[GH #175] Slightly reorganise current and add new sort benchmarks #950

Merged
merged 8 commits into from Mar 24, 2013

Conversation

Projects
None yet
2 participants
@paultcochrane
Contributor

paultcochrane commented Mar 24, 2013

These patches do some slight reorganisation of the existing sort benchmarks and add two new sort benchmarks: one for ResizableIntegerArray and one for ResizableFloatArray.

Here are the timing results on my Debian wheezy (amd64) system (Intel(R) Core(TM)2 Duo CPU T8300 @ 2.40GHz), just in case anyone's interested:

time ./parrot examples/benchmarks/sort_ffa.pir
Sorted 1000000 floats

real    0m0.797s
user    0m0.704s
sys 0m0.028s

time ./parrot examples/benchmarks/sort_fia.pir
Sorted 1000000 integers

real    0m0.666s
user    0m0.648s
sys 0m0.020s

time ./parrot examples/benchmarks/sort_rfa.pir
Sorted 1000000 floats

real    0m0.730s
user    0m0.704s
sys 0m0.020s

time ./parrot examples/benchmarks/sort_ria.pir
Sorted 1000000 integers

real    0m0.754s
user    0m0.680s
sys 0m0.020s

paultcochrane added some commits Mar 24, 2013

[GH #175] sort.pir benchmark reports how many elements were sorted
This gives us output which can be checked in the benchmarks.t test, such
that the sort.pir test can be added to the benchmarks test suite
[GH #175] sort_ffa.pir benchmark prints number of elements sorted
It's good to see that a program gives some output to the user, so that we
know that *something* happened.  Now we have this output and can add the
relevant search in the benchmarks.t test suite.
[GH #175] Renamed sort.pir benchmark to sort_fia.pir
This makes it more obvious that this benchmark is for sorting
FixedIntegerArrays in much the same way that sort_ffa.pir benchmarks sorting
FixedFloatarrays
@leto

This comment has been minimized.

Show comment Hide comment
@leto

leto Mar 24, 2013

Owner

Looks great! I don't think we will see much of huge difference in the timings for these benchmarks when run on a single commit of Parrot, but my hunch is that they will be useful to track the general performance of Parrot for various different commits.

Owner

leto commented Mar 24, 2013

Looks great! I don't think we will see much of huge difference in the timings for these benchmarks when run on a single commit of Parrot, but my hunch is that they will be useful to track the general performance of Parrot for various different commits.

leto added a commit that referenced this pull request Mar 24, 2013

Merge pull request #950 from paultcochrane/ptc/sort_benchmarks
[GH #175] Slightly reorganise current and add new sort benchmarks

@leto leto merged commit 6a7f23f into parrot:master Mar 24, 2013

@paultcochrane paultcochrane deleted the paultcochrane:ptc/sort_benchmarks branch Mar 7, 2014

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment