360 SSE and SSE2 Optimizations Appendix E
25112 Rev. 3.06 September 2005
Software Optimization Guide for AMD64 Processors
E.4 Moving Data Between XMM Registers and GPRs
Optimization
Store a register that needs to be spilled in memory, rather than moved to a different register file.
Application
This optimization applies to:
• 32-bit software
• 64-bit software
Rationale
While register moves within a given register file are very efficient (XMM to XMM, GPR to GPR),
moves between register files (XMM to GPR, GPR to XMM) are not. .