It depends on the type of sequence you are trying to amplify (and the species). If it is a coding sequence and it is not spliced (i.e. prokaryotic or eukaryotic single exon), then cDNA from a sample where the gene is expressed can be used for template. But to be honest, why bother making RNA, reverse transcribing it and making sure the gene is expressed in the first place? Unless you have a ready cDNA sample, go with gDNA. Genomic DNA is easy to extract, and you don't even need to worry about integrity as your amplicon is so short. Chromatin can work, albeit less efficiently - but again, why? It is harder to make.