Dumping Gimple After Optimization

[Daniel Berlin]

On 4/5/07, Mostafa Shahto <mshahto@gmail.com> wrote:
>
> I've tried -fdump-tree-all before, and yes, I'm getting different types of
> files, including Gimple. But in the gimple file, there is no Copy
> Propagation Optimization or CSE. How can i get an optimized (all possible
> optimizations) gimple file or gimple-like file??
> I'm new to developing under Linux, so regarding you advice, I've run the
> following command (my C++ file is mm4x4.cpp):
> $gcc mm4x4.cpp -fdump-tree-all tmp.c -o mm4x4
> But i'm getting an error about tmp.c. Is this what Andrew ment with tmp.c?
> or does tmp.c stand for something else? Because I really need the SSA file
> too.

if you do gcc mm4x4.cpp -fdump-tree-all -o mm4x4, you should get a
bunch of files named mm4x4.cpp.<dump file number>.<pass name>

Most passes dump the IR as it looks after their pass.
Any of the passes in the middle are dumps that are in SSA form.

> Btw, Thank you Andrew
> Thank you,
> Mostafa Shahto
>
> Andrew Haley wrote:
> >
> > Mostafa Shahto writes:
> >  >
> >  > Hey,
> >  > I have a couple of question regarding dumping gimple:
> >  > 1)How can I get Gimple after a certain optimization pass (say loop
> >  > unrolling)?
> >
> > -fdump-tree-*.
> >
> >  > 2)Is there a way to get Gimple after all optimization passes?
> >
> > -fdump-tree-all dumps all of the passes.
> >
> >  > 3)If neither of 1) and 2) is possible, what presentation in GCC allows
> >  > getting an optimized low level code in a similar form to Gimple.
> >  > 4)I'm using the -fdump-tree-ssa command when compiling with gcc, but
> > I'm not
> >  > getting a *.ssa file in the folder,
> >
> >  $ gcc -S -fdump-tree-all tmp.c -O
> >  $ ls tmp.c*
> > tmp.c  tmp.c.t26.ssa
> >
> > Andrew.
> >
> > --
> > Red Hat UK Ltd, Amberley Place, 107-111 Peascod Street, Windsor,
> > Berkshire, SL4 1TE, UK
> > Registered in England and Wales No. 3798903
> >
> >
>
> --
> View this message in context: http://www.nabble.com/Dumping-Gimple-After-Optimization-tf3505050.html#a9866437
> Sent from the gcc - Help mailing list archive at Nabble.com.
>
>