Which GPU is best for this application

Friends:

I am reltive new to CUDA and GPU. I am trying to implement some signal processing algorithms on NVIDIA GPU using CUDA. But I am still not sure out of the various series of CUDA -Emabled NVIDIA GPUs, which one should I buy. It should be optimum in terms of cost and performanace. Basically I would be doing following calculations:

Generation of wideband signals

• Conversion of time domain signals into

frequency domain via FFT.

• Compute 64 sets of 64-point FFT

• Computation of 33 covariance matrices in

frequency domain.

• Computation of eigenvalues and eigenvectors

• Computation of initial estimates of number of

sources

• Computation of initial estimates of DOA

• Computation of 33 transformation matrices and

focusing on central frequency

• Computation of eigenvalues and eigenvectors

• Computation of number of sources

• Computation of final estimates of DOA

Can any one of you please let me know which of the CUDA-compatible NVIDIA’s GPUs will be best options. CUDA compatible GPUs are lsited here: http://www.nvidia.com/object/cuda_learn_products.html
My budget is between $200-500

Thanks in advance,

Heshsham
heshsham2001@yahoo.co.in