Amiga's New SDK: A First Glance

Mike Bouma writes: "Recently it began raining news coverages about Amiga`s new OS in the mainstream press like CNN`s Digital Jam, The New York Times and Gamersdepot. The first impressions of the new SDK have been very positive. Lars Thomas Denstad has written a small article about his experiences with the new SDK so far."

Comments

TAO Similarities

[naden]

I remember seeing Taos in a magazine from years back, June 1994 to be exact. This whole text, is some mine, some quoted from Chris and mainly excerpts from the magazine. The magazine is called Edge, a UK videogaming magazine. I noticed the same raytraced picture was done using Taos, and there are numerous mentions of Taos systems.

"This month Edge got a glimpse of the future, thanks to a demonstration of the Taos OS. In a nutshell, Taos enables programs coded on any machine to run on any other machine - in parallel, across any available processors in the system."

"Taos ie even more amazing when you realise that it is the product of one man's efforts, coding for his own benefit, rather than cumulative efforts of some corporate programming team"

The men denoted as the "Three Wise Men" were Chris Hinsley (inventor of Taos), Tim Moore and Francis Charig - directors of Taos Systems.

This operating system was targeted at the console industry, where Chris had the idea of producing an operating system that would manage games and aid code portability. The first step was a macro set which Chris constructed for the assemblers of all the platformers he was writing on. Rather than write in the native assembler language, he wrote in the macro language he defined; he then devised a translator which would take the binary equivalent of that macro set and translate it, on the fly, into the instructions of a particular machine.

The Taos kernel which is typically around 16K, is loaded into the processor at boot time. That kernel is specific to that particular processor. If the kernel finds it needs a translator tool, it brings in the translator as well. The application then gradually builds itself in memory: as a processor in the network needs to call functions it brings them in and binds the application.

All programs are compiled or assembled into VP code and are kept in this form on disk. The VP code is translated into the native code of the processor on which it is run only when it is needed. The translation occurs as the VP code is loaded from the disk, across the network, and into the memory of the target processor. (Note this implies distribued computing.)

However, this doesnt slow the system down: most processors can actually translate VP code into native code faster than VP code can be loaded from disk and sent across the network. And VP code is often more compact than native code; it takes up less disk space and is loaded faster.

For instance, if you had a console that booted from CDROM, a CD would be pressed so that the first thing it did would be to load up the appropriate version of Taos, place it in memory and set it running. Then it would load the game code, which would run the operating system. The operating system would then load the specific tools required for that game and execution of the game would begin.

Access to custom chips is taken care of automatically by Taos using a method called dynamic binding: individual chips are supported by VP libraries, which allow for a tool for that particular processor to be accessed by the system; the tools are bound in during runtime as they are needed. Dynamic binding also enables several processes to share tools, which is very memory efficient.

"This 'virtual processor' works like a 16-bit register RISC microprocessor, explains Chris" "But it isnt an emulated technology; it actually translates into native code and it's the native code which runs and al the translations take place during the load time of the

It is quite an effort to think of their feature list so many year ago ..

Hardware Independance / Load balancing / Heterogenous processing / Dynamic Binding / Multi-threading / Parallelism as well as support for MPEG / Postscript and real time polygon rendering.

In my opinion this guy is a genius, that relegates Linus to quite a mediocre status. I mean this OS is good by todays standard. I mean Linux is even now not brilliant at parallel processing and this OS can not only parallel process tasks but delegate them to entirely different chips.

To put it into perspective, at the start of 1994 only 7 million people in US had computers with CDROM drives.

I think he deserves a universal sympathy award for not patenting some of these concepts. Had these been patented you wonder whether technologies like Java and companies like Transmeta would still exist.

I hope Amiga does well .. they were always a favourite.

Naden
(member of www.it-guys.com)