>>>>> Vladimir Makarov writes: Vlad> I did investigate the current status of the infrastructure on future Vlad> mainstream processor Core2 (> 11% slower compiler, worse code and bigger Vlad> code size). That is the reason why I started this.
You do not believe that this is a concern of others? You do not believe that this will be addressed after the merge? This could be an example of GCC optimization becoming more aggressive with increased accuracy, causing increased register pressure problems, which is particularly detrimental to GCC for IA-32. Why don't we analyse and fix any problems instead of trying to keep GCC's infrastructure weak and stupid to cover up its inadequacies? Complaining about and blocking the merge of df does not solve the problem, it only delays it. David