As with many textbooks on science, I use this book as a tool (in this case a borrowed tool), not reading the book from cover to cover, but rather using it to read up on whatever part of it I happen to need at the moment. I have certainly not read all parts of the book, but find those parts I have read a nice mathematically inclined description of the central conepts and methods of numerical analysis. The text's focus is very much on the what and why theoretical aspects of numerical analysis, rather than on the how-to aspects of numerical methods in practice.
The book does unfortunately suffer from a lay-out style that seems to have become popular with text books in the mathematical sciences the last fifteen or so years. While the pages certainly look, stylish, clean and modern, I find myself having difficulties trying to get an immediate overview of what is main text, suplementary text, definitions, theorems, examples etcetera.
Once we are past the lay-out issues, the text gives a thorough introduction to the mathematical aspects of numerical problem solving. Despite its thickness, the book does not waste space by endless babbling the way many American textbooks tend to do, but gives a fairly succinct account for its material, while giving ample space to pseudo-code descriptions of the numerical algorithms.
For practical use when working with numerical scienctific computing, this book does need to be complemented with a more hands-on, how-to-do-it book of numerical methods, but my impression currently is that it serves its purpose as a readable text on the underlying theoretical aspects.