Processor question

[temporary]

I have a question;

If you get an accelerator board for an Amiga 1200, just a regular 1230, as an example. What happens to the original processor in the Amiga? Does it have any function left, execpt maybe in some part of the boot process?

I assume not, since you get no info on it with the cpu command in shell. But still, it would be interesting to know exactly what happens to it.

I guess one idea is to pull it out and see what happens, but since I know those old pins can be a bit fragile, perhaps it's best to avoid doing so :-D

[Piru]

The original CPU is shut down completely when the accelerator's CPU is in use. The original CPU isn't used at all, not even at power on.

IIRC the A1200 expansion slot has some CPU override pins etc which the accelerator use to override the original CPU.

[temporary]

Alright, thanks!

[bloodline]

Piru wrote:
The original CPU is shut down completely when the accelerator's CPU is in use. The original CPU isn't used at all, not even at power on.

IIRC the A1200 expansion slot has some CPU override pins etc which the accelerator use to override the original CPU.

Is the CPU halted (i.e. sleep mode if they had such a thing 15 years ago), Or is it still drawing full current?

I've suspected that it's still on, but I've not checked the 020 temp.

What does intregue me, is how the A600 030 accellerator works since the board fits over the 68k... I would have imagined there would be conflicts :-?

[Linchpin]

I always wondered that too, you'd think the system would either see nothing or both...

Thats why im not an Electronics engineer...

[bloodline]

LinchpiN wrote:
I always wondered that too, you'd think the system would either see nothing or both...

Thats why im not an Electronics engineer...

Well, it's been my assumption that the clip over socket holds the 68k rest pin high so the 68k is in a permanent reset... but then the 030 would not be able to see any reset signals from the motherboard (does it need to?) :-?

[Zac67]

You can't keep _RST high as the driver pulling it down would be overstressed or even burn out.
I guess they use/mimic some bus request/bus grant stuff to get the 68k out. The '030 requests the bus and just never gives it back. (In contrast to other architectures the M68k itself is fully DMA capable since it is possible to arbitrate for the bus.)

Making use of both CPUs would imply SMP logic in both OS and system/turbo board, which simply isn't there. Running a PPC concurrently is somewhat different as it has an OS of its own, nothing that would be called SMP.

[bloodline]

Zac67 wrote:
You can't keep _RST high as the driver pulling it down would be overstressed or even burn out.
I guess they use/mimic some bus request/bus grant stuff to get the 68k out. The '030 requests the bus and just never gives it back. (In contrast to other architectures the M68k itself is fully DMA capable since it is possible to arbitrate for the bus.)

I wasn't aware the 68000 bus was as sophisticated as that, I'll have to check my datasheets :-)

[Wol]

Hi all,


ON power up, both processors go thru reset stuff.

The eccelerator processor then asserts BUS REQUEST, after
a short finish up time, the existing  processor releases
the buses (goes into high impeadance state) and asserts
BUS GRANT. The eccelerator processor then has full contol
over the buses....and so on.


Wol..

[KThunder]

i always though it was kind of strange that the amiga has so many accelerators and has always had the possibility for multiproc stuff. i know that the a3000 and a4000 cpu slots have the capability of shutting down the host cpu or not. and a500 a1000 and others do since they have the cpu in place anyway and bypass it.
amiga is probably the only computer in the world were most people have more than one cpu but noone ever can use more than one. maybe if it was written into the os.

[Zac67]

W/o large caches there's not much point in putting two CPUs on the same bus, they'd be waiting for RAM access most of the time. An '040/060 dual system with (separate) cache(s) would be something different though. Makes me think of Dave's Gemini research, which had never been able to run AmigaOS afaik.

[bloodline]

Zac67 wrote:
W/o large caches there's not much point in putting two CPUs on the same bus, they'd be waiting for RAM access most of the time. An '040/060 dual system with (separate) cache(s) would be something different though. Makes me think of Dave's Gemini research, which had never been able to run AmigaOS afaik.

I think he stated that it would run a "special" version of exec.

[temporary]

But let's not forget old Alice, Lisa, Paula etc. Perhaps not central processor units, but processor units never the less  :-P