> * vacc3: vectorised with loop in C++
I shaved a bit off this with a simple micro-optimization. In vacc3a replace:
p = std::max(p, 0.0);
p = std::min(p, 1.0);
with:
if(p<0.0)p=0.0;else if(p>1.0)p=1.0; //Clip
I called it vacc6, and a couple of runs are shown below. It is worth
about 4-5% (on the median time).
(Maybe this dilutes your point about comparing vectorization approaches;
but I think it is important to note that loops offer you this possibility.)
Darren
expr min lq median uq
max
1 vacc1(age, female, ily) 10180.614 10531.933 10888.4910 11289.1540
42388.079
2 vacc2(age, female, ily) 330.790 337.499 362.4295 404.7460
516.647
3 vacc3(age, female, ily) 44.270 46.742 51.5535 55.7330
72.879
4 vacc4(age, female, ily) 56.375 58.562 66.2860 71.2345
97.836
5 vacc5(age, female, ily) 45.483 49.820 53.0070 58.3720
81.354
6 vacc6(age, female, ily) 44.304 45.610 49.7165 56.5450
69.215
expr min lq median uq
max
1 vacc1(age, female, ily) 10406.705 10594.7065 10928.0790 11216.0370
15318.001
2 vacc2(age, female, ily) 327.804 331.9610 354.8430 363.3185
462.903
3 vacc3(age, female, ily) 44.013 45.7835 49.1725 56.7800
124.539
4 vacc4(age, female, ily) 56.354 58.3935 62.1805 69.9180
106.800
5 vacc5(age, female, ily) 44.897 46.5435 49.9435 57.9220
72.421
6 vacc6(age, female, ily) 44.124 45.0205 47.0645 54.1135
80.783
_______________________________________________
Rcpp-devel mailing list
Rcpp-devel@lists.r-forge.r-project.org
https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/rcpp-devel
