiger Server or Leopard (any version).
in theory you should be able to grab the kexts from tiger server and install them into tiger client

and I imagine you need the universal build of tiger server for these kexts


this has been quite fun to see

as I have a feeling that these cards work with the G5 Quad by complete fluke

in that we very fortunate that they have an fCode ROM but I would say its its not for G5's per-say more POWER Servers and the such like, but being fCode means it does with a G5, and then Leopard (and Tiger Server) has drivers for these cards because the Intel Xserves use LSI SAS HBA's for their 3 SAS bays IIRC, so it all just comes together to happen to work for us :)

(much in the same way that AHCI drives do happen to work in PPC Leopard, in that there just happens to be a driver for em thankfully)


@DearthnVader it might be worth seeing if you can find a PCI-X card that has an fCode ROM, as that would allow a much more direct connection, and let you use the 64 bit part of the PCI Slots in a G3 B&W or G4 :)
 
  • Like
Reactions: Amethyst1
in that we very fortunate that they have an fCode ROM but I would say its its not for G5's per-say more POWER Servers and the such like, but being fCode means it does with a G5, and then Leopard (and Tiger Server) has drivers for these cards because the Intel Xserves use LSI SAS HBA's for their 3 SAS bays IIRC, so it all just comes together to happen to work for us :)
That's what you can be quite sure about, that this card has not been made with the G5 in mind. Doesn't seem to keep this 15 $/€ part from beeing faster than anything that (for a fortune of money) was.

@DearthnVader it might be worth seeing if you can find a PCI-X card that has an fCode ROM, as that would allow a much more direct connection, and let you use the 64 bit part of the PCI Slots in a G3 B&W or G4 :)
As far as i see it, the PCI-X card posted earlier uses the same chipset as the one i have. So why souldn't the fCode work for it too? Most likely, as the 3041E, it has two separate areas on the flash. One for the firmware and one for the "boot services"/bios, which, according to the SASFlash-manual, can be x86, fCode or EFI BSD or even combinations of them. ...which did not work for me. I, as described above, had to remove the x86-part of the bios to get the card booting in the Quad.
 

Attachments

  • SASFlash_user_maual.pdf
    309.3 KB · Views: 211
  • Like
Reactions: Amethyst1
@DearthnVader it might be worth seeing if you can find a PCI-X card that has an fCode ROM, as that would allow a much more direct connection, and let you use the 64 bit part of the PCI Slots in a G3 B&W or G4 :)

As far as i see it, the PCI-X card posted earlier uses the same chipset as the one i have. So why souldn't the fCode work for it too? Most likely, as the 3041E, it has two separate areas on the flash. One for the firmware and one for the "boot services"/bios, which, according to the SASFlash-manual, can be x86, fCode or EFI BSD or even combinations of them. ...which did not work for me. I, as described above, had to remove the x86-part of the bios to get the card booting in the Quad.

I did give it some thought, but I thought the chips would be different, as one is PCI-E and one is PCI-X.

If @flyproductions thinks the chips are the same I'll order one of these too?
 
Hmmm...

The LSI PCI-X card looks to be keyed for 3.3v PCI, where as my QS it keyed for 5v PCI.

Card won't even physically fit!

MDD, same deal 5v PCI.

G5 looks to be compatible, 3.3v 64bit PCI.

I did notice that some where not universal voltage keyed! as you say bit of a non stater for G4's, thats a bit of a bugger then!

as it does look like you can get them to work in PCI-X G5's

certainly work in MacOS X and there is talk of an fCode ROM being available, so in theory that should complete the package so to speak to make em fully bootable


 
  • Like
Reactions: Amethyst1
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.