This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
| Other format: | [Raw text] | |
Richard Kenner wrote: > > Why does it use over 200 MB of memory and take over 15 minutes of CPU time on > a moderately-fast machine? I think it is a generation of DFA pipeline hazard recognizer (it is implemented in genautomata.c). It takes 2m50s on a 900Mhz Itanium2 machine. It is a small part of all compiler build time. But the advantages of usage of DFA pipeline hazard recognizer is bigger. The code generated by gcc with the DFA pipeline hazard recognizer is 3% faster for SPECINT2000 on Itanium2 and gcc itself is 60% faster. Generally speaking, genautomata.c makes a lot of work what gcc did earlier during compiling each file. It is made once during building the compiler. I understand that is inconvenient for gcc developers but gcc users win a lot which is more important with my point of view. I think there are still opportunities to decrease memory consumption and it will be on my TODO list. Vlad
| Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
|---|---|---|
| Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |