Thu Nov 6 03:36:39 PST 2008 Simon Marlow <[EMAIL PROTECTED]>
* Run sparks in batches, instead of creating a new thread for each one
Signficantly reduces the overhead for par, which means that we can
make use of paralellism at a much finer granularity.
M ./compiler/prelude/primops.txt.pp +6
M ./includes/StgMiscClosures.h +1
M ./rts/Capability.c -7 +22
M ./rts/Capability.h -1 +5
M ./rts/Linker.c +1
M ./rts/Prelude.h +1
M ./rts/PrimOps.cmm +22
M ./rts/Schedule.c -30 +4
M ./rts/Sparks.c -4 +6
M ./rts/Sparks.h -2 +2
M ./rts/package.conf.in -5 +3
View patch online:
http://darcs.haskell.org/ghc/_darcs/patches/20081106113639-12142-d8372a684994409b8b9fe8d0f4f9cc7f749459ce.gz
_______________________________________________
Cvs-ghc mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/cvs-ghc
