Referential compression algorithms are one of the main strategies to cope with the exponential growth of DNA data available to scientists. One of the techniques used to build a referential compression is sequence alignment, which in turn requires a lot of computing. In this article we present UdeAlignC, a fast alignment algorithm for the compression of DNA reads. We demonstrate that our algorithm is 2× faster than prominent state of the art tools while optimality is only reduced by 5.6%. We also implement a GPU-accelerated version and show local speedups of up to 12×. Source code available at https://bitbucket.org/BioMRcomp/udealignc.