Page 10 of 26

Re: Q_Liberator malaise

Posted: Tue Jan 14, 2025 10:12 pm
by Peter
Since all Q40 and Q60 have an on-chip MMU, another possibility might be to set up special memory regions for self-modifying programs and mark them as writethrough. The MMU allows much more granularity in specifying regions than the TTRs, so in principle that should probably work. The OS could provide EXEC_SM or something to start self-modifying programs.That way, only the QLiberated stuff would slow down, but not the whole machine.

But programming the MMU is not trivial and - if I have to do it myself - would be another distraction from the already huge hardware challenge.
Also I was considering usage of the 68EC060 for future machines, which are cheaper and easier to get, but do not contain an MMU.

Re: Q_Liberator malaise

Posted: Tue Jan 14, 2025 10:15 pm
by Peter
tofro wrote: Tue Jan 14, 2025 9:59 pm Easily done, and if they don't, nothing is lost, but if they do, we can use the code as-is without investing a lot of time.
That "easily done" still requires someone more capable in assembler programming than myself. ;)

Re: Q_Liberator malaise

Posted: Wed Jan 15, 2025 1:35 pm
by RalfR
BTW: During examining the SMSQ/E sources, I have found a few *_asm, where special things for Q_Lib were done (as far as they are documented, all *_asm):

sbsext/ext: CALLsq
sbsext/ext: file
sbsext/tk2: file
smsq/sbas: base
smsq/sbas: init
smsq/sbas: putp
smsq/sbas: aldat
smsq/recent: things

Re: Q_Liberator malaise

Posted: Wed Jan 15, 2025 4:07 pm
by Derek_Stewart
It seems to me that there are the SMSQ/E and QLIB_RUN source code to analyse.

I am rapidly, getting out of my comfort zone with assembler source code. I need to do more studying.

Re: Q_Liberator malaise

Posted: Thu Jan 16, 2025 7:58 pm
by Peter
Hi Derek,
Derek_Stewart wrote: Sat Jan 11, 2025 3:23 pm The QXL has a 68EC040 40/50 Mhz CPU but the 68040 halfves the clock speed to 20/25 Mhz, runs the same version of version of SMSQ/E.
I never had many problems relating to Qliberator compiled files or SBASIC files.
Coming back to this again: If you have a QXL at hand, could you use the cachemode tool to check if it basically boots/works with copyback?
Not very likely, but if it does, that would give us one more test target for QLib - and a faster QXL of course.

All the best
Peter

Re: Q_Liberator malaise

Posted: Fri Jan 17, 2025 7:08 am
by Derek_Stewart
Peter wrote: Thu Jan 16, 2025 7:58 pm Hi Derek,
Derek_Stewart wrote: Sat Jan 11, 2025 3:23 pm The QXL has a 68EC040 40/50 Mhz CPU but the 68040 halfves the clock speed to 20/25 Mhz, runs the same version of version of SMSQ/E.
I never had many problems relating to Qliberator compiled files or SBASIC files.
Coming back to this again: If you have a QXL at hand, could you use the cachemode tool to check if it basically boots/works with copyback?
Not very likely, but if it does, that would give us one more test target for QLib - and a faster QXL of course.

All the best
Peter
HI Peter,

I have 1 QXL waiting on some RAM to be delivered and 2 blank PCBs that need all the parts, which is one part: QXLDEC : ST 20AS25HB1 "GENERIC ARRAY LOGIC", is proving challenging to find, and the QXL Altera Glue Chip, with the same issue as the Super Gold Card.

When the QXL RAM has arrived, I will test the cachemode software out.

There is a 68040 to 68060 adaptor board for 68040 CPU systems from the Amiga World, I was always going to build one o see if the QXL could be run the CPU at full speed, as the 68040 halves the QXL Clock speed of 40/50Mhz to 20/25Mhz. Where as the fitting of the 68060 runs the CPU at full board clock speed 40/50Mhz.

could produce more QXL boards, but I do not think there is a demand. Anyway, the Q68 amd QIMSI Gol is better.

Re: Q_Liberator malaise

Posted: Fri Jan 17, 2025 9:05 am
by XorA
could produce more QXL boards, but I do not think there is a demand.
ISA slots are getting rare and expensive!

Re: Q_Liberator malaise

Posted: Fri Jan 17, 2025 10:07 am
by Peter
XorA wrote: Fri Jan 17, 2025 9:05 am ISA slots are getting rare and expensive!
Also the ISA cards, so if I want to make a new Q60 affordable, I need to re-implement at very least IDE and SER in PLD/FPGA for a new Q60. For ethernet, it is even worse, as the obsolete RTL8019 chip with it's ancient 5V levels must be added to the mainboard or a newly designed I/O card. Unfortunately, if I want to stay compatible with the original Linux kernel, I can not just switch to more modern I/O. So a new Q60 without ISA is way more work than the original machine... a QLib solution would at least provide more motivation.

Re: Q_Liberator malaise

Posted: Fri Jan 17, 2025 10:39 am
by Derek_Stewart
Hi

The ISA expansion slot on the Q60, always caused confusion with the PCB ISA technology, I had people moaning PCI would be better than ISA, but one actually used it to the full potential.

Only having, Multi I/O, Network Cards for use with the ISA style expansion.

No else produced anything that connects the Q60 Expansion connector.
XorA wrote: Fri Jan 17, 2025 9:05 am ISA slots are getting rare and expensive!
With regards to the QXL, I had 7 working boards and 2 blank PCBs, I only have 1 working board requiring more ram and the blank PCBs, seems people still want the QXL even PC ISA is obsolete and not readily available.

The remaining QXL board is reserved for a customer.

I have a few ISA based motherboards and an Amstrad 486 PC that can run the QXL.

So the QXL is still wanted.

Re: Q_Liberator malaise

Posted: Fri Jan 17, 2025 5:21 pm
by Pr0f
What about leveraging something like the ECB bus - there are a good number of cards and backplanes for that?

https://www.retrobrewcomputers.org/doku ... :ecb:start