After getting no feedback on my last mail so far, here is a series of proposed patches for the stable-2.0 branch. In total, they speed up the unzip benchmark by a factor of about 4.7, but it's still about 1.8 times slower than what we'd get for using generic (non-fixnum) operations.
- [PATCH] Take some lowhanging fruit to speed up R6RS fixnu... Andreas Rottmann
- Re: [PATCH] Take some lowhanging fruit to speed up R... Ludovic Courtès
- Re: [PATCH] Take some lowhanging fruit to speed ... Andreas Rottmann
- Re: [PATCH] Take some lowhanging fruit to sp... Andreas Rottmann
- Re: [PATCH] Take some lowhanging fruit t... Ludovic Courtès
- Re: [PATCH] Take some lowhanging fr... Andreas Rottmann
- Re: Take some lowhanging fruit to speed up R6RS fixnum op... Andy Wingo
- Re: Take some lowhanging fruit to speed up R6RS fixn... Andreas Rottmann
- Re: Take some lowhanging fruit to speed up R6RS fixn... Andreas Rottmann
- Re: Take some lowhanging fruit to speed up R6RS fixn... Andreas Rottmann
- R6RS fixnum arithmetic optimizations Andreas Rottmann
- [PATCH 2/3] Several optimizations for R6RS f... Andreas Rottmann
- [PATCH 1/3] Add a few benchmarks for R6RS fi... Andreas Rottmann
- [PATCH 3/3] Add `fixnum?' VM primitive Andreas Rottmann
- Re: [PATCH 3/3] Add `fixnum?' VM primiti... Andy Wingo
- Re: [PATCH 3/3] Add `fixnum?' VM pr... Andreas Rottmann
- define-inlinable Ludovic Courtès
- Re: define-inlinable Andreas Rottmann
- Re: define-inlinable Ludovic Courtès
- Re: define-inlinable Andy Wingo
- Re: define-inlinable Ludovic Courtès