Page 5 of 10
Re: New Q60 Boards
Posted: Mon Jan 27, 2025 7:26 am
by Derek_Stewart
HI Peter,
Development of QDOS Classic was only an idea.
I build a Q60 in 2002, because I liked Assembly Language, and the 68060 CPU looks a nice processor to programme, this still holds true now.
Re: New Q60 Boards
Posted: Mon Jan 27, 2025 7:34 am
by Derek_Stewart
HI,
I read in this message thread, the everyone wants Q60, great, this can be done, with the last design.
But as in 2002, when with a friend who lived locally, we devised a production process for manufacturer of the Q60 in 2 forms:
1. Cased Q60 in available cases at the time of manufacturer only Beige PC Midi Tower Cases were available.
2. Q60 Board option: with PCB, CPU, I/O Card, RAM.
This was the only options because, if the Q60 board was supplied only without, CPU, I/O Card, RAM then I can not guarantee that the parts fitted will work.
Part of the Q60 manufacture was the testing of the hardware and soak testing of the operating system under load. Linux was used for this as Linux could access all the 128Mb ram.
Today the same thing still applies.
Re: New Q60 Boards
Posted: Mon Jan 27, 2025 7:56 pm
by ql_freak
Sorry for quoting myself, but I have just copied the C68 runtime and documentation archives from Dilwyn´s page to my QPC2 installation and am just reading changes_doc.
ql_freak wrote: Sun Jan 26, 2025 10:15 pm
...(are there any programs, which can use the FPU, perhaps C68 compiled code, I've heard later versions had support for the FPU, at least for the 68040, the 060 lacks some features of the 040)...
YES! I have remembered correctly :-D
Code: Select all
C68 COMPILATION SYSTEM Change Log
Release 4.20a (18th November 1995)
...
7. The floating point support routines will now use hardware
floating point for multiply and divide if it is present.
This is transparent in normal use beyond the speed up when
such hardware is present. The TECHREF.DOC manual is
updated with details of the underlying implementation.
Note that this implementation is fully compatible with the
package by Simon Goodwin that allows floating point
hardware to be used from within SuperBasic programs.
...
And even better, the "FP Save" routine from George Gwilt, which Peter has suggested, has been added even in an earlier version :-D AND in
Version 4.21:
Code: Select all
3. Support for generation of inline hardware floating point
instructions added, although use at this release is still
experimental (due to GWASS being experimental).
Version 4.22:
Code: Select all
5. The IEEE support routines upgraded to work with George
Gwilt's FPSAVE floating point support routines.
Re: New Q60 Boards
Posted: Mon Jan 27, 2025 9:02 pm
by Derek_Stewart
Hi,
There has been cases where C programmes are cross compiled with QDOS-GCC, rather than C68. That with compiled optimisation, a 30% speed increase was achived.
I used supply George Gwilt's FPU code for Q60 with FPU fitted.
The same FPU code works with a QXL fitted with a 68040 with FPU.
However, only programme specific software tskes advantage of the 68060 FPU, as normal SMSQ/E operation does use the FPU in the CPU.
Or maybe I did not know how to use it.
Re: New Q60 Boards
Posted: Tue Jan 28, 2025 4:38 pm
by ansar
Hi Derek,
Although I am still waiting for Q68, can you supply me a Q60, in order to test my new 64 bit compiler and interpreter? Just pm cost etc.
Regards,
ansar
Re: New Q60 Boards
Posted: Tue Jan 28, 2025 6:08 pm
by Derek_Stewart
ansar wrote: Tue Jan 28, 2025 4:38 pm
Hi Derek,
Although I am still waiting for Q68, can you supply me a Q60, in order to test my new 64 bit compiler and interpreter? Just pm cost etc.
Regards,
ansar
I have the Q68 boards and I am in the process of constructing the Q68.
I need to get the Q68 cases made along with QL Network circuitry.
Q60 not in production at present.
Re: New Q60 Boards
Posted: Sat Feb 01, 2025 11:20 am
by Peter
Derek_Stewart wrote: Mon Jan 27, 2025 9:02 pm
There has been cases where C programmes are cross compiled with QDOS-GCC, rather than C68. That with compiled optimisation, a 30% speed increase was achived.
QDOS-GCC can compile code with FPU support, but the assembler then fails. Apparently, using -m68881 (or any other commandline option which selects an FPU architecture) generates an unrecognized "gwass" commandline option.
It also seems impossible to assemble QDOS-GCC >=68020 code, even without FPU code generation.
Re: New Q60 Boards
Posted: Sat Feb 01, 2025 4:23 pm
by Derek_Stewart
Hi Peter
I only repeat what others have quoted, their claims seems to relate to C compiled programmes with QDOS-GCC.
Whether this correct, I can not verify.
Maybe I should only Superbasic and Assembly Language.
I try to understand all this programming, but I am just an enthusiastic user that has had limited success building hardware.
Re: New Q60 Boards
Posted: Sat Feb 01, 2025 4:27 pm
by XorA
It also seems impossible to assemble QDOS-GCC >=68020 code, even without FPU code generation.
When I looked at the support for this ages ago it looked to be incomplete code.
Re: New Q60 Boards
Posted: Sat Feb 01, 2025 4:28 pm
by tofro
Peter wrote: Sat Feb 01, 2025 11:20 am
Derek_Stewart wrote: Mon Jan 27, 2025 9:02 pm
There has been cases where C programmes are cross compiled with QDOS-GCC, rather than C68. That with compiled optimisation, a 30% speed increase was achived.
QDOS-GCC can compile code with FPU support, but the assembler then fails. Apparently, using -m68881 (or any other commandline option which selects an FPU architecture) generates an unrecognized "gwass" commandline option.
It also seems impossible to assemble QDOS-GCC >=68020 code, even without FPU code generation.
Actually, I don't think this is a problem (rather some sort of advantage or case of forced self-restraint): I don't think we want the world of QL software which is small enough already, further sub-divided in programs that can only run on certain CPUs and FPUs.