Intel FP Optimization Differences

different optimizations are required for different family members

• 486 - avoid an excessive number of FXCH instructions since they require execution by the FPU and thus reduce the FP performance
• Pentium - FXCH instructions can be strategically paired with other instructions to improve performance
• P6 (Pentium II/III/M) - FXCHs are essentially free since they are handled by the register renaming hardware
• Pentium 4 - avoid FXCHs since they consume slots in the trace cache and there are also issue slot restrictions (use SSE2 instructions instead where possible)