68060 based QL system?

[Peter]

Because you opened the door. If it doesn't interest you that this door is opened, then it doesn't matter what threads we start - unless someone reproduces all your many years of work (which isn't going to happen) then the matter is moot.

Firstly, the largest part of my Q68 time went into the painful work of trying and debugging several different approaches for an FPGA CPU that is reasonably fast, compact, and can actually execute QL code. The final result was contributed back into an open source project. Maybe someone else would not easily get the same performance and QL specific system integration, but basically, my largest timeloss does not have to be repeated.

Secondly, I did a lot of driver development, everything was implemented OS-less by me first, and later moved into SMSQ/E. Again, the result is open source. I'm not sure my hardware work took longer than that.

Nobody would expect you to support these extended, out of band uses of your hardware. However, people aren't going to use the hardware because it's closed, only available as complete units with mark-ups that mean only 5% or 10% of the market is even being tapped.... It's an interesting curio. If the programmed Q68 FPGAs were made available for, say, €25, then your IP would be protected AND the tech could be put to other uses.

If a reply from me is desired, new topic, please.

[Derek_Stewart]

Hi,

Just to this topic back on track.

Can a DMA chip be interface to the Q60 ISA slot or would connection would have to be made to the CPU Socket.

[Peter]

Hi Derek,

DMA would need the CPU socket. Anyway the work effort to design a DMA controller for the Q60 would be much bigger, and the result much less helpful than a different bus system for peripherals.

Also, we see under Linux what could be done by improving SMSQ/E, instead of the hardware. E.g. the SMSQ/E harddisk driver does not even support interrupts, which would be possible, if smoother background behaviour is desired. (And better background behaviour would be the only gain from DMA, it can not improve peripheral datarate on the Q60.)

Peter

[Peter]

After seven years, I changed my personal vote from "Original Q60 with video updates" to "New 68060 system design"

[Dave]

When a successor to the Q60 appears and the Q60 is a previous generation product three times over, maybe you'll consider releasing the files and equations as open source?

[Dave]

E.g. the SMSQ/E harddisk driver does not even support interrupts, which would be possible, if smoother background behaviour is desired. (And better background behaviour would be the only gain from DMA, it can not improve peripheral datarate on the Q60.)

Even the most rudimentary DMA implementation would be a nightmare to plan out. Flip side, it could really smooth out some of the user interface stutter we've all come to know and..... love?

[Pr0f]

I favour a new design - with perhaps a blitter and or interleaved access to video frame buffer.

As resolutions get higher, and colour depth increases - the amount of data that has to be moved about is much larger - and this is what slows down the Q68 when chosing the higher video modes. In this case a bespoke DMA solution just for the video memory or main memory to video memory may be a simple ask given it has more specific requirements than a general purpose DMA channel.