I think you're mistaken (and in the wrong thread). I can't say I've seen "many" claims from people about having achieved 64bit support for that obsolete chip (none in fact). If you want ML with graphics acceleration on X3100 GPU, you need MLPF to run ML in 32bit mode with DP1 kernel and a mix of DP1/10.7.5 kexts and graphics frameworks. The original ML-related MLPF thread remains active in OS X 10.8 forum subsection.
I understand that MLforAll may have achieved 64bit support in MacPostFactor but until it's eventually published, you're limited to 32bit MLPF.
https://forums.macrumors.com/threads/1325709/
Regarding the 64bit kexts, Google will easily provide you links...