Computer Organization and Design: The Hardware/Software Interface, by Patterson & Hennessy, Morgan Kaufmann, 3rd Edition. Finish §4.4; §4.7-4.8.
Alternatively, Edition 1 is available on-line in draft form. Roughly speaking, you can find the equivalent material in Chapter 3. Note that Edition 1 was written before Kepler (and its predecessor, Fermi) were introduced.
Another alternative is to read Scalable Parallel Programming with CUDA, by J. Nickolls et al., ACM Queue 6(2):40-53, March/April 2008. Read through page 48 (If reading the html version, read through Related Work) Though this paper is somewhat dated (pre-Kepler and pre-Fermi), many of the discussions are relevant today.