68060 based QL system?

Nagging hardware related question? Post here!

Do you want a 68060 based QL system?

No, better improve speed of FPGA based 68K
12
44%
Original Q60 with video updates
5
19%
New 68060 system design
10
37%
 
Total votes: 27
User avatar
Peter
Super Gold Card
Posts: 637
Joined: Sat Jan 22, 2011 8:47 am

68060 based QL system?

Postby Peter » Fri Apr 06, 2018 9:40 am

Any native 68K hardware including 68060 has become slower than emulation on PCs. I wonder how people see 68060 systems today.


User avatar
tofro
QL Wafer Drive
Posts: 1368
Joined: Sun Feb 13, 2011 10:53 pm
Location: SW Germany

Re: 68060 based QL system?

Postby tofro » Fri Apr 06, 2018 11:19 am

How realistic is a QL system that would provide a sort of QL base board for Amiga/Atari ST-based Vampire/Apollo accelerators (http://www.apollo-accelerators.com)? They claim to be 3 to for times faster than a 68060. The board is a memory/(FPGA-)CPU/HDMI/SD board for Amigas and STs. A bit hard to come by, but faster than anything 68k-compatible on the market.

Tobias
Last edited by tofro on Fri Apr 06, 2018 11:22 am, edited 2 times in total.


martyn_hill
Gold Card
Posts: 380
Joined: Sat Oct 25, 2014 9:53 am

Re: 68060 based QL system?

Postby martyn_hill » Fri Apr 06, 2018 11:21 am

Hi Peter

Thanks for the initiative...

Given my limited knowledge at this level, I think I'd need to have it explained in a little more detail as to the relative benefits/limitations of each option before I could form and share an informed opinion.

For example:
a) What potential clock-speeds can an FPGA vs native 680x0 reach?
b) Would the choice limit the potential peripheral/device options (e.g. addition of HW support for USB. Also, what 'standard' video output options are available under each option?)
c) How much new SW development would the differing approaches require - this is and always has been a significant factor in success of HW projects for us.
d) Similarly, which option offers the most compatibility with existing SW or HW?
e) (Something that interests me, perhaps more than it would some other users) Which option maintains the option to develop simple HW add-ons most readily (e.g. allowing for an 8-bit expansion interface at 'legacy' bus-speeds.)
f) Relative cost and time for development...

When I think about my own drivers for wanting a new QDOS/SMSQE compatible HW platform, the ability to connect it to the rest of the (peripheral) world using standard connections/protocols is probably more significant than raw performance, for example. Plus the opportunity to develop my own amateur projects in both SW and HW - the latter being facilitated by having standard I/O options like SPI or I2C and USB (were SW support ever to become available.)

All that said, I think we have some truly amazing platforms available already, including the Q68, which I haven't even begun to fully explore yet.

M.


User avatar
Peter
Super Gold Card
Posts: 637
Joined: Sat Jan 22, 2011 8:47 am

Re: 68060 based QL system?

Postby Peter » Fri Apr 06, 2018 12:38 pm

tofro wrote:How realistic is a QL system that would provide a sort of QL base board for Amiga/Atari ST-based Vampire/Apollo accelerators (http://www.apollo-accelerators.com)? They claim to be 3 to for times faster than a 68060.

To electrically build such a base board seems relatively realistic (within the contraints we always have in the QL scene) but:
  • You easily pay over 500 € to actually get a fast Vampire accelerator
  • If you look at anything that was available, the factor was 1.2, not 3, versus Amiga 68060 accelerators. I doubt it would be any faster than a QL-specific 68060 board at all.
  • Like the open-source 68K FPGA cores, their closed-source core will probably have QL-specific bugs, which don't surface on Amiga/Atari. No way we could fix them.
  • The Vampire plugs into a 68000 socket - that would make video circuitry with SMSQ/E driver support very slow.


User avatar
tofro
QL Wafer Drive
Posts: 1368
Joined: Sun Feb 13, 2011 10:53 pm
Location: SW Germany

Re: 68060 based QL system?

Postby tofro » Fri Apr 06, 2018 1:23 pm

Peter wrote:
  • You easily pay over 500 € to actually get a fast Vampire accelerator

Mine (that I have in the Amiga 600) cost €285 to my desk in Germany
Peter wrote:
  • If you look at anything that was available, the factor was 1.2, not 3, versus Amiga 68060 accelerators. I doubt it would be any faster than a QL-specific 68060 board at all.

  • Can't really refute you on that - Other than that it's blazingly fast, as I have no 68060 to compare against.
    Peter wrote:
  • Like the open-source 68K FPGA cores, their closed-source core will probably have QL-specific bugs, which don't surface on Amiga/Atari. No way we could fix them.

  • That's true for just about anything - On the other hand, there is a professional development team
    Peter wrote:
  • The Vampire plugs into a 68000 socket - that would make video circuitry with SMSQ/E driver support very slow.

  • The Vampire comes with HDMI video circuitry - Basically, the only thing missing to a full-blown computer is the keyboard and a case.

    Which reminds me I must try the QL-Amiga with the Vampire.

    Tobias


    User avatar
    Peter
    Super Gold Card
    Posts: 637
    Joined: Sat Jan 22, 2011 8:47 am

    Re: 68060 based QL system?

    Postby Peter » Fri Apr 06, 2018 1:41 pm

    martyn_hill wrote:a) What potential clock-speeds can an FPGA vs native 680x0 reach?

    At the moment, over 100 MHz is a realistic figure for a new design. Performance will of course depend on much more than clock-speed (architecture, cache, ...).
    martyn_hill wrote:b) Would the choice limit the potential peripheral/device options (e.g. addition of HW support for USB. Also, what 'standard' video output options are available under each option?)

    The original Q60 would keep it's limitations, except different video timings for partial flatscreen support (VGA or by VGA-HDMI converter).
    The other two choices do not limit USB/video considerations (but one should be as realistic as always for QL hardware.)
    martyn_hill wrote:c) How much new SW development would the differing approaches require - this is and always has been a significant factor in success of HW projects for us.

    For Q60 no SW development required at all. For the other two options, it depends more on additional peripherals than on the core system, which could be held close to Q60/Q68 to minimize SW development.
    martyn_hill wrote:d) Similarly, which option offers the most compatibility with existing SW or HW?

    The Q60 has issues with Qliberated software which are probably related to the 68060. Same has to be expected for a new 68060 design. FPGA allows to stay closer to original QL, if desired.
    martyn_hill wrote:e) (Something that interests me, perhaps more than it would some other users) Which option maintains the option to develop simple HW add-ons most readily (e.g. allowing for an 8-bit expansion interface at 'legacy' bus-speeds.)

    Independent from the poll choice, rather an individual design decision. Q60 has a legacy expansion bus (ISA standard). Q68 has an even simpler legacy expansion bus. So it is doable and exists.
    martyn_hill wrote:f) Relative cost and time for development...

    Time for hardware development: Q60 = Small (but challenges purchasing obsolete parts and manufacture) Other two poll choices = High
    Costs: 68060-based always higher than FPGA-based


    User avatar
    Peter
    Super Gold Card
    Posts: 637
    Joined: Sat Jan 22, 2011 8:47 am

    Re: 68060 based QL system?

    Postby Peter » Fri Apr 06, 2018 2:09 pm

    tofro wrote:
    Peter wrote:[*]The Vampire plugs into a 68000 socket - that would make video circuitry with SMSQ/E driver support very slow.[/list]

    The Vampire comes with HDMI video circuitry

    I know - but of no use since SMSQ/E does not support that video circuitry. (Reminds me of Jochen Merz' attempts with SMSQ/E on the Milan.)
    tofro wrote: Basically, the only thing missing to a full-blown computer is the keyboard and a case.

    In detail, it will probably be not quite as simple as that. But with the current non-availability and prices, it's not worth a closer look for me.
    tofro wrote:Which reminds me I must try the QL-Amiga with the Vampire.

    Do you mean QDOS Classic from Marc Swift?


    martyn_hill
    Gold Card
    Posts: 380
    Joined: Sat Oct 25, 2014 9:53 am

    Re: 68060 based QL system?

    Postby martyn_hill » Fri Apr 06, 2018 2:15 pm

    Thanks Peter!

    Assuming then that the FPGA option effectively represents an upgraded Q68, I'll vote for that option :-)


    User avatar
    Peter
    Super Gold Card
    Posts: 637
    Joined: Sat Jan 22, 2011 8:47 am

    Re: 68060 based QL system?

    Postby Peter » Fri Apr 06, 2018 3:22 pm

    I should add that open-source 68K FPGA cores, as used on the Q68, are significantly slower at same clockspeed than the commercial "Apollo" core mentioned by Tobias. So while FPGA technology in principle already allows to outperform the 68060, this is beyond reach for hobbyist designs at the moment. However, some improvements are realistic, like cache.


    Derek_Stewart
    QL Wafer Drive
    Posts: 1021
    Joined: Mon Dec 20, 2010 11:40 am
    Location: Runcorn, Cheshire, UK

    Re: 68060 based QL system?

    Postby Derek_Stewart » Fri Apr 06, 2018 4:12 pm

    Hi,

    I have a Vampire 2 on a Amiga A600, it plugs into the 68020 CPU socket, quite easy to fit. Worked first time. But I have not had much time to play with it.

    If tge Supergold card had its CPU in a socket, the A600 Vampire 2 may fit, but it might not work with the QL.

    There is an Amiga A500 Vampire 2 board that plugs into the A500 68000 CPU socket. Not comptiable with the 68008.

    I would like to have a modern Q60, or a Q60 tthat can handle TFT monitors.


    Regards,

    Derek

    Return to “Hardware”

    Who is online

    Users browsing this forum: No registered users and 12 guests