Benchmark for standard MRI reconstruction

[aTrotier]

Hi,
Thanks to @tknopp we know have access to a large number of non uniform FFT implementation.

What are your thoughts about some MRI specific benchmarks to help the user choose the right one for example :

• 2D and 3D acquisition
• Large matrix > 256^3
• Multi-channel reconstruction (don't know if it will make a difference)
• CPU / GPU

Aurélien

[tknopp]

I am already working on it: https://github.com/JuliaMath/NFFT.jl/tree/master/benchmark

It will not be MRI specific though but I will focus on common sizes used in the NFFT literature.

At this point what is more important is the interface and the correctness. So I use these wrappers to test if AbstractNFFTs has the right interface.