Welcome, Guest. Please login or register.

Author Topic: ColdFire Project?  (Read 7455 times)

Description:

0 Members and 1 Guest are viewing this topic.

Offline Plaz

Re: CofFire Project?
« on: September 27, 2007, 01:50:35 AM »
The coldfire mail list still has some activity as discussions continue about possible solutions. Basically there about about 6 instructions that behave differently on coldfire that can't be traped. I think it can be solved (Elbox must have done it) but it would take alot more work using a card much different than the current prototype.

Plaz
 

Offline Plaz

Re: CofFire Project?
« Reply #1 on: September 27, 2007, 02:09:09 AM »
Quote
Well, could a recompilation system do it? Using AROS you could make an OS which would, rather than JIT emulation, re-compile the apps. The coldfire is similar enough that it could work.


The trouble is those wayward instuctions also may exist in the OS not just the apps. So just like pokemon, you got to catch them all. :-P

Some say JIT is the only way and why bother, but I think a prescann-replace solution could work. It would be similar to the way vmware handles code on non-100% compatibly x86 families.

Plaz
 

Offline Plaz

Re: CofFire Project?
« Reply #2 on: September 27, 2007, 02:36:14 AM »
Quote
How would an OS compiled for the Coldfire not run on the Coldfire? Note, I said AROS, not AmigaOS, so you could compile it for the right cpu faily.


Oh, I misunderstood. Yes that would be different. You just have to edit the 68K source code before compiling replacing the offending code with coldfire acceptable code. There are instruction on the web about which instructions you need to replace and how in order to compile 68K code for coldfire.

Plaz
 

Offline Plaz

Re: CofFire Project?
« Reply #3 on: September 27, 2007, 03:06:03 AM »
Quote
There are already 400mhz v5 coldfire CPUs inside HP printers, oddly enough there is no information about it in the freescale site


Freescale will make custom coldfires to suit your needs. Maybe HP walked in with a bag of money and said... "we need one that does this", and freescale built them one to HP's specs. Too bad amigaland doesn't have the same kind of resources. We're stuck with normal wares off the shelf. ........ Heeeeyyyyy, what if DiscreetFX.......... nah.

Plaz
 

Offline Plaz

Re: CofFire Project?
« Reply #4 on: September 27, 2007, 01:19:11 PM »
Quote
I think the AROS code is in c or c++ so a simple recompile to Coldfire or 68k/Coldfire mixed code should do the trick without editing the source. That only would be needed if the source is in assembler.


Generally Coldfire and 68K code is equal. But some 68K instructions don't exist on coldfire. In other cases (and I'm mega simplifying here) InstructionA makes coldfire turn left, while the same instruction makes a 68020 turn right. To make clean coldfire builds, you'll need to edit legacy 68K code to replace these missing and misinterpeted instructions. Luckily the list of edits will probably be small.

Plaz

 

Offline Plaz

Re: CofFire Project?
« Reply #5 on: September 27, 2007, 01:23:10 PM »
Quote
So how come Elbox Dragon is 040 speeds then?


Some say they are probably burning much of their cycles running a JIT compiler to handle those instruction differences I mentioned between CF and 68K.

Plaz
 

Offline Plaz

Re: CofFire Project?
« Reply #6 on: September 27, 2007, 05:25:02 PM »
Quote
Any body here who want's to bring the AROS 68k port out of unmaintained, where it sits for long now?


I'm told there is some one who's started some work again. I've also stuck my nose in for a look.

Plaz