The difference in how the brain stores information and how computers do makes it difficult to gauge the similarity. The brain is more efficient, combining memories to make recalling an entire scene easier. Sometimes, though, this doesn’t work as planned, and you’re left wondering what that word you were looking for was. Neural connection for bit, the commonly referred-to amount is 2.5 petabytes, but if neural “compression” could be simulated with computer data, we may not be as far off.
In actuality, the brain has what on a computer would be partitions. Your brain only stores so much language for easy access, which is why we often forget a word or name. The meaning of that word, though, is not forgotten, as you can often give the definition. This is because the memory of that word is in longer term memory, tucked in to another area of the brain.
Google has been able to analyze things in a similar way to the brain, taking large sets of data and seeing what a general algorithm could learn from it. The computer network running this program for a month was able to learn how to distinguish a cat on any web page. PBS Idea Channel has a great video about this here. To a human, though, of really any age, this would be a very simple task, after seeing maybe half a dozen cats and being told what they are.
My point on that last note, that this task of learning independently from a data set is easy for a human, yet difficult for modern computers, is that what it seems from the modern form of computing is that the brain has its strengths for what can be calculated, and the CPU has its own.