Loop optimisation with identical counters

Sebastian Fischer fischer at nii.ac.jp
Sat Nov 6 01:56:01 EDT 2010

> Which proposal do you mean?

I referred to Christian's questions whether it is possible to generate  
`ff` and `gg` from `f` and `g`. If `h` is similar to `g`, then `hh`  
could reuse `ff` while with an inlining approach something like `ff`  
would be duplicated.

> I'm not sure something like that is feasible without knowing the  
> call sites.

I agree. One would need to generate variants for different call sites  
and reuse variants for similar call sites.


More information about the Glasgow-haskell-users mailing list