AMD 250 Computer Hardware User Manual


 
Appendix E SSE and SSE2 Optimizations 357
Software Optimization Guide for AMD64 Processors
25112 Rev. 3.06 September 2005
E.2 Zeroing Out an XMM Register
Optimization
When it is necessary to zero out an XMM register, use an instruction whose format matches the
format required by the consumers of the zeroed register.
Application
This optimization applies to:
32-bit software
64-bit software
Rationale
When an XMM register must be set to zero, using the appropriate instruction helps reduce the chance
of any performance penalty later.
Table 21 shows the different possible consumers of an XMM register and the corresponding
instruction that should be used to zero out the register.
Table 21. Clearing XMM Registers
Producer of Zero Example Consumers of Zero
xorpd xmm1, xmm1 cmppd xmm1, xmm2
cmpsd xmm1, xmm2
comisd xmm1, xmm2
maxpd xmm1, xmm2
maxsd xmm1, xmm2
ucomisd xmm1, xmm2
subsd xmm1, xmm2
xorps xmm1, xmm1 cmpps xmm1, xmm2
cmpss xmm1, xmm2
comiss xmm1, xmm2
maxps xmm1, xmm2
maxss xmm1, xmm2
ucomiss xmm1, xmm2
subss xmm1, xmm2