LMS algorithms parallelization in GPGPU environment