I did the simple fix of replacing all files that had PIC code with versions that dont . The most serious slowdowns are with dive_1 which is not used that much and add/sub_n which is , here we had to fallback to C code. mul_basecase did not need changing , so there should be not too much of a slow down in general.
On Tuesday 27 October 2009 05:43:49 Bill Hart wrote: > I don't know whether to comment on this one or not. But my > understanding is that Darwin does not support 32 bit PIC relocations. > The *only* way to fix the problem is to delete all the assembly files > which use PIC code, or rewrite them in the very peculiar Darwin way (I > couldn't follow the docs). Or rewrite the files so they don't need PIC > code at all. But one is a performance problem. The other two are > majorly large undertakings. > > I hope I'm commenting on the right issue here. I'm not on my own > computer and can't check anything. So I'm commenting on everything > from (my poor) memory. > > Bill. > > 2009/10/26 Jason Moxham <ja...@njkfrudils.plus.com>: > > And I should of said this fix will apply to any machine which returns a > > config.guess of > > core2-apple-darwin* > > or > > penryn-apple-darwin* > > > > and no others > > If this covers more machines than necessary then we will have a slight > > slowdown on them. > > > > On Monday 26 October 2009 00:41:46 Jason Moxham wrote: > >> svn trunk 2478 should fix everything , > >> so a configure ABI=32 should work now > >> > >> yasm failed configure because in 32bit mode it doesn't like core2 as a > >> cpu type but prefers i686 , and when we force a cpu type with build=.... > >> this overrides yasm autodetect > >> > >> Thanks > >> Jason > >> > >> On Sunday 25 October 2009 23:58:14 Jason Moxham wrote: > >> > first line is the problem > >> > > >> > checking build system type... penryn-apple-darwin9.8.0 > >> > > >> > clearly we need penryn and core2 for the cpu > >> > what about the darwin version ? > >> > I can do darwin* or darwin9.* or ? > >> > We need to know which versions have this different asm code for PIC > >> > > >> > if you do > >> > ./configure --build=core2-apple-darwin > >> > this will force a build with the current code > >> > > >> > Thanks > >> > Jason > >> > > >> > On Sunday 25 October 2009 23:51:37 Case Vanhorsen wrote: > >> > > Coming up, I've done a make distclean and right now I'm running > >> > > > >> > > $ ./configure ABI=32 >conou 2>&1 > >> > > > >> > > ... > >> > > > >> > > OK, file conou attached to this mail. Offhand I can't spot anything > >> > > suspicious, but then I'm definitely no configure expert;-). > >> > > > >> > > > >> > > Thanks, > >> > > > >> > > Alex > >> > > > >> > > On Sun, Oct 25, 2009 at 4:40 PM, Case Vanhorsen <cas...@gmail.com> > > > > wrote: > >> > > > Alex, > >> > > > > >> > > > Can you capture the output of ./configure ABI=32? > > --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "mpir-devel" group. To post to this group, send email to mpir-devel@googlegroups.com To unsubscribe from this group, send email to mpir-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/mpir-devel?hl=en -~----------~----~----~----~------~----~------~--~---
