Closed Astal4 closed 2 weeks ago
Don't use -Ofast, it can screw up math code.
If that doesn't help I have no idea, I can't debug this myself because I have no PPC hardware
Retrying with -O2 -maltivec instead
No difference. still issues with -O2
Depending of the PPC system, altivec doesn't work as it is from those old sources.
I've tested some altivec for MorphOS platform (G4), and one of the functions that does funny things with models is BlendJoints: Try to copy paste the SIMD_Generic version equivalent to the SIMD_Altivec source.
Or try to compile a version that only uses Generic Simd code.
Hi. Same problem with a Quad G5 running Void ppc64. I replaced the whole SIMD_Altivec.cpp file for SIMD_generic and all the models are still stuck with the T-Pose. I can't interact with them either. Was a solution ever found?
EDIT
Compiling and running in a 32bit chroot solved the problem for me (Void ppc64). It appears that the problem il not related to altivec.
This problem should be fixed in the following branch: https://github.com/DanielGibson/dhewm3/tree/fix-ppc64be
@Astal4 @FuryOusFrank could you please test that branch to see if it works for you, and to make sure there are no additional 64bit Big Endian problems we haven't found yet? Thanks in advance! :)
See also #626
@Astal4 @FuryOusFrank could you please test that branch to see if it works for you, and to make sure there are no additional 64bit Big Endian problems we haven't found yet? Thanks in advance! :)
Thanks! My G5 is not working at this time. Overheating caused by the age of the liquid cooling system. I'll look into it as soon as I can.
fix merged
Hello! I've got a PowerMac G5 Quad that I'm playing with, it's running Gentoo, and on the X1900 GT, all models are stuck in a T-Pose. It's a funny glitch, but still a glitch.
Built with
-Ofast
and-maltivec
, this is from the stable release.