• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

A high-quality SATA PCI 2.5" hard card, to celebrate SATA's 20th birthday

ArmorAlley

Well-known member
The chipset is PCI and the firmware is designed specifically for PCI PPC macs, it would be both extremely difficult, and the end result would be something bottlenecked by the host's bus speed worse than PCI cards are. It would be an entirely new product, more complex than the original.
This is what I feared. Thanks for this.
 

rabbitholecomputing

Vendor The First
If I order one, will it come with the PM39LV040 or MX29LV040 chip to use the 1S2 ROM for booting System 7? I'd really appreciate the boost for my Power Macintosh 9500.
The board has an AM29LV040 soldered down from the factory, so yes. While that's not one of the two ICs you cite, it will also work.
 

Phipli

Well-known member
As @Phipli already stated, that's not a practical possibility, because you'd need to create an 040 bus to PCI bridge, which would be an absurdly expensive thing to do, to say nothing of the engineering costs associated with bringing something like that into existence.
And the ROM driver is for openfirmware, which doesn't exist on 68k macs. So you'd have to write new low level drivers.
 

Melkhior

Well-known member
As @Phipli already stated, that's not a practical possibility, because you'd need to create an 040 bus to PCI bridge, which would be an absurdly expensive thing to do, to say nothing of the engineering costs associated with bringing something like that into existence.
I agree that it's (emphatically) not a practical possibility, but the culprit would be the software IMHO (as it almost always is). Xilinx has both PCI and PCIe root port available for their FPGA, and there is bridge chip available both ways. Hardware-wise, mapping a PCIe root port in the slot space of the Mac and using the superslot for the PCIe space would probably not be particularly complex. Adding a SATA PCIe chip would be easy. Artix-7 have native pins for PCIe (GTP), so any board breaking those out would be enough. The Trenz TE0712 is would be my choice (I've looked at it for a hypothetical SBusFPGA 2.0). You could probably assemble a board like that for less than 500€, with most of the cost sunk into the TE0712.

However, for SATA on 68040 (or even '030), a shortcut would be to use said GTP pins to drive SATA directly. This is supported in the Litex infrastructure using the LiteSATA IP. Extra hardware is just a handful of capacitors and the SATA connectors. You still need a board like the TE0712 (the Ztex I use doesn't expose GTP pins). That was my intention for SBusFPGA under NetBSD, but is still theoretical (the TE0712 has been out of stock for months if not a couple of years, and isn't cheap). Probably less than 450€ for the assembled board, again the TE0712 is the expensive bit.

However in either case, one would need to write the appropriate drivers(s). Pure SATA might be easier, but is still a fairly complex piece of software that may or may not fit in the SCSI Manager framework. SATA on PCIe is even worse, as in addition to SATA one would need configuration and enumeration of the PCIe domain.

The hardware is quite feasible, the software would require a LOT of efforts.
 

herd

Well-known member
With all due respect, this is not relevant to the conversation in any way, shape, or form. We're discussing a two port SATA card, which is more than sufficient for most use cases.

I applaud your accomplishment and appreciate your support of the community. If it's not too far off topic, I would like to express my interest in a 64-bit PCI storage option for PowerPC Macs. There are a few existing cards that can approach the 267MB/s theoretical speed of this bus, but (to my knowledge) none are bootable. If a product were available that would let me boot a newer drive (SATA, SAS, nvme, etc.) that is significantly faster than IDE or firewire, I would buy several.
 

kerobaros

Well-known member
If a product were available that would let me boot a newer drive (SATA, SAS, nvme, etc.) that is significantly faster than IDE or firewire, I would buy several.
I am 100% on board with your need for speed, but I would love to know what you'd use all that throughput for (other than setting benchmarks on fire)
 

Phipli

Well-known member
Are there instructions somewhere for obtaining and flashing the firmware needed to use this on a G4 Macintosh?
 

bigmessowires

Well-known member
I had skimmed that, but it's a long thread discussing multiple brands of PCI cards, chip sets, and ROM versions and it's hard to discern what information is still current or relevant to the RHC card. The linked post is talking about physically replacing the ROM chip, if I understand. I'll take a closer look later.
 

dosdude1

Well-known member
I had skimmed that, but it's a long thread discussing multiple brands of PCI cards, chip sets, and ROM versions and it's hard to discern what information is still current or relevant to the RHC card. The linked post is talking about physically replacing the ROM chip, if I understand. I'll take a closer look later.
The best thing to do is flash my patched/compressed version of the SeriTek ROM, which can fit onto the original 128K EEPROM that a lot of these Sil3112 cards have. Though this RHC one should have a 512K EEPROM installed already if I'm not mistaken. Regardless, the same compressed ROM can be used. You can flash either from a PC using Flashrom, or using my patched SeriTek firmware updating utility on the target machine from within OS9. With the latter option, you simply install the card into your desired machine, boot OS9, download and run the tool, and from there simply select the option to flash the card. Latest resources can be found on my post here: https://68kmla.org/bb/index.php?thr...ng-easier-way-using-flashrom.7013/post-472774
 
Top