This is for the people who use Mac OS X 10.3 and would to use the CVS build as of November 9, 2004.
how difficult is it to make blender using both cpus of the g5s here?
I’m also curious… but about Altivec optimization as well. In testing on my machines between G3s and G4s the speed appears to be purely based on CPU clock-speed and doesn’t appear to be optimized for processor functions. Part of the G3->G4 abilities is Altivec, and the G4->G5 increases are also processor-specific.
Is this something similar to Antares Intel-Optimized builds? Should Mac people be optimizing specific Blender versions for their processors? If that is the case, I am willing to permanently host any user-based processor builds for Mac users…
your build use a dynamic library in /usr/local (freetype) which i have elsewhere so i cant test, but do you have international fonts working ?
I’m interested in that as I have currently some problems with fgtl for the officials builds.
akator, blender is not altivec optimized, and wont be before some long time (you need a special coding for that, and a special coder to understand).
The builds are tuned to run on G4, that’s all, but in a manner that they can run on any G3-G5.
Hmm, shouldn’t you be able to soft link to /usr/local/freetype?
ln -s /usr/local/freetype(?) /your/path/to/the/lib
the case of freetype is a bit special. fgtl need to be linked against a specific version of freetype to run, so for testing fgtl, i must be sure i use the same version of lib
The gcc 3.5 in Tiger next March will have automatic code vectorisation so no longer will you need to be a ‘special coder’. I think Blender is compiled using gcc so next year should herald som vast improvements for G4/G5 users. From http://www.apple.com/macosx/tiger/xcode.html
At the heart of Xcode 2.0 is Apple’s version of gcc 3.5, the next generation of the industry-standard gcc compiler. The new compiler helps you get more performance from your existing code by using a number of advanced optimization techniques. Auto-vectorization, a technique borrowed from the world of supercomputing, helps you to unlock the power of the Velocity Engine in every PowerPC G4 and G5 system without writing vectorized code. Other optimization tools include support for feedback-directed optimization and inter-module analysis.
I thought there was auto dual proc support too but I guess not unless Tiger handles that itself.
I’ve been using fink, had some success with it, the bigger problem has been getting the builds to work when using old darwin pre-compiled libs…
I’m guessing some SConstruct clean up on my end could take care of that…
PopeStewart, if your built with Scons, can you make your SConstruct file available?
As for the SConstruct, you can find it at http://sulley.dm.ucf.edu/~pope/blender/SConstruct
Doh! I actually was looking for your config.opts, however I got it solved…
Actually got a dynmic build to complete, however I noticed one problem, the face/edge/vert select buttons are gone in the latest cvs release?
Something going on here?
the build i sent has that in it.
Yes, my mistake, must be the long hours lately, the most recent build I generated had it as well.
Thanks for all your assistance, perhaps I’ll post a link to my static build here (if it’s all right with you)