MOS8_030
Well-known member
I'm looking back through some of my old work emails from ~2005 and we were running two Apple test suite tools. One suite was called "iSys" and the other suite was called "INDY". We only ever tested A8 with 10.3 on the MDD G4 systems. I believe the "INDY" software was the old stuff.
Here's an interesting bit I found from that time from an engineer at Sonnet:
"These are two different things as it turns out. The boot ROM does have to change, typically for PVR changes, but also for cache reporting and processor speed reporting - the new PLL multipliers were not part of the old speed tables.
So, one set of changes provides support in the boot ROM. This takes the boot process from initialization, memory configuration, cache configuration, some testing, and then into Open Firmware.
Open Firmware then looks for the official boot device; if OS X is selected, it loads a bootstrap program ("BootX") which does some prep, then reads the kernel image from disk into memory, and jumps to it.
Once the OS X kernel starts booting, Apple reads the PVR again with a mfspr instruction and then looks up in a processor table all the features and characteristics of the processor. Interestingly, they do not use the information for the boot ROM data structures, but read it straight out of the processor.
Frankly, we may have had some hand in that, as OS X was able to boot on some old, old machines (9500 class) using our processor upgrades that never had the PVR value set correctly in the boot data structures.
So, this is a second set of changes, this time in the OS X kernel - and very minor, as I stated. I don't see any way around it though - Apple reads the pvr, and there is no intervention in the code stream until it is used to look up the processor info in the table.
I routinely have to make custom kernels for new processors (7457, 7447A for example) before Apple provides official support, but we never thought much about it not being part of future OS X versions till the Apple WWDC announcement.
I'm happy to provide any further info on the issue, as we really like the 7448 so far and would very much like to use it in our products when it becomes available."
And here's a note I sent to my boss around the same time.
"We are using a custom kernel now with 10.3.5. That's because the A7PM wasn't out when 10.3 was released. However, I'm sure 10.3.9 supports it.
(Your laptop requires at least 10.3.4 I believe)
To get A8 to work we had to do a firmware update so the system (hardware) would recognize the part. That's a fairly trivial update.
The same custom 10.3.5 kernel works with A8. (I don't know why.)
However, I've tried to boot 10.4.x with A8 and the OS won't load.
I assume because the OS doesn't recognize the PID.
So, yes, in order to run A8 on 10.4.x Apple will have to update the kernel. He's absolutely right that because of this it may be very hard or impossible to make a 7448 upgrade for older Macs.
Of course Apple has always discouraged upgrades anyway, so any support
for 7448 wouldn't be done just for them....
He (and we!) can only hope that Apple will use the 7448.
It makes it kinda hard for the upgrade companies to move forward.
If you would like more information I can make some discreet inquiries
to the Northern folks. I'd like to get a kernel for A8 and 10.4 anyway."
(I never did get any support for 10.4 to run on the 7448)
I also found some comments about getting the MPC8641D (dual-core 7448) to run on the Mac back then, but that was after extensive hardware/software hax.
Here's an interesting bit I found from that time from an engineer at Sonnet:
"These are two different things as it turns out. The boot ROM does have to change, typically for PVR changes, but also for cache reporting and processor speed reporting - the new PLL multipliers were not part of the old speed tables.
So, one set of changes provides support in the boot ROM. This takes the boot process from initialization, memory configuration, cache configuration, some testing, and then into Open Firmware.
Open Firmware then looks for the official boot device; if OS X is selected, it loads a bootstrap program ("BootX") which does some prep, then reads the kernel image from disk into memory, and jumps to it.
Once the OS X kernel starts booting, Apple reads the PVR again with a mfspr instruction and then looks up in a processor table all the features and characteristics of the processor. Interestingly, they do not use the information for the boot ROM data structures, but read it straight out of the processor.
Frankly, we may have had some hand in that, as OS X was able to boot on some old, old machines (9500 class) using our processor upgrades that never had the PVR value set correctly in the boot data structures.
So, this is a second set of changes, this time in the OS X kernel - and very minor, as I stated. I don't see any way around it though - Apple reads the pvr, and there is no intervention in the code stream until it is used to look up the processor info in the table.
I routinely have to make custom kernels for new processors (7457, 7447A for example) before Apple provides official support, but we never thought much about it not being part of future OS X versions till the Apple WWDC announcement.
I'm happy to provide any further info on the issue, as we really like the 7448 so far and would very much like to use it in our products when it becomes available."
And here's a note I sent to my boss around the same time.
"We are using a custom kernel now with 10.3.5. That's because the A7PM wasn't out when 10.3 was released. However, I'm sure 10.3.9 supports it.
(Your laptop requires at least 10.3.4 I believe)
To get A8 to work we had to do a firmware update so the system (hardware) would recognize the part. That's a fairly trivial update.
The same custom 10.3.5 kernel works with A8. (I don't know why.)
However, I've tried to boot 10.4.x with A8 and the OS won't load.
I assume because the OS doesn't recognize the PID.
So, yes, in order to run A8 on 10.4.x Apple will have to update the kernel. He's absolutely right that because of this it may be very hard or impossible to make a 7448 upgrade for older Macs.
Of course Apple has always discouraged upgrades anyway, so any support
for 7448 wouldn't be done just for them....
He (and we!) can only hope that Apple will use the 7448.
It makes it kinda hard for the upgrade companies to move forward.
If you would like more information I can make some discreet inquiries
to the Northern folks. I'd like to get a kernel for A8 and 10.4 anyway."
(I never did get any support for 10.4 to run on the 7448)
I also found some comments about getting the MPC8641D (dual-core 7448) to run on the Mac back then, but that was after extensive hardware/software hax.
Last edited by a moderator: