At 09.53 2011.10.25, you wrote:
>I think I'm losing confidence in Microchip's compiler expertise...
>
>What's the point again of defining the SFR addresses in the linker =20
>instead of in the C include files?
>It seems to prevent a significant slew of "obvious" optimizations, and =20
>there is still a huge per-chip source include as well, so it's not =20
>really moving any chip dependencies to link time...
>
>     digitalWriteFast(3,1);  // compiles to LATxSET =3D b
>9d00135c:       24020001        li      v0,1
>9d001360:       3c03bf88        lui     v1,0xbf88
>9d001364:       ac6260e8        sw      v0,24808(v1)
>     digitalWriteFast(3,0);  // compiles to LATxCLR =3D b
>9d001368:       3c03bf88        lui     v1,0xbf88
>9d00136c:       ac6260e4        sw      v0,24804(v1)
>
>Love that double load of identical constants!
>
>Is this the sort of thing that Microchip's "global optimization" fixes?

I hope so. Oh, and that was digitalWriteFast, I have seen more horrific
output from the compiler. :D

This is an example (but there could be more and better) that asm is not
finished on the PIC32, where actually it's even easier than in dsPIC or
PIC.

I prefer to code in asm, if I am in a hurry I do write a function in C,
develop, debug.. when I'm happy with it, I rewrite it in asm, maybe even
taking a look at the disassembly of it.


>Grr.
>BillW
--=20
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.