so, I've got this for loop which does a really simple thing:
the question is.. is it possible, at all, to optimise this further? Or do I have to go to assembly? just trying to analyse sine waves.. real quick.Code:for(i = 0; i < 4096; i++){ int sineindex = i % 16; total += soundsamples[i] * sinetable[sineindex]; }
Leave it up to your compiler. I suppose you could try to unroll your loop with 16 assignments in a single pass, but again, most compilers will optimize better than you can.
Quzah.
Hope is the first step on the road to disappointment.
You're taking a dot product with a sine wave. If you do that at a series of different time shifts, that's a convolution. A convolution can be optimized with an FFT, yielding O(n/log n) speedup.
More information on what you are attempting would help me suggest further optimizations. What you have here, for its obvious purpose, isn't too bad, assuming the compiler optimizes that modulus as it should be doing.
Code://try //{ if (a) do { f( b); } while(1); else do { f(!b); } while(1); //}
Hmm, looks like there's not much I can do then.. I'm only doing this for one series at a time. I am using FFT (in particular kiss fft) but for an unrelated set of operations..
thanks
