It has been a while since I last tried to use MATLAB testbench and model generated by the FFT core. However, it seems like in the newer version of the FFT (not sure when exactly this was done), the FFT MATLAB model is in .mexw32 format instead of the .dll format (that was used in previous version). Needless to say, I had trouble running the MATLAB testbench and model since my MATLAB version isn't recent enough to recognize .mexw32 files.
After spending sometimes at MATLAB support group, it seems like the .mexw32 file can be changed easily to .dll file (simply by changing the extension manually). Please see the MATLAB thread:
http://www.mathworks.com/matlabcentral/newsreader/view_thread/157605 Also, to run the MATLAB testbench, I had to do the following:
(1) Copy SVSfftmodel.mexw32 from the FFT's library (
C:\altera\72\ip\fft\libraries) to a working directory. Change the newly copied .mexw32 file to a .dll file (by simply changing the file extension).
(2) Comment out line 28 of <FFT_Variant>_model.m => The original line added the default FFT library path to the current working path, that way, MATLAB knows where to find the .mexw32 FFT model. By removing that line, you will be forcing MATLAB to use the copied .dll instead.
(3) Run the HDL simulation to generate blksize_report.txt and inverse_report.txt.
Then you should be able to run the testbench. Hope this saves you some time.
