Abstract Background DNA marker profiles play a crucial role in the identification and registration of germplasm, as well as in the distinctness, uniformity, and stability (DUS) testing of new plant variety protection. However, selecting minimal marker sets from large-scale SNP dataset can be challenging to distinguish a maximum number of samples. Results: Here, we developed the CoreSNP pipeline using a “divide and conquer” strategy and a “greedy” algorithm. The pipeline offers adjustable parameters to guarantee the distinction of each sample pair with at least two markers. Additionally, it allows datasets with missing loci as input. The pipeline was tested in barley, soybean, wheat, rice and maize. A few dozen of core SNPs were efficiently selected in different crops with SNP array, GBS, and WGS dataset, which can differentiate thousands of individual samples. The core SNPs were distributed across all chromosomes, exhibiting lower pairwise linkage disequilibrium (LD) and higher polymorphism information content (PIC) and minor allele frequencies (MAF). It was shown that both the genetic diversity of the population and the characteristics of the original dataset can significantly influence the number of core markers. In addition, the core SNPs capture a certain level of the original population structure. Conclusions CoreSNP is an efficiency way of core marker sets selection based on Genome-wide SNP datasets of crops. Combined with low-density SNP chip or genotyping technologies, it can be a cost-effective way to simplify and expedite the evaluation of genetic resources and differentiate different crop varieties. This tool is expected to have great application prospects in the rapid comparison of germplasm and intellectual property protection of new varieties.