> All of the FP registers in a disassembly look like <fp inst> ST, or > ST(i) for i in 0-7. You can also look at the trace file and grep for > FR[i], which is the FP register i. (Generate trace file via the > :trace-file option to compile-file.)
Ok, thanks. It seems that the code is only using floating pointer registers 0, 1, and 2. So I'll agree with you that your initial explanation is not highly satisfactory. Which I guess leaves us with no real explanation? In any case, this is no longer the bottleneck in my code (I've changed other things), so it's become more of a wanting to understand issue than a critical performance issue. Cheers, rif
