On Thu, 18 Feb 1999, Daniel C. Sobral wrote: > Lucky you. -O2 *does* break world for many people. Eventually, it > might break your world too, and there is a great chance you'll first > spam -current before changing -O2 to -O and trying again. Or there > would, if we stopped hitting on this nail. :-)
Odd..I've been running -O2 -mno-486 -pipe for at least a year now, without noticing any world problems (or weird behaviour from binaries) other than the commit-related breakage everyone else has seen due to legitimate bugs. What would cause some people to have problems with -O2 and others not? I also run my kernels -O3 because they work for me, but I understand the potential problems with this and drop it back if I ever have problems. This reminds me of something I need to benchmark a bit more: I was testing the optimization of egcs 1.1.1 and found that -O3 produced executables which were significantly slower than -O2, no matter what architecture settings I used. From memory, for -O2 binaries I was getting ~20% speed improvements on my simple test suite compiling with "-mpentium -march=pentium" compared to "-mno-486" on GCC 2.7.2.3 Kris > -- > Daniel C. Sobral (8-DCS) > d...@newsguy.com > d...@freebsd.org > > Well, as a computer geek, I have to believe in the binary universe. > > > To Unsubscribe: send mail to majord...@freebsd.org > with "unsubscribe freebsd-current" in the body of the message > ----- (ASP) Microsoft Corporation (MSFT) announced today that the release of its productivity suite, Office 2000, will be delayed until the first quarter of 1901. To Unsubscribe: send mail to majord...@freebsd.org with "unsubscribe freebsd-current" in the body of the message
