Abstract: Methods and structures are disclosed for using a graphics processing unit (GPU) to detect and organize pulses in an efficient, parallel manner. A received signal is divided into a plurality of sub-batches. Each of the plurality of sub-batches is processed in parallel by detecting a number of pulses present within each of the plurality of sub-batches. A scanning algorithm process for combining the pulse information of the detected pulses includes providing pulse information from detected pulses within a plurality of subgroups with each subgroup including 2n sub-batches, with n equal to a number of iterations of performing the providing, and repeating the iterations, increasing n by 1 for each iteration, until 2n is equal to the total number of sub-batches in the plurality of sub-batches. An output array is generated having a size based on a total combined pulse count and including sequential pulses from the plurality of sub-batches.