Re: [Mingw-users] GCC 4.3.0 20080502 (alpha) Released - Please helpus test it!

[Brian Dessent <brian@xxxxxxxxxxx>]

Stéphane Larouche wrote:

> (no flag): (331+151)kB, 142s
> -O1: (272+138)kB, 87s
> -O2: (268+138)kB, 84s
> -O3: (269+139)kB, 82s
> -ffast-math: (326+150)kB, 115s

Specifying -ffast-math without -O doesn't make any sense because in the
absence of -O the compiler performs no optimization.  So all the above
really shows is that -ffast-math improves the unoptimized case from 142s
down to 115s, but you can't meaningfully compare "-ffast-math" to "-O2"
as that's apples and oranges; you should compare "-O2" to "-O2
-ffast-math".

Also, you need to specify a relevant -march (with 4.3 you can use
-march=native for autodetection of the host's capabilities) otherwise
only the default 386 instruction set is used which is quite limiting to
mathematical performance.

Brian

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference 
Don't miss this year's exciting event. There's still time to save $100. 
Use priority code J8TL2D2. 
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
MinGW-users mailing list
MinGW-users@xxxxxxxxxxxxxxxxxxxxx

You may change your MinGW Account Options or unsubscribe at:
https://lists.sourceforge.net/lists/listinfo/mingw-users