What is the "batch" in CUFFT?

[_Big_Mac]

The CUFFT library manual is not very elaborate on this. It only says it’s for doing multiple FFTs in parallel. Does that mean multiple FFTs on the same exact dataset? Or on different parts of the input array?

Is it like this?
External Media

Does it mean that the input (and output) array needs to be the size of FFTSIZE*BATCH?

I’m interested in 1d R2C transforms for turning audio from time domain to frequency domain (spectrum analysis).

[vpodlozhnyuk]

Single batched 1D plan execution transforms N vectors – one transform per each vector in the batch. Stride between vectors is currently always zero (non-adjustable yet)

[_Big_Mac]

How are the vectors represented, are they rows or columns of a 2d array?

Like this?

External Media

Excuse my ignorance :)

Oh and about the stride, the manual says:

For 1D complex‐to‐complex transforms, the stride between signals in a

batch is assumed to be the number of cufftComplex elements in the

logical transform size. However, for real‐data FFTs, the distance

between signals in a batch depends on whether the transform is inplace

or out‐of‐place. For in‐place FFTs, the input stride is assumed to

be 2*(N/2+1) cufftReal elements or N/2+1 cufftComplex elements.

For out‐of‐place transforms, the input and output strides match the

logical transform size (N) and the non‐redundant size (N/2+1),

respectively.