adjoint simulation time 5 times longer

[poutine-dejeuner]

I am running a 2D simulation over a (280, 375) grid and need to compute the gradient. The forward run takes around 40 secs and the adjoint around 180 sec. I got 2 questions

1. are there some ways to reduce the simulation time (besides symmetries)?
2. is it normal that the adjoint run takes 5 times longer than the forward run?

[stevengj]

Do you have very closely spaced frequencies?