Re: [KOS] optimization oddity

[Richard M. <ric...@cr...>]

On Sunday 14 March 2004 23:28, Dan Potter wrote:
> On Mar 15, Jim Ursetto wrote:
> > I apologize if this stuff is obvious, esoteric, and/or offtopic. :)
>
> I find it really interesting anyway. :) It's good to know (well, in a
> "better know now than later" sort of way) that GCC isn't as good as I
> had thought it was originally. I wonder too how much of these missed
> optimizations are due to SH4 vs other arches. It seems like some of them
> have got to be the responsibility of a higher-up piece of code.

GCC does the same thing with my SSE wrapper classes -- in certain 
situations it insists on writing an object out to memory that could spend 
its entire life in a single SSE register.  On the other hand, I've seen it 
turn high-level expressions involving multiple objects and potential 
temporaries into 3 or 4 vector instructions -- which is very satisfying to 
see, but makes some of the missed optimizations even more painful.