[Bug middle-end/31249] pseudo-optimzation with sincos/cexpi

["dominiq at lps dot ens dot fr" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #8 from dominiq at lps dot ens dot fr  2007-03-20 13:57 -------
> The only reason why cexp is slow on PPC darwin is because the ABI is stupid. 
> Complex float arguments are passed via the GPR and returned also the same way
> instead of via the FPRs.  So you will get a transfer of registers.  This is
> also true of PPC64 darwin, why they made the same mistake twice I have no idea,
> guess they did not expect people to use complex that much.

Is this also true for complex double on 32 bit architectures (i.e., 4 GPRs)
or do you mean the GPR is used to pass a pointer?

> ... .  The number of cycles used up by this issue
> can add up with both sides of the function having this hazard.

You are comforting my prejudice against using procedures in critical loops.

Now if you cannot convince darwin people to fix the problem, I cannot how I
could.  My short term interests are:

(1) to understand how this reverse optimization is triggered.

(2) to know what are the non-Linux platform that are affected beside
Darwin.

(3) to get a work around less hackish that what I did in my first example.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31249