Re: serious (5x) link-time regression in 3.3.5-12 on ia64

2005-04-04 Thread David Mosberger
Duraid, How about collecting some profiles (e.g., with qprof or q-syscollect)? That might be a quick way to get some idea as to what's causing the regression. --david > On Sun, 3 Apr 2005 16:04:48 +0200, Matthias Klose <[EMAIL PROTECTED]> > said: Matthias> according to the ch

Re: serious (5x) link-time regression in 3.3.5-12 on ia64

2005-04-03 Thread Matthias Klose
according to the changelog, nothing really has changed ... CCing to the ia64 list. Duraid Madina writes: > Hi Mathias, Gerhard and others, > > I want to report a nasty performance regression in 3.3.5-12. Moving > from: > > gcc version 3.3.5 (Debian 1:3.3.5-11) > > to: > > gcc version 3.3

serious (5x) link-time regression in 3.3.5-12 on ia64

2005-03-29 Thread Duraid Madina
Hi Mathias, Gerhard and others, I want to report a nasty performance regression in 3.3.5-12. Moving from: gcc version 3.3.5 (Debian 1:3.3.5-11) to: gcc version 3.3.5 (Debian 1:3.3.5-12) leads to a large reduction in C++ software build speeds, due to something that apparently ma