|Opened on:||Tuesday, September 4, 2018 - 04:27|
|Last modified:||Wednesday, September 12, 2018 - 23:40|
The new devprev release seems a bit sluggish at times, so I tried
comparing 20180829-6ed3ad0d20-win_x86-devprev to 20180725-d14dae34c3-win_x86.
On puzzle 1567, I used Rav3n_pl Fuzes v1.5.5 with medium wiggle power.
Starting with a score of 15,345, I ran main and devprev three times, each time increasing the wiggle factor setting used by the recipe. (The Foldit wiggle power setting was medium for all three tests.)
While devprev had a slightly higher score each time, it also took longer.
Interestingly, wiggle factor 4 produced slightly better results than wiggle factor 5, but only by a fraction.
The table below summarizes the results, showing the gain, elapsed time (in hours), and point per hour (PPH) for each test.
While devprev seems a little slower based on this limited testing, the results don't seem to explain the sluggishness I was seeing in 1567. The symptoms included it being difficult to freeze structures by double-clicking on them, an very slow performance in some recipes.