Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
Re: [henshin-dev] Comb-Pattern Benchmark
Hi Matthias,

my guess is also that it is caused by the matching order. I tried out turning off the automatic variable sorting but that didn't help either :(.

Cheers,
Christian


2014-04-25 13:29 GMT+02:00 Matthias Tichy <matthias.tichy@xxxxxxxxx>:
Hi,

It is not only much slower it also looks exponential (sparese grid) now when it was probably linear previously. Could it be that there was a change in the order of the matching which caused this?

Btw, this is my data for comparison:

Benchmark for matching comb pattern in full grid...
GridLen PatLen Matches Time
60 10 3009 232
60 20 2419 713
60 30 1829 308
60 40 1239 698
60 50 649 1345
60 60 59 2335

Benchmark for matching comb pattern in sparse grid (no matches)...
GridLen PatLen Time
100 10 13
100 20 62
100 30 222
100 40 553
100 50 1142
100 60 2210
100 70 3926
100 80 6374
100 90 9936
100 100 15097

/mtt

From: Christian Krause <henshin.ck@xxxxxxxxx>
Reply-To: Henshin developers mailing list <henshin-dev@xxxxxxxxxxx>
Date: Sonntag, 13. April 2014 15:10
To: Henshin developers mailing list <henshin-dev@xxxxxxxxxxx>
Subject: [henshin-dev] Comb-Pattern Benchmark

Dear Henshin developers,

we have seen some great performance improvements of Henshin, particularly of the match finder, in the past. Unfortunately, it seems like we have also lost some of the matching speed. Specifically, in the comb-pattern benchmark documented at


we have two benchmarks: the last two for matching the comb pattern in the full and the sparse grid, that are significantly slower than documented. So if any of you has time to investigate why this is so much slower now, it would be great. Unfortunately, there is no prize to win ;).

Cheers,
Christian

_______________________________________________
henshin-dev mailing list
henshin-dev@xxxxxxxxxxx
https://dev.eclipse.org/mailman/listinfo/henshin-dev


Back to the top