Well, realistically, if you want to find eigenvalues and eigenvectors, and its not too fast. (I dont think EEG is fast), then you need a processor with double-precision numbers to do that, and program the algorithm on it.

The most obvious processor to use is an ARM processor. There are others too, please have a look at IpExtreme. If the target of your work is an FPGA, it is more recommended to use an FPGA with an embedded processor, and not write in Verilog. There is also the possibility to use a microprocessor, and not do anything in Verilog or FPGA.

Can you give some background on what you are trying to do ?.

Best Regards,

Henri.

I want to process an EEG signal in which noise or artifacts are present and ORICA algorithm is a technique to remove those artifacts to get clean EEG signal. So in this algorithm i need to calculate inverse and inverse square root of matrices and since SVD is best for complex calculations like these and also it is fast. But problem here is in implementation of SVD. Actually i want to implement this design finally on FPGA.

Well,  my advice is to either take an FPGA with a suitable (ARM) processor with a decent floating-point unit, and implement it on that processor, or, use a cheaper FPGA, with an external microprocessor, and implement it on that.

In my opinion, the C-code to run it on a processor is already quite complicated. (You could look for some pre-existing library - dont bet the firm on it.) Doing in Verilog means basically doing a processor first, and the C-code second, and -if you dont  license- that's more or less a mission impossible.

Best Regards,

Henri.

Ok. Thanks for your advice. I will look forward to this procedure. And i also know its kindof impossible on verilog thats why am looking for options. Even i cant use IP CORE directory which VIVADO design suite provides for SVD.