Yahoooo.
I have done it.
I have written all the code : ram vectors and interrupt handlers in assembler language and in c language stored in rom and it is loaded in ram as fastcode.
After that i do the copying of fastocde, data and bss. stack filling. in startup.s . Then i do in assembler the remapping of the ram to adress 0x0000 0000. just before i jump into my main. The increase in speed is very noticable. Now i see the raw speed of the SAM7S for the first time.
When i have put in the comments, i will place the example code on sourceforge and i will post here on the date of placement.
Now i have standard 9 interrupts running. 1 FIQ for system interrupt through the PIT and a maximum of 8 IRQ interrupts for devices as standard but can be expanded up to 32. I use the AIC. I use separate handlers for each interrupt for now but i will have to add the saving of the registers as well. I use the irq stack and the system stack. I have the FIq stack, irq stack, supervisor stack and the user /system stack.
Later on i want to put the SWI handler as well inside the fastcode.
Time to sleep to load up for a programming challenge tomorrow.