Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Guide to flashing PC SIL3112 SATA cards for Mac
- Thread starter harrymatic
- Start date
- Sort by reaction score
No, because I am going to finally register my firmware what I wrote with authorities in my country - and toast that guy from Hong Kong for good. I am in Europe since a while. And finally, the firmware is now mine, mine, mine.
The good news: working (again) to re-write the code and unify the 3112 / 3152 / 3114. MacOS9 included.
Started earlier - but ultimately wanted to play safe and did not wanted to conflict with than-employer.
Other good news: there seem to be a smaller batch of 3124. These are unfortunately PCI, not PCI-X.
However, 3124 has no problem with sleep on Quicksilver. Going to write the MacOS-9 driver but maybe it won't support port multipliers on "9", just only on "X" as everyone knows.
All recent 3112, 3114 and 3152-s need a similar regulator that what we had back in the FirmTek days.
Otherwise no sleep on QS. Unless you re-solder the regulator.
My idea of the neat 16-bit Macronix chip on SeriTek/1SE2 turned to be a major piracy-roadblock: not a single 1SE2 was pirated. Unfortunately the Chinese manufacturers now fell back to cheap EPROMS and cheap regulators.
Hint: anyone who want to make a "pirated" SeriTek/1S2 work in Quicksilver better to get a soldering iron, a MIC29150 3.3V Regulator and a lot of luck. And ask me nicely before you do.
I have many more plans, hope I have all the energy.
Last edited:
Welcome to the forum--your work would definitely be appreciated here! I've been hoping to get a bootable 3124 card working, but the firmtek ROM is buggy in my experience.
Can you please describe the bug? I wrote that boot code, so I can hopefully fix it.
Explanation: if you take the risk, remove the old regulator and solder the new one... and ask me to use my firmware and drivers for your own benefit and to spread and to appreciate my work: that's OK, feel free (and if we ever meet- you ow me an espresso).
However if you steal it for business and even mask the "SeriTek" on the screenshot in your auction... and you sell hundreds if not thousands cards yearly for at least $55 profit / card (without even taking care of QuickSilver Mac and soldering a new regulator) - than it's not only not OK, but I am going to report that case to the police.
Just got an e-mail from MacOS 9 Lives. Somehow I can't log in there, but Ataman Honcho Honchev is the same as (S)ATAMan. Since 1999 or even earlier. ;-)
Apologies for using this forum for "MacOS 9 Lives" messaging - but there is some web site problem there for me.
Hope they fix it. They know my e-mail, it's still the same.
I disappeared from MacOS 9 Lives for a while because first of "long" COVID (I got rid of THAT virus-bug, but it wasn't easy) first, than due concerns regarding my than-employer. Now there are no concerns. ;-)
Apologies to "MacOS 9" guys, I promise to be a good boy and much more diligent.
Apologies for using this forum for "MacOS 9 Lives" messaging - but there is some web site problem there for me.
Hope they fix it. They know my e-mail, it's still the same.
I disappeared from MacOS 9 Lives for a while because first of "long" COVID (I got rid of THAT virus-bug, but it wasn't easy) first, than due concerns regarding my than-employer. Now there are no concerns. ;-)
Apologies to "MacOS 9" guys, I promise to be a good boy and much more diligent.
As reported in the other thread to this topic, i have some FS8860 33C on my 3114 based card working nicely in a Quicksilver. And, as it has the same form factor as well as pinout as the stock AMS1117, it offers a "little" more pleasant look over the MIC29150. I just checked and wake from deep sleep seems to work too.
But, as the wiebeSATA firmware is used on the card - of course - no OS 9 operation. Would be nice to see, if, with your firmware all this works with OS 9 too. 😉
The difference between the 3112 and 3114 is some errata regarding a microcode bug in 3114, also affecting the 3512.
Back than I had to restrict writes on the 3512 (that's the cardbus card) being active on one channel only.
That did work (kind of), but we never got the official report from our Silicon Image contacts.
Probably because they either resigned or were laid off. The usual brainless operation of a brainless corporation...
(When the workers are fired to let the managers manage each other).
In any case I probably know now what is the workaround - the only problem that I don't have a PCI-based 3512.
So going to test the 3114 first and than let's see, what happens with 3512.
Otherwise I have the documentation and 3114 is by no means more difficult than, say, Vitesse.
Currently working precisely on 3114, but the way I don't need to handle million versions (like external, internal, 3112, 3114, 3512) on both "9" and "X".
The next will be probably the 3124 for "9". If anyone experiences bugs with the current FirmTek boot process involving 3124 on macOS X, please report it here or on " '9' lives " or both.
I got a 64-bit PCI silicon image 3124 card to try in a quicksilver with Mac OS X 10.4.11. The card works fine with the driver installed (SiI3124_2.0.3_Sil_Pkg.pkg) but when flashed with the firmtek ROM (SeriTek2SE4_5.3.2.dmg) there will be little glitches in audio playback, spotlight indexing will crash, and trying to copy large amounts of data will usually fail. The logs show messages for some of these errors. I've tried a couple different 3124 cards in a few different G4 machines with the same result. If I flash the card back to the stock ROM and re-load the driver then it works fine again, except it's not bootable. The firmware instructions state, "If your current firmware is working for you, FirmTek does not recommend updating the firmware." I have not been able to find any of the older versions to test.
It would be nice to get these cards working because they're significantly faster than the 3112 cards:
I've been testing them with different SSDs attached. Could the SSD be "too fast" for what the firmware is expecting?
Yes, that i can confirm! Only one of the ports, the lower one at the card's rear end, seems to work.
Thanks, that will be re-tested. I have to consult with the open source Linux driver to see, how they manage the DMA.
It very well could be, the manual has some inconsistence and / or my understanding was not good.
According the description in the reference manual 3124 has one of the weirdest set of API-s with certain elements of SG list playing a "kangaroo" role. I had weeks of head scratching how to implement these "kangaroo"-s, the worst was in Open Firmware. But as I recall from the before-COVID-weeks when out of curiosity I looked at the 3124 Linux driver - there was no trace of "kangaroo". Same about 3132, it has the same internal as 3124 (just being a PCIe).
It never worked for FirmTek properly - so out of frustration I just double-buffered everything, avoiding the "kangaroo".
That worked... except that my 3132 driver was dog slow. Time to look again.
I mean, SIMULTANOUSLY. I.e. both ports should work - but writes never happen the same time.
When two drives are in RAID-0 and you write to that RAID from the PowerBook G4, one port will silently wait until the other port does not finish the actual transaction. This is transparent to the end-user, except that the performance is quite worse than expected (and defeats the purpose of the RAID-0).
Such thing happened only on 3512 and we had 3512 only inside of cardbus card. So I was thinking that this problem is related to cardbus controller. Probably I was wrong.
I have a G4 notebook in my other home as well as the FirmTek 3512 cardbus card. Will test it later on. So far, the suggested workaround was already implemented, but I am not ready to test yet (today).
If everything is right, the new 3114 driver for "X" will be ready to debug next week.
The "9" driver is already written, but I will start with "X" on the 10.5.8 and dual-proc G4 MDD 1.42 (400 FW, the original processor was upgraded).
Why I am starting with 10.5.8? Because that setup is the easiest for two-machine debugging via FireWire.
I have two machines with the "Stealth" serial port for "9", just lazy to set everything up. Maybe that will come as well.
The development itself is done under 10.13.6 (probably will move to Ventura this year) under Parallels, Parallels has a Snow Leopard VM. I like that because I am browsing, e-mailing, etc. on a relatively up-to-date system while the development is on a VM inside that system. These days a 43" TV is 200 Euro at "Norma" opposite the e.LeClerc, so I am happy.
With fwkpfv I can see all the driver output from the 10.5.8 inside a terminal window on the 10.13.6
Not sure how to deal with the "Stealth" port of "9" yet... I am sure there is a hack to get the data from serial port of "Stealth" to the USB.
Anyone knows the internal code names for Silicon Image 3512 and Silicon Image / CMD 680?
The original Silicon Image internal code names for two of their first gen. SATA ASIC-s are:
"Athena" (3112)
"Taurus" (3114)
For the second gen.:
"Sable" (3124)
I am thinking to name the corresponding drivers matching the internal code names.
Before these code names appeared in the "model" property generated by Open Firmware driver.
Even I don't remember what, say, "WhoNowsTek/1SD4KE-M24" could mean, LOL.
So going away from "internal" (no hot-swap) vs. "external" (hot-swap) and from differentiating how many internal or external connectors that card has.
Maybe, say, use "Athena2" versus "Taurus4" versus "Sable4" naming, the last digit indicating the number of SATA channels?
The original Silicon Image internal code names for two of their first gen. SATA ASIC-s are:
"Athena" (3112)
"Taurus" (3114)
For the second gen.:
"Sable" (3124)
I am thinking to name the corresponding drivers matching the internal code names.
Before these code names appeared in the "model" property generated by Open Firmware driver.
Even I don't remember what, say, "WhoNowsTek/1SD4KE-M24" could mean, LOL.
So going away from "internal" (no hot-swap) vs. "external" (hot-swap) and from differentiating how many internal or external connectors that card has.
Maybe, say, use "Athena2" versus "Taurus4" versus "Sable4" naming, the last digit indicating the number of SATA channels?
Hmm, maybe it’s just the WiebeSATA-firmware. But with the 3114-card, i only get a working connection with the lower rear connector. Nothing on the upper one as well as on any of the two on the upper side of the card. While with the (otherwise nearly identically looking) 3112-card i have, both of the connectors are working.
I have two of those 3114s, which are both showing these same symptoms.
i bought one of those cards from local338 and i couldn't gt it working right in my g3 or g4 at the time. if you guys have the firmware and whatnot, maybe i can give it a try with flashing it again to get the sata up and running again. i will post pics of it along with the chips and the "special" soldering it has
*edit*
here are the pics of the card i bought
*edit*
here are the pics of the card i bought
Last edited:
Yes, I know "local338" - he is the HK pirate. In 2019 I wrote him to stop what he is doing.
Not only I did not achieve anything - eBay would not remove his listings either.
Thanks for the pictures. Would be interesting to swap a card with you later on.
Could be well that your card has hardware defect. The job "local338" did with the regulator is not bad tho.
Not only I did not achieve anything - eBay would not remove his listings either.
Thanks for the pictures. Would be interesting to swap a card with you later on.
Could be well that your card has hardware defect. The job "local338" did with the regulator is not bad tho.
I've updated flashrom/directhw/pciutils to work in Mac OS X 10.4/10.5 on PowerPC but I don't know if it works since I don't have a satasii PCI card.
https://68kmla.org/bb/index.php?thr...ng-easier-way-using-flashrom.7013/post-475781
Not reallly, myself and many others had issues with the card, it only worked somewhat and was pretty much unusable.
I "fixed" the issue by soldering an OEM regulator on the card, but that makes it unusable in my QS.
If you get the 64bit 3124 cards working with OS 9 I'd be interested to test those too!