The crash log has a sufficient backtrace. What it looks like is that the JIT got in an infinite loop and exhausted the stack. However, if that's truly what occurred, that should happen on G3 also. The debug version has no AltiVec code in it either, so that can be exonerated as a cause.
My suspicion is that FPR6 enabled something that behaves badly on boxed, but I won't be able to debug this further until the end of next week.