One of the largest biometric databases in Chile is fingerprints, which is also the most widely used biometric in the national context. In this sense, it is relevant to have efficient algorithms in execution time on this biometrics. In the current context of High Performance Computing, there are co-processors, such as GPUs (Graphic Processing Units) capable of accelerating algorithms efficiently with low monetary cost, low electrical cost and physical space. In this work, we propose the acceleration of a biometric fingerprint algorithm on a multi-GPU hardware platform.