Thanks for the explanation Anton.  This almost also points to David's gcc
setup being the problem, so unless he can prove there's a difference on a
clean install, I will assume the flag is not needed.

DaR

> -----Original Message-----
> From: Anton Ertl [mailto:[EMAIL PROTECTED]]
> Sent: Tuesday, September 10, 2002 10:46 AM
> To: [EMAIL PROTECTED]
> Subject: Re: [gforth] OS X 10.2 builds and benchmarks
>
>
> Dennis Ruffer wrote:
> > What does the -fno-gcse flag turn off anyway?
>
> Global common subexpression elimination.  This optimization moves code
> around (typically out of loops), but in gcc >2.95 it is confused by
> labels-as-values (what Gforth uses for threaded code).  It then moves
> code from somewhere into all primitives.
>
> This should be independent of MacOS X and G3/G4 (actually it also
> happens on Linux/Intel), and just depend on gcc version.
>
> - anton
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to