[Bug tree-optimization/53346] [4.6/4.7/4.8 Regression] Bad vectorization in the proc cptrf2 of rnflow.f90

[ubizjak at gmail dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=53346

--- Comment #10 from Uros Bizjak <ubizjak at gmail dot com> 2012-05-18 15:11:53 UTC ---
(In reply to comment #8)
> (In reply to comment #7)
> > Fixed.
> 
> Unfortunately, the loop in original rnflow test still gets vectorized, with no
> change in the runtime:

With -O2 -ftree-loop-distribute-patterns -ftree-vectorize, the runtime is still
the same:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls   s/call   s/call  name    
 43.76      9.70     9.70       64     0.15     0.15  cptrf2_
 40.69     18.72     9.02     6685     0.00     0.00  trs2a2.2054
  7.35     20.35     1.63       64     0.03     0.03  gentrs_
  2.21     20.84     0.49       64     0.01     0.01  cptrf1_
  1.44     21.16     0.32        1     0.32    12.32  matsim_
  1.17     21.42     0.26     6685     0.00     0.00  invima.2045
  0.81     21.60     0.18       64     0.00     0.00  cmpcpt_
  0.54     21.72     0.12        1     0.12     9.85  evlrnf_