In article <2025Oct2.224440@mips.complang.tuwien.ac.at>,   
   Anton Ertl  wrote:   
   >dxf  writes:   
   >>For 3DUP I believe this is the one to beat:   
   >>   
   >>: 3DUP ( a b c -- a b c a b c )  dup 2over rot ;   
   >>   
   >>With NTF/LFX the locals version will break even.   
   >   
   >As we already discussed in the thread including   
   ><2021Sep11.083507@mips.complang.tuwien.ac.at>, NTF/LXF produces the   
   >same (optimal for the calling convention used by NTF/LXF) code for   
   >3DUP versions using the data stack, return stack, or locals.  That's   
   >because the actual data flow is always the same, and NTF/LXF can see   
   >this data flow in all three cases.   
      
   The problem with 3DUP is that it is actually used in context.   
   What is the data that is going to 3DUP ped? In my view this   
   amounts to double use of data that is in registers (32 in the riscv)   
   anyway, after an optimiser does his thing.   
      
      
      
   >- anton   
   --   
   The Chinese government is satisfied with its military superiority over USA.   
   The next 5 year plan has as primary goal to advance life expectancy   
   over 80 years, like Western Europe.   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)