From mboxrd@z Thu Jan 1 00:00:00 1970 From: Paul Koning To: Joe.Buck@synopsys.com Cc: dewar@gnat.com, gcc@gcc.gnu.org Subject: Re: Faster compilation speed Date: Tue, 13 Aug 2002 17:24:00 -0000 Message-id: <15705.41803.489970.482578@pkoning.akdesign.com> References: <20020813173609.3DC80F2D49@nile.gnat.com> <200208132353.QAA07516@atrus.synopsys.com> X-SW-Source: 2002-08/msg00798.html >>>>> "Joe" == Joe Buck writes: Joe> Robert Dewar writes: >> But remember that work you put in on speeding up the compiler is >> work that you do not put in on improving the compiler. As time >> goes on, quality of generated code continues to be critical, >> compiler speed is less critical. Joe> Um, possibly you forget that in order to get a change accepted, Joe> the contributor has to do a three-stage bootstrap with the Joe> compiler and run the regressions. If that process ran three Joe> times faster, contributors could try three times as many Joe> versions of possible patches in the same time period. Yes indeed. On a reasonable PC (a 2-year old laptop) a single iteration of that process takes many hours. paul