GCC trunk SPEC2000 performance

[Diego Novillo]

On Thu, 20 Jun 2002, Andreas Jaeger wrote:

> David Edelsohn <dje@watson.ibm.com> writes:
> 
> > 	To avoid the argument that GCC 3.2 performance regressions were
> > reported too late in the release cycle, does anyone know why the graphs on
> > Andreas Jaeger's SPEC CPU2000 performance tracking page show:
> >
> > 1) 181.mcf, 254.gap, 172.mgrid, performance has fallen precipitously.  The
> > trunk produces worse code than GCC 2.95, 3.0, and 3.1.
> 
> My pages contain the diffs between single runs so if somebody likes to
> investigate...
> 
Similarly, with SPEC95: tomcat, hydro2d and fpppp have also
fallen below 3.1 levels.


> > 2) "peak" (-O3 -funroll-all-loops -fprefetch=loop-arrays) produces slower
> > code than base (-O2) for many tests.
> 
> And one more point: 301.apsi does not build on peak at all.
> 
May be this is related to the SPECfp95 failures we've been having
for a while now?  turb3d, apsi and wave5 all fail to compile at
-O3 with similar errors:

-----------------------------------------------------------------------------
/home/dnovillo/perf/toribio/spec95/gcc/i686/inst/bin/g77 -c -o turb3d.o   -O3 turb3d.f
Error  from 'specmake  build 2> make.err > make.out':
turb3d.f: In program `main':
turb3d.f:54: warning:
         CALL VERIFY()
              ^
Reference to unimplemented intrinsic `VERIFY' at (^) (assumed EXTERNAL)
turb3d.f: In subroutine `xyfft':
turb3d.f:1266: internal compiler error: RTL flag check: REG_POINTER used with unexpected rtx code `mem' in copy_rtx_and_substitute, at integrate.c:2065
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://www.gnu.org/software/gcc/bugs.html> for instructions.
specmake: *** [turb3d.o] Error 1
-----------------------------------------------------------------------------


Diego.