spergel profiles are much slower than galsim

[beckermr]

I've heard reports that the spergel profiles are much slower than galsim.

To Do:

• [x] add benchmark comparison to galsim
• [ ] profile to find optimizations
• [x] implement those optimizations

[ismael-mendoza]

quick note, when I checked it was a sum of Exponential and Spergel profile (nu=-0.6) -- might be worth taking a look at the timing of each of these separately, but my guess is that it's very likely coming from the Spergel...

[beckermr]

We've determined that this is from the conversion of half light radius to scale radius. So we're going to focus our optimization on those bits of the code.

[ismael-mendoza]

is this still relevant after #130 ? i forgot about the status, perhaps we can close this issue

[beckermr]

There is still one other thing here I think.

[ismael-mendoza]

i see, do you agree if I check off the first and third item based on #125 and #130? so just the second one is missing (?)