Aaargh. A system crash just ate my reply. Re-attempting:
But that seems almost like a parallel project to trag's, at least according to my poor tech reading skils. Aren't you going to have a clash of bus line requirements? Or would you be depending on the FPGA to arbitrate that?
Further thought: would it be possible to attach a new Cardbus cage (and chipset if required) directly to the FPGA and have it appear to the CPU bus as a PBX/T-Rex combo? Or does that lead you down the black hole of reverse engineering poorly documented ASICs? I mean, it's not as though dead & dying 1400s & 5300s for cannibalizing are unobtanium.
) but isn't that kinda good enough? VM appears as RAM to the OS and apps, neh? Is there some sort of overhead with using RAM-as-disk-as-VM-as-RAM - and even if there is, is it worth bothering about?
A wild brainfart appears: RAM on a PCMCIA card for RAM-challenged Powerbooks*. I know that we can already do much the same thing with CF, but is there any speed advantage** using RAM*** rather than flash?
* 190, 1400, 5300, 2400, 3400 - ie, those with Cardbus/PC Card slots, but proprietary & limited RAM.
** Presumably more so on Cardbus (PCI) than PC Card (PBX-T-Rex)
*** Raw ICs (w/wo sockets), or headers for SIMM/DIMM/SO-DIMM
OK, I can see the utility of that: you would have access to all those MacOS compliant cards - and more importantly their drivers - from back in the day. And at a guess, you would be re-using ROM code from the 1400 to enable the PBX/T-Rex in other machines, yeah?Design the card with the PBX Bridged '030 headers / for the 1400's T-Rex enabled PCMCIA Card Cage
But that seems almost like a parallel project to trag's, at least according to my poor tech reading skils. Aren't you going to have a clash of bus line requirements? Or would you be depending on the FPGA to arbitrate that?
Further thought: would it be possible to attach a new Cardbus cage (and chipset if required) directly to the FPGA and have it appear to the CPU bus as a PBX/T-Rex combo? Or does that lead you down the black hole of reverse engineering poorly documented ASICs? I mean, it's not as though dead & dying 1400s & 5300s for cannibalizing are unobtanium.
Correct me if I'm wrong (as usual
) but isn't that kinda good enough? VM appears as RAM to the OS and apps, neh? Is there some sort of overhead with using RAM-as-disk-as-VM-as-RAM - and even if there is, is it worth bothering about?A wild brainfart appears: RAM on a PCMCIA card for RAM-challenged Powerbooks*. I know that we can already do much the same thing with CF, but is there any speed advantage** using RAM*** rather than flash?
* 190, 1400, 5300, 2400, 3400 - ie, those with Cardbus/PC Card slots, but proprietary & limited RAM.
** Presumably more so on Cardbus (PCI) than PC Card (PBX-T-Rex)
*** Raw ICs (w/wo sockets), or headers for SIMM/DIMM/SO-DIMM