The most straightforward way to compare them is to find the common gene ids between them. However, you'll have to keep in mind the caveat that different probes targeting the same gene (probes from different GPL) can give different results. Given that data from different GPL for similar experiments are likely to have large batch differences (since they would be from different experiments) probe performance differences might be the least of your problems. Another way to approach things would be to get the probes behind each GPL, map them to genes yourself, and then apply the data behind each experiment to the commonality of your mappings.