------- Comment #4 from kargl at gcc dot gnu dot org 2007-01-11 17:25 ------- (In reply to comment #3) > Paul, Steve, Please let me know if you see an issue with the above analysis. > Hopefully, the three-loop strategy of (i) determine mask size, (ii) populate > mask and (iii) conditionally execute loop makes sense?
Roger, I haven't looked too closely at the guts of the forall implementation, but your analysis and 3-loop strategy to fixing the bug seems reasonable. In fact, most of the forall code may be the original code implemented by pbrook. I particularly like the concept of "make it work and then worry about optimizations". One thing to keep in mind is that the forall statement is a parallel assignment operation. You may be able to take advantage of the tree-vectorize work for the inner loop. -- kargl at gcc dot gnu dot org changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |kargl at gcc dot gnu dot org http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30404