BGE    CR4, loadge16       // Load for len(sep) >= 16
	SUB    R6, R16, R9         // 16-len of sep
	SLD    $3, R9              // Set up for VSLO
	MTVSRD R9, V9              // Set up for VSLO
	VSLDOI $8, V9, V9, V9      // Set up for VSLO
	VSLO   ONES, V9, SEPMASK   // Mask for separator len(sep) < 16

loadge16:
	ANDCC $15, R5, R9 // Find byte offset of sep
	ADD   R9, R6, R10 // Add sep len