Page 2 of 8

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 6:45 am
by Dave
New idea, though it is not asymmetric processing. A bridge board that puts 128K of common memory between two QLs. Both can have read and write access to all of it. I wonder, besides high speed two player gaming, what it could be used for.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 10:43 am
by Martin_Head
Dave wrote: So far we have (including IRC discussion and computer lab discussion:

A 6502 and Beeb tube implementation
80486DX and run Windows 3.1 in a window on the QL? RiscPC style!
ATmega, and the variety of IO/datalogging it could provide.
A large FPGA as the ultimate dev board that can be anything you want.
Transputers.... because transputers!
Turning the QL into the 2nd CPU and plugging it into a RiscPC.
What about a graphics processor. The QL operating system just tells the second processor to display a character at a particular position, or draw a line between two points. Then gets on with something else, while the second processor gets on with it. The QL side would not get bogged down dealing with high screen resolutions, or colour depths.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 2:09 pm
by Silvester
There was the humble Miracle MIDI interface where the Z80 waited for OS code to be uploaded from the ROM port: http://www.dilwyn.me.uk/docs/hardware/QLperip.zip

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 2:33 pm
by RalfR
Have the Miracle MIDI I/F ever run succesfully? I have never read about that.

And not that all. What about in conjunction with the Miracle WIN? Seems to be quite problematic to connect.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 4:51 pm
by Derek_Stewart
Hi,

How about connecting 2 QLs together via an interface to combine the 2 CPUs and machine resources.

I once thought about a ROM Port connection, to have a shard memory in the Rom Port on each QL. But the QL is read only without special interfacing like on the Miracle Hard Drive.

The Aurora had an enhanced Rom Port connection, with more possibilities.

But looking at the Issue 8 discussion, Rom Port is being removed.

I have some of the Miracle MIDI interfaces, I was going to connect a MIDI keyboard and use the Tracker software.

Just never got around to it, maybe next on the list.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 5:22 pm
by Dave
Derek_Stewart wrote:But looking at the Issue 8 discussion, Rom Port is being removed.
Issue 8 is definitely retaining the ROM port.

What we're doing is ending the motherboard between the ROM port and joystick port. An extender will implement the ROM port, and can either be used or - if you're not using the ROM port - removed so you can fit more of your expansion card inside the QL. The extender is neat because it can be bare and just implement the ROM port and an internal expansion area, or it can contain 16MB of DRAM.

Have you worked out a schematic, or got clear photos of the MIDI board? It sounds like it would be a very interesting thing to see. MIDI has been whispered in my ear more than a few times in the last 3-4 years.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 5:26 pm
by Dave
Martin_Head wrote:What about a graphics processor. The QL operating system just tells the second processor to display a character at a particular position, or draw a line between two points. Then gets on with something else, while the second processor gets on with it. The QL side would not get bogged down dealing with high screen resolutions, or colour depths.
HehehehahahahAHAHAHAHAHA!!!!!

Ok, so.... Sorry.... *deep breath*

Don't watch this space, but do watch another one down list for after the Issue 8 is finished.

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 6:04 pm
by RalfR
Derek_Stewart wrote:Hi,I have some of the Miracle MIDI interfaces, I was going to connect a MIDI keyboard and use tge Tracker software.

Just never got around to it, maybe next on the list.
Great to hear. I fear, Tracker was Turbo compiled (but with an older version). Any chance to contact the author to get the source code?

Re: Asymmetric multiprocessing...

Posted: Sat May 18, 2019 10:48 pm
by Derek_Stewart
Hi Ralf,

I doubt the author of the Tracker is available. But maybe Martin Head's Decompiler can restore the source code.

Re: Asymmetric multiprocessing...

Posted: Sun May 19, 2019 4:50 am
by 1024MAK
Maybe a 6809 (Hitachi 6309) so that OS-9 or UniFlex can be run...

Incidentally was/is OS-9/68K available for the QL?

And what about the other way round with a more modern fast processor (maybe in FPGA) being used as the main processor with the QL being used as the I/O processor... Maybe even a super fast 68K, Z80 (clocked at 20MHz actual CMOS MPU or faster if a FPGA version) or 6502 (W65C02S–14 clocked at 14MHz actual CMOS MPU or faster if a FPGA version)...

Oh, and for a basic system you don’t need dual port RAM. One third party Z80 second processor system for the Acorn BBC B used two PIO chips back-back for processor to processor communications. In this case a 6522 VIA for the 6502 and a 8255A for the Z80A.

Mark