diff options
| author | Simon Marlow <marlowsd@gmail.com> | 2009-03-13 13:51:16 +0000 |
|---|---|---|
| committer | Simon Marlow <marlowsd@gmail.com> | 2009-03-13 13:51:16 +0000 |
| commit | 4e3542263207ae49963811aeb84927027e7bb61d (patch) | |
| tree | 90814c02010977694748704fc96bf4347bf2104f /rts/sm/GCThread.h | |
| parent | 8fda9784f1eff82aa243073fd31b181080be3109 (diff) | |
| download | haskell-4e3542263207ae49963811aeb84927027e7bb61d.tar.gz | |
Use work-stealing for load-balancing in the GC
New flag: "+RTS -qb" disables load-balancing in the parallel GC
(though this is subject to change, I think we will probably want to do
something more automatic before releasing this).
To get the "PARGC3" configuration described in the "Runtime support
for Multicore Haskell" paper, use "+RTS -qg0 -qb -RTS".
The main advantage of this is that it allows us to easily disable
load-balancing altogether, which turns out to be important in parallel
programs. Maintaining locality is sometimes more important that
spreading the work out in parallel GC. There is a side benefit in
that the parallel GC should have improved locality even when
load-balancing, because each processor prefers to take work from its
own queue before stealing from others.
Diffstat (limited to 'rts/sm/GCThread.h')
| -rw-r--r-- | rts/sm/GCThread.h | 10 |
1 files changed, 6 insertions, 4 deletions
diff --git a/rts/sm/GCThread.h b/rts/sm/GCThread.h index 3ee2757f19..aacef82cbc 100644 --- a/rts/sm/GCThread.h +++ b/rts/sm/GCThread.h @@ -15,6 +15,7 @@ #define GCTHREAD_H #include "OSThreads.h" +#include "WSDeque.h" /* ----------------------------------------------------------------------------- General scheme @@ -81,13 +82,14 @@ typedef struct step_workspace_ { StgPtr todo_free; // free ptr for todo_bd StgPtr todo_lim; // lim for todo_bd - bdescr * buffer_todo_bd; // buffer to reduce contention - // on the step's todos list + WSDeque * todo_q; + bdescr * todo_overflow; + nat n_todo_overflow; // where large objects to be scavenged go bdescr * todo_large_objects; - // Objects that have already been, scavenged. + // Objects that have already been scavenged. bdescr * scavd_list; nat n_scavd_blocks; // count of blocks in this list @@ -95,7 +97,7 @@ typedef struct step_workspace_ { bdescr * part_list; unsigned int n_part_blocks; // count of above - StgWord pad[5]; + StgWord pad[3]; } step_workspace ATTRIBUTE_ALIGNED(64); // align so that computing gct->steps[n] is a shift, not a multiply |