Our dynamic fixed-point arithmetic techniques ensure accurate phasing of TX-Phase under precision constraints. We compare the phasing accuracy of TX-Phase, measured by the SER, with an alternative implementation (with standard fixed points) without our dynamic scaling techniques for fixed-point matrices and vectors. The evaluation is conducted on 424 trio samples (Chromosome 20) from the UKB data set, using reference panels of varying sizes from the remaining cohort. Markers indicate the mean SERs, and error bars represent 95% confidence intervals. We also include the accuracies of SHAPEIT4 for comparison, which are shown to be comparable with those of TX-Phase. Without our dynamic fixed-point techniques, the accuracy of TX-Phase significantly deteriorates.