314 Instruction Latencies Appendix C
25112 Rev. 3.06 September 2005
Software Optimization Guide for AMD64 Processors
C.5 3DNow!™ Technology Instructions
Table 16. 3DNow!™ Technology Instructions
Syntax
Encoding
Decode
type
FPU
pipe(s)
Latency Note
Prefix
byte(s)
imm8
ModRM
byte
FEMMS 0Fh 0Eh DirectPath FADD/FMUL/
FSTORE
22
PAVGUSB mmreg1, mmreg2 0Fh, 0Fh BFh 11-xxx-xxx DirectPath FADD/FMUL 2
PAVGUSB mmreg, mem64 0Fh, 0Fh BFh mm-xxx-xxx DirectPath FADD/FMUL 4
PF2ID mmreg1, mmreg2 0Fh, 0Fh 1Dh 11-xxx-xxx DirectPath FADD 4
PF2ID mmreg, mem64 0Fh, 0Fh 1Dh mm-xxx-xxx DirectPath FADD 6
PFACC mmreg1, mmreg2 0Fh, 0Fh AEh 11-xxx-xxx DirectPath FADD 4
PFACC mmreg, mem64 0Fh, 0Fh AEh mm-xxx-xxx DirectPath FADD 6
PFADD mmreg1, mmreg2 0Fh, 0Fh 9Eh 11-xxx-xxx DirectPath FADD 4
PFADD mmreg, mem64 0Fh, 0Fh 9Eh mm-xxx-xxx DirectPath FADD 6
PFCMPEQ mmreg1, mmreg2 0Fh, 0Fh B0h 11-xxx-xxx DirectPath FADD 2
PFCMPEQ mmreg, mem64 0Fh, 0Fh B0h mm-xxx-xxx DirectPath FADD 4
PFCMPGE mmreg1, mmreg2 0Fh, 0Fh 90h 11-xxx-xxx DirectPath FADD 2
PFCMPGE mmreg, mem64 0Fh, 0Fh 90h mm-xxx-xxx DirectPath FADD 4
PFCMPGT mmreg1, mmreg2 0Fh, 0Fh A0h 11-xxx-xxx DirectPath FADD 2
PFCMPGT mmreg, mem64 0Fh, 0Fh A0h mm-xxx-xxx DirectPath FADD 4
PFMAX mmreg1, mmreg2 0Fh, 0Fh A4h 11-xxx-xxx DirectPath FADD 2
PFMAX mmreg, mem64 0Fh, 0Fh A4h mm-xxx-xxx DirectPath FADD 4
PFMIN mmreg1, mmreg2 0Fh, 0Fh 94h 11-xxx-xxx DirectPath FADD 2
PFMIN mmreg, mem64 0Fh, 0Fh 94h mm-xxx-xxx DirectPath FADD 4
PFMUL mmreg1, mmreg2 0Fh, 0Fh B4h 11-xxx-xxx DirectPath FMUL 4
PFMUL mmreg, mem64 0Fh, 0Fh B4h mm-xxx-xxx DirectPath FMUL 6
PFRCP mmreg1, mmreg2 0Fh, 0Fh 96h 11-xxx-xxx DirectPath FMUL 3
PFRCP mmreg, mem64 0Fh, 0Fh 96h mm-xxx-xxx DirectPath FMUL 5
PFRCPIT1 mmreg1, mmreg2 0Fh, 0Fh A6h 11-xxx-xxx DirectPath FMUL 4
PFRCPIT1 mmreg, mem64 0Fh, 0Fh A6h mm-xxx-xxx DirectPath FMUL 6
PFRCPIT2 mmreg1, mmreg2 0Fh, 0Fh B6h 11-xxx-xxx DirectPath FMUL 4
PFRCPIT2 mmreg, mem64 0Fh, 0Fh B6h mm-xxx-xxx DirectPath FMUL 6
PFRSQIT1 mmreg1, mmreg2 0Fh, 0Fh A7h 11-xxx-xxx DirectPath FMUL 4
Notes:
1. For the PREFETCH and PREFETCHW instructions, the mem8 value refers to an address in the 64-byte line to
be prefetched.
2. The byte listed in the column titled ‘imm8’ is actually the opcode byte.