d@0: d@0:
d@0:d@0: d@0: Next: Multi-dimensional Array Format, d@0: Previous: Other Important Topics, d@0: Up: Other Important Topics d@0:
In order to get the best performance from FFTW, one needs to be d@0: somewhat aware of two problems related to data alignment on x86 d@0: (Pentia) architectures: alignment of allocated arrays (for use with d@0: SIMD acceleration), and alignment of the stack. d@0: d@0: d@0: d@0: