I have seen similar issues while trying to fix a memory overflow error while compiling one of the newer CPUv6 C++ bms, where cc1plus42 consumes quite a bit of time.
-Dibyendu AMD Open64 Compiler Dev -----Original Message----- From: "C. Bergström" [mailto:cbergst...@pathscale.com] Sent: Sunday, January 23, 2011 12:35 AM To: open64-devel; path64-dev Subject: [Open64-devel] Compilation times Problem On a sufficiently large C++ test program compilation time is quite slow. Here's a quote from the end user for why it matters "cause you know ... in development, I compile more often than I run the damn thing". You can see below that most of the time is spent in gcc. Solutions a. Don't use gcc fe (eg psclang) b. Find the bottleneck in building whirl from libspin (I think transversing the AST tree like we are now is really inefficient and probably unsolvable - See below) c. Don't use c++ ;) 32bit appears to win a little, but not much. We can probably switch to using smaller portable types and save some time. pathCC is a ELF 64-bit LSB executable openCC is a ELF 32-bit LSB executable Thoughts? Anyone else care about this? Timing [phoenix3]# time /opt/ekopath/4.0.7/bin/pathCC -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m9.399s user 0m8.497s sys 0m0.881s [phoenix3]# time /opt/ekopath/4.0.7/bin/pathCC -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m9.397s user 0m8.470s sys 0m0.911s [phoenix3]# time /mnt/psc/crap/open64/bin/openCC -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m7.726s user 0m7.020s sys 0m0.666s [phoenix3]# time /mnt/psc/crap/open64/bin/openCC -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m7.760s user 0m7.080s sys 0m0.635s [phoenix3]# time g++ -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m6.700s user 0m4.887s sys 0m0.456s [phoenix3]# time g++ -I. -I../boost-trunk libs/phoenix/test/core/primitives_tests.cpp real 0m5.743s user 0m4.590s sys 0m0.468s ----------- /opt/ekopath/4.0.7/lib/4.0.7/x8664/cc1plus42 ... real 0m2.022s user 0m1.945s sys 0m0.076s # I didn't look into why it's doing this yet /opt/ekopath/4.0.7/lib/4.0.7/x8664/cc1plus42 ... Segmentation fault /opt/ekopath/4.0.7/lib/4.0.7/x8664/wgen42 ... real 0m0.598s user 0m0.233s sys 0m0.359s /opt/ekopath/4.0.7/lib/4.0.7/x8664/inline real 0m0.087s user 0m0.054s sys 0m0.033s /opt/ekopath/4.0.7/lib/4.0.7/x8664/be real 0m1.359s user 0m1.286s sys 0m0.070s as --64 primitives_tests.s -o primitives_tests.o real 0m0.073s user 0m0.060s sys 0m0.013s time /usr/bin/ld ... real 0m0.045s user 0m0.028s sys 0m0.016s ------------------------------------------------------------------------------ Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)! Finally, a world-class log management solution at an even better price-free! Download using promo code Free_Logger_4_Dev2Dev. Offer expires February 28th, so secure your free ArcSight Logger TODAY! http://p.sf.net/sfu/arcsight-sfd2d _______________________________________________ Open64-devel mailing list Open64-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open64-devel ------------------------------------------------------------------------------ Special Offer-- Download ArcSight Logger for FREE (a $49 USD value)! Finally, a world-class log management solution at an even better price-free! Download using promo code Free_Logger_4_Dev2Dev. Offer expires February 28th, so secure your free ArcSight Logger TODAY! http://p.sf.net/sfu/arcsight-sfd2d _______________________________________________ Open64-devel mailing list Open64-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/open64-devel
