Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: ARM version of Chess Tiger for Palm?

Author: Christophe Theron

Date: 14:19:51 07/04/03

Go up one level in this thread


On July 03, 2003 at 22:36:55, Rodney Topor wrote:

>On July 02, 2003 at 22:09:53, Christophe Theron wrote:
>
>>On July 02, 2003 at 16:25:47, Michael Conway wrote:
>>
>>>To take advantage of the new ARM Palms running OS 5, many software developers
>>>have programmed native ARM code (ARMlets) to replace their calculation-intensive
>>>Motorolla 68k (OS 4) functions.  As this circumvents the Palm 68k interpreter
>>>(PACE) built into OS 5, performance increases in ARM processor Palms have been
>>>enormous.
>>>
>>>Are there plans to upgrade Chess Tiger for Palm in this manner?  Would it be
>>>very difficult to do?
>>
>>Yes, and I'm working on it.
>>
>>Yes, it is difficult to do. ARMlets are toys, not real programs. For example you
>>cannot have global variables in ARMlets.
>
>I don't understand this.  I thought ARM was a more modern, more flexible
>architecture than M68k.  For example, it has no 32KB jump limit.  Why should it
>be harder to program for than the older architecture?
>
>Rodney Topor



Because at this time the tools provided by PalmSource and Metrowerks to compile
C programs to native ARM code are in their infancy.

They have huge and almost absurd limitations that make them useless for any
serious work. For example no global variables (or if you want globals you need
to mess with the CPU's registers!).

That's why PalmOS 5 is said to support "ARMlets", not "real ARM programs".

Support for complete programs compiled from C in native ARM code is said to be
the goal of PalmOS 6.

Actually the tools are going to be released very soon, and complete ARM programs
will also work in PalmOS 5.

This is very similar to what happened when Intel released the 386: it had
tremendous possibilities, but it took several years for the operating systems of
that time (DOS for example) to start taking advantage of its possibilities.

It took several years to create a fully multitasking operating system with
preemptive multitasking and virtual memory on the x86 architecture, something
the 386 had been designed for (the 286 was supposed to allow it but did not
offer enough features actually).



    Christophe



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.