Dirac came from a later generation than Einstein. For example during Einstein's annus mirabilis (1905) I doubt very many physicists knew of tensor calculus considering it was only developed a decade beforehand. During that time the mathematics physicists learnt was very much the traditional topics of analysis: series expansions, special functions, integral equations, calculus of variations, quadratic forms and of course partial differential equations. Hilbert himself actually wrote a textbook with Richard Courant covering these topics: Methods of Mathematical Physics. As such I don't really think many physicists of Einstein's generation knew lots of mathematics simply because it wasn't taught at the time. In the decades since physicists gradually has generally become more "mathematised" so to speak. In Dirac's generation tensor calculus and group representation theory became important, than in general abstract algebra and functional analysis, then as differential geometry became more well developed that gets taught and closer to the modern day you can find top theoretical physics students learning all sorts of abstract mathematics like algebraic topology.