After some more research it seems that this is an issue of not having an
even multiple of input samples to taps. For others, here is where I found
the answer: http://www.dspguide.com/ch18/2.htm
On Fri, Feb 10, 2017 at 11:19 AM, Dave NotTelling
wrote:
> I am trying to use the FFT filter in some c++ code and I'm having issues
> with seg faulting. The code below will work, but only because the output
> buffer is twice as large as the input buffer. It doesn't have to be twice
> as large, but it does have to be some number larger than the input buffer.
> Why is that? Also, how do I know what the output buffer size should be?
>
> Thank you!!
>
>
> [code]
>
> #include
> #include
> #include
> #include
>
> int main(){
> // Generate a set of zeroed out taps
> std::vector > taps(512, std::complex(0.0f,
> 0.0f));
>
> // Number of random samples to generate
> const int sampleSize = 3;
> // Create a vector of 32fc samples. Don't initialize it to anything,
> just let it have
> // random data.
> gr_complex * input = (gr_complex *)malloc(sizeof(gr_complex) *
> sampleSize);
>
> // Place to store the output samples from the FFT
> gr_complex * output = (gr_complex *)malloc(sizeof(gr_complex) *
> sampleSize);
>
> // Create a filter to run the input samples through
> gr::filter::kernel::fft_filter_ccc * filter = new
> gr::filter::kernel::fft_filter_ccc(1, taps, 1);
>
> // Filter the input samples
> filter->filter(sampleSize, input, output);
>
> // Free up the malloc'd input and output storage
> free(input);
> free(output);
>
> delete filter;
>
> return 0;
> }
>
> [/code]
>
>
___
Discuss-gnuradio mailing list
Discuss-gnuradio@gnu.org
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio
