[Freemint-list] Three CPU targets for *everything*

Miro Kropáček miro.kropacek at gmail.com
Mon Feb 13 01:19:48 MSK 2017


>
> You miss the point. Wouldn't the correct sequence of events first be to
> suggest the change, discuss it and then implement it?
>
Maybe yes but you know how it is in this list -- some things get ignored no
matter what so I could have ended up waiting, doing nothing and after some
time I'd give up.

What about the Milan040 and Hades040? The AB does not need special
> attention in the kernel, the generic 040 build has always worked. If I
> understand the 040/060 merge correctly, shouldn't all of these machines
> - AB040 included - be able to run the 040/060 kernel?
>
Again, Milan (and Hades, as it's the same build) are still separate
kernels. At one point, Milan and Aranym will be merged into min4060.prg,
yes. But that's not the most important thing right now.

So basically this change is to improve mintloader? Can I still put
> mint.prg in the autofolder? After all, my Milan only runs exactly one
> kernel.
>
Yes you can but after my work on autobuilds you'll see why you will perhaps
prefer do it in the new way. :)

My experience is that optimizing for the 040 or 060 does not result in
> measurable increase in performance compared to 020-060. The quality of
> the code is what matters.
>
Exactly my point. 10 years ago I was overwhelmed by the CT60, I wanted
everything "made" for it but in the end, it doesn't really matter.

So in a nutshell, my change forces only three cpu targets (000,020-60,5475)
for every component (except the kernel...), reduces the number of kernels,
makes xaloader/usbloader load xaaes.km/usb.km again (but I kept it backward
compatible, so yes, your xaaes040.km will load too) and makes mintloader
SYSDIR aware.

-- 
MiKRO / Mystic Bytes
http://mikro.atari.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.atariforge.org/pipermail/freemint-list/attachments/20170213/9d1832f9/attachment.html 


More information about the Freemint-list mailing list