On Saturday 04 May 2002 03:18, Christian Jönsson wrote: > Hej. > > I'm just test building the gcc sources in a windows xp/cygwin > environment, below is the output of cygcheck.exe -c -s -r. > > I'm astonished of the time it takes to bootstrap/build gcc-3.1 and run > the testsuite on this windows xp/cygwin system, with 1.8 GHz proc and > 1GB mem, compared to sparc32/linux with dual 75 MHz proc with 320 MB > mem... > > Does it really take so my effort by adding the cygwin "layer" so that > these two systems take about equal time to compile and check > gcc-3.1/3.2? > > Especially, am I perhaps doing something stupid here? > > > Cheers, > > /ChJ > I haven't tried any worst-case setup, but gcc-3.1 definitely builds faster under cygwin on my 3rd-hand P4/1.7/768MB at the office than on this P-III laptop, on which the full bootstrap/check cycle is about 4 hours ( a little over double the time taken under SuSE-7.3 Personal). P4 of course is more sensitive to issues such as alignment, but I won't beat that to death, when you haven't told us what type of delay you are seeing. In the old days, under NT4, gcc builds under cygwin would spend a great deal of time at less than 10% CPU on a PPro/200, and it would take all day to perform bootstrap/check. Nowadays, the Windows performance meter shows mostly over 90% CPU usage on a single CPU box, and typically about 60% system time, or whatever Windows calls it, when I specify -Os -mpreferred-stack-boundary=4 -march=pentium[34] in BOOT_CFLAGS, having rebuilt binutils accordingly. If you have chosen slower options, you should get higher CPU usage. Low CPU utilization could come from fragmented disk or registry, or having virus checking turned on. The way my employer configures McAfee, it takes 4 times as long to run bash scripts. -- Tim Prince
-- Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple Bug reporting: http://cygwin.com/bugs.html Documentation: http://cygwin.com/docs.html FAQ: http://cygwin.com/faq/
