codecov[bot]
commented
3 years ago Codecov Report
Merging #281 (76f7859) into master (c57a2be) will increase coverage by
0.02%. The diff coverage is100.00%.
@@ Coverage Diff @@
## master #281 +/- ##
==========================================
+ Coverage 77.86% 77.88% +0.02%
==========================================
Files 43 43
Lines 18389 18394 +5
==========================================
+ Hits 14319 14327 +8
+ Misses 4070 4067 -3 | Impacted Files | Coverage Δ | |
|---|---|---|
| gap/base/recognition.gi | 69.02% <100.00%> (+0.23%) |
:arrow_up: |
| gap/perm/giant.gi | 90.35% <0.00%> (+0.67%) |
:arrow_up: |
The first commit fixes a bug in RandomElm. This fix is needed to call RandomElm on a group which does not store its random elements.
The second commit removes a TODO. The following is of course not a proper benchmark, but just a comparison of two runs of
tst/testall.g. I think it's good enough to warrant that commit though. I originally wrote the comparison only fortst/testquick.gbut usingtst/testall.ggives the same overall picture.Storing the random elements makes the whole test suite and most test files finish roughly 10% earlier than without storing random elements.
There are a few tests which run almost two times as fast with storing random elements enabled by default compared to without it being enabled.
There are a few test files where both approaches take roughly the same time.
Ignoring those test files which take less than 10ms to complete, PermLargeBasePrimitive is the only test file where not storing random elements is significantly faster, namely roughly 5%. Edit: With the
slowandveryslowtests running too,veryslow/MatFDPM.tstis the only other test file where not storing random elements is significantly quicker for me: it takes 20s without storing random elements compared to 33.5s when storing random elements.Here are timings from runs of
tst/testquick.gon my machine. I ran the tests without storing random elements first to exclude the possibility that the whole speed difference came from heating up my CPU. I'm currently also running the full test suite on my machine.With storing random elements
Without storing random elements