home .. forth .. misc mail list archive ..

Firewire/3D/embedded


A Firewire 3.2Gb/s optical (or double line 6.4Gbs) has PCI 64 type speeds 
surely in the future supporting just a firewire type interface instead of 
the rest would enable Misc access to all periphial devices planned foir it 
(just about everything if you include the wireless version including hard 
drives).  This would cut out the required inrterfacing for seperate IDE, 
floppy etc etc and proabably wouldn't add muck more coding than would be 
needed to support just the common devices.

>Using a standard OS Microsoft Arcade core, or something smaller >should be 
>able to write one game to suit PC and handheld.

Thats why I mentioned the new Wireless Application Protocol based on a basic 
sub sub etc set of WIN-CE,  I have been thinking that this + basic direct-X 
(and other WIN-CE game API's used in Sega Dream cast) would be all we need 
and be fulluy standardised with other systems.  Something that iTv could 
have done.

By the way what happened at ITV I lost a couple of posts and couldn't find 
mention of what happened but people were a bit glommy about it?


Now back to #D ;).  There is a huge gap in the market out there, huge, the 
colour game boy is a bit parsay, there is no proper 3D colour system out 
there.  So there is oppertunity to have say an ITV WAP machine plus with 3D 
game capability.  Nobody is touching this market (though we hear Nint has 
plans).  That and the embedded markets are the few market gaps left.  The 
embedded market is an easy entry target, the game market is a hard entry 
target.  Companies spend 100's of millions just to launch a machine.  But in 
the handheld market low tech reings and developing a better machine is 
easier, and contract (network) manufactures are available.  The only other 
problem is a regular supply of game software, without this people wont buy 
(unless its like a Playstation 2 in the hand), by adopting an existing 
OS/emulator you can have a regular supply of masses of games.  Using a 
derivative of an existing OS (java windows CE) you could relaease for both 
the machine and the PC market at the same time, using an emulator or 
complete you allready have hundreds of titles to offer.  There is more to 
this than I can mention art the moment because I would like to do a plug in 
VM for internet browsers.

Now the voxel question, given a decent 1 cycle integer operations (doubling 
misc chip transistor count) another small amount of transistors would 
probably support the interger based routines required on Voxels, note I'm 
talking about 10's of thousands transistors to do Voxel 3D instead of the 
millions they need for polygons, sounds very misc to me.  So for 20-30+ 
thousand transistors total it might be possible to transform a 15 thousand 
transistor i21 into a high speed voxel machine.  But of course I'm just 
bouncing the voxel hardware idea for those wondering what to do for an 
i/p/f32.  The existing hardware would still be able to do some. I think the 
memory problems with voxels are over comable, given two memory banks (one 
for data and one for code) one PC-100-133 etc for data and the other s-ram 
etc for code execution, the random access hit of sram can be reduced, as 
well as large bandwith.  Also I think the memory requirements of voxels can 
be greatly reduced.

The PC maybe have the desktop and the internet, but the handheld is still up 
for grabs.

Thanks.

Wayne.


______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com