diff options
| author | Duncan Coutts <duncan@well-typed.com> | 2011-06-01 19:48:15 +0100 |
|---|---|---|
| committer | Duncan Coutts <duncan@well-typed.com> | 2011-07-18 16:31:14 +0100 |
| commit | fa8d20e6d85212290b633159b6ef2d77fb1c4021 (patch) | |
| tree | 0dd7f7926d2c3a482e451a691f794fde2cae9a38 /rts/Sparks.c | |
| parent | 556557ebee2758acade603e25a8a16266dea791d (diff) | |
| download | haskell-fa8d20e6d85212290b633159b6ef2d77fb1c4021.tar.gz | |
Classify overflowed sparks separately
When you use `par` to make a spark, if the spark pool on the current
capability is full then the spark is discarded. This represents a
loss of potential parallelism and it also means there are simply a
lot of sparks around. Both are things that might be of concern to a
programmer when tuning a parallel program that uses par.
The "+RTS -s" stats command now reports overflowed sparks, e.g.
SPARKS: 100001 (15521 converted, 84480 overflowed, 0 dud, 0 GC'd, 0 fizzled)
Diffstat (limited to 'rts/Sparks.c')
| -rw-r--r-- | rts/Sparks.c | 8 |
1 files changed, 6 insertions, 2 deletions
diff --git a/rts/Sparks.c b/rts/Sparks.c index d358ae6660..26b8199035 100644 --- a/rts/Sparks.c +++ b/rts/Sparks.c @@ -64,8 +64,12 @@ newSpark (StgRegTable *reg, StgClosure *p) SparkPool *pool = cap->sparks; if (!fizzledSpark(p)) { - pushWSDeque(pool,p); - cap->spark_stats.created++; + if (pushWSDeque(pool,p)) { + cap->spark_stats.created++; + } else { + /* overflowing the spark pool */ + cap->spark_stats.overflowed++; + } } else { cap->spark_stats.dud++; } |