Has anyone used singular spectrum analysis before? How does it compare to PCA?

I have some time series data in the form of (x1, x2, ..., xn, y) for each example, and I am wondering whether SSA would be good to use as a pre-processing step. I don't know very much about it yet, so I am wondering if anyone here has any experience using it and how that worked out.

Does SSA get computationally expensive for large amounts of data? Could I use multivariate SSA here for better accuracy?