Pranab-
While I wouldn't say they're totally worthless, there are so many other sources of problems out there that are better that I'd at least put McGraw-Hill at a very low priority.
Kaplan is better, Princeton Review is better,
Magoosh is better, official ETS questions are obviously the best... I can't imagine that most students would require more problems than the aforementioned 4 companies can provide.
The reason I say McGraw-Hill is so bad is that I've seen numerous problems of theirs that test concepts that aren't tested on the GRE. Or they have answer choices that are entirely unrealistic in that some of them are too obviously wrong or they are so incredibly far apart than anyone could just guess the answer by its size rather than doing any work. Many of them are solvable in one step, which is infrequently seen on the GRE. A lot of them contain mistakes or are just wrong. All of these issues can cause problems for students who are not aware that they're studying things they don't need to study, or are being convinced that the GRE's problems are easier than they really are, etc. So in a sense, doing McGraw-Hill problems could possibly make GRE students worse, rather than better. At the very least they'd be better served doing problems from other sources.
I may be making some fairly strong statements but I believe they are correct. The problem in this post is an example of what I'm talking about and I can easily be proven wrong if anyone can produce an official ETS question that requires the student to know the area of a sphere. Feel free to post it below.