why the paramter "batch" is needed in cufftPlan1d?

Accelerated Computing CUDA CUDA Programming and Performance
1

There is no such parameter in fftw or mkl’s fft2dfti wrapper. Why it is needed?

2

Because fftw and mkl are written for scalar processors and cufft is written for massively parallel gpus.

3

Thanks! Then it means I have to manually devide the data for parallism?

4

Or how to select the optimal batch value?

5

You can certainly do single transforms (batch=1). But you can get better speed on average if you can rephrase your algorithm to perform lots of transforms together.

6

In this case, the transforms should be the same length?

Does cufft show higer efficiency than cpu version fft program when batch=1?