diff options
author | Mattias Andrée <maandree@kth.se> | 2023-12-17 13:23:51 +0100 |
---|---|---|
committer | Mattias Andrée <maandree@kth.se> | 2023-12-17 13:23:51 +0100 |
commit | 2e7b4df9f7dfd6a4a6796cd2fcee010ea78427ea (patch) | |
tree | a321a6a0b4bc93cbc1b7704239a675c490383b6d /config.mk | |
parent | Tell the user whether signals and errors are signed or unsigned (diff) | |
download | libsyscalls-2e7b4df9f7dfd6a4a6796cd2fcee010ea78427ea.tar.gz libsyscalls-2e7b4df9f7dfd6a4a6796cd2fcee010ea78427ea.tar.bz2 libsyscalls-2e7b4df9f7dfd6a4a6796cd2fcee010ea78427ea.tar.xz |
Miscellaneous improvements
Signed-off-by: Mattias Andrée <maandree@kth.se>
Diffstat (limited to 'config.mk')
-rw-r--r-- | config.mk | 13 |
1 files changed, 13 insertions, 0 deletions
@@ -23,6 +23,19 @@ LDFLAGS = # which returns the index, plus 1, of the least significant set bit # or 0 for 0, you can add -DHAVE_BUILTIN_FFSLL to CPPFLAGS; if you # are using GCC>=3.4 or clang>=5, this is added automatically. +# Additionally, you can add -DUSE_INTERPOLATION_SEARCH to CPPFLAGS +# if you want the library to use interpolation search instead of +# binary search, this is probably a bad idea on platforms that do +# not support division natively, even on amd64, it does not seem to +# make any difference at the moment. If you add -DUSE_INTERPOLATION_SEARCH, +# you may also add `-DINTERPOLATION_SEARCH_FLOAT=long double` (default), +# `-DINTERPOLATION_SEARCH_FLOAT=double` or `-DINTERPOLATION_SEARCH_FLOAT=float` +# to specify which floating point type interpolation search should use +# if it cannot used integers; which option is best depends on the +# platform, as some platforms can work natively with any of the types +# and those `float` is most performant, and other platforms (such as +# i386) can only work natively with a specific type (`long double` in +# the case of i386) and thus performs best with that type. DOWNLOAD = curl -- |