Re: [oc] Would OR1K becomes next 8051 or 6502?

["Damjan Lampret" <lampret@opencores.org>]

>
> 1. To make OR1K more debuggable with software and hardware,
>     I think gdb is also important!!
>
Sure. Perhaps it would be also good to bundle or1ksim with gdb and then you
can use gdb front-ends to debug code on a simulator. Quite useful for
porting operating system.

>   Is it good or bad? I think it's good!!! Then all embedded chips may
>    be OR rather than ARM or MIPS. Then maybe one day ARM or MIPS
>    will perhaps give up their ISA patents.
>
Possibly when it will be too late for them. ;-)

> 3. If I want to minimize area without a MMU, the memory configuration
>     should be what? After all, MMU is big for embedded usage and
Either you have MMU or not. If not then I guess it makes sense that virt
addresses == phy addresses. If you have a MMU then you select how much TLBs
you need (current design supports either no mmu or 2x16 TLBs).

>     Virtual address=physical address may be too simple.
>    But I think it is still feasible!
>
> 4. OR16(16-bit) ISA could be made more stand-alone and
>    we could make it embedded into OR1K with one OR16 decompressor
>     block.
>
If we want a complete 16-bit insn subclass then it first should be defined.
Currently only a couple of very oftenly used insns are 16-bit. So code is
much smaller (but this is not yet supported in assembler; assembler treats
them as 32 bit insns with lower 16 bits set to 0!). It would make perhaps
sense to make a complete 16 bit insn class. Anyone want to define it?

regards, Damjan