------- Additional Comments From stevenb at suse dot de 2004-11-04 00:28 ------- Subject: Re: [4.0 Regression] jump threading on trees is slow with switch statements with large # of cases
> However, there's clearly an algorithmic problem in this code. There is. The loop predictors are quadratic in the loop nest depth. Honza already suggested predicting the loops from the innermost one to outer, and just not predict the outer loops with such an expensive algorithm. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=15524
