Consider fine tuning rts options for ghc-bin
It's folklore that increasing the size of the allocation area with the -A
rts option improves the performance of GHC.
I'm wondering if by fine-tuning ghc-bin's default rtsopts whether we can improve the experience of the average user.
Right now ghc-bin is compiled with "-with-rtsopts=-K512M -H -I5 -T"
.
Contrast haskell-language-server which is compiled with "-with-rtsopts=-I0 -A128M -T"
My suggestion is that we try compiling head.hackage with a variety of -A
values and see if it makes much of a difference to time and memory usage.