I think one has to remember computers are very specialized for a specific task, Brains and CPU's are not comparable in a fundamental way, sure they both can do math, but there is a significant difference. Machines no matter what kind of equipment we give them are never truly self-aware entities, they are mathematical models programmed into their foundations of hardaware and software, even the ones that "learn" aren't truly "Aware", they are just processing meaningless data in ways we told them to interpret it. Brains are conscious living things capable of adaptation, re-orginization and self-repair, CPU's are not. CPU's also don't have to worry about their users feeding them crap components via their diet during their growth and abusing the 'processing circuitry', they also don't have to worry about temperature and the different "dirty" and unpredictable environments brains have to work in. All the materials for a CPU are 'best of breed' available at the time and handpicked by specialists, to perform the specialized tasks they were designed for. So whenever CPU's advance, just remember they didn't evolve from the original chips through a process of evolutionary re-orginization, they started over from scratch with different tools and materials every time. So in this way computing power can't be directly comparable to human brains because, the materials, tools, and designs are radically different from one another, so different they have to start over and use new blank slates rather then fixing or re-organizing the original.
A computers power is meaningless without humans to tell them what is which. To give them the proper equations and mathematics and make templates for them to make it 's ultimate output into a humanly understandable or usable form that services some purpose, design or need of ours. The design foundation or framework for a computer to do things comes from us, without us they couldn't do what they will eventually do. I'm sure sometime in the future AI and hybrid AI (biological+artificial) intelligence will be common place, each specialized to do a particular task.
It is obvious that the brain is a computational and mathematical machine, it just works in a fundamentally different manner. Language encoding and decoding is mathematical process, so even if humans are not very fast at math, their brain does do some seriously fast mathematical computations they are not consciously aware of, especially in relation to vision processing encoding and decoding of images. If you've ever been to the eye doctor, to check your eyes sometimes he squirts this liquid on your eyes which makes the bloodvessels on your eyes appear to your sight, normally your mind naturally filters these blood vessels obstructing your vision out with a mathematical algorithm so the 'artifacts' don't disrupt your normal vision function.
The brain is very good at doing very complex processing: for instance, we can make sense of the odd jumble of shapes and shadows we are looking at and recognize it as a face. We can recognize the shapes that make up a coffee mug, despite having never seen it before from that precise angle, and we can work out how far away it is and how to pick it up. We can track a person walking across a crowded room, and we can disentangle what they are saying from the hubbub around them. We can work out where a sound is coming from, despite the fact that echoes of the same sound are coming from all directions off the surrounding walls.
All these processes are complex. The tasks mentioned may seem easy to you, but that is only because you are using a frighteningly sophisticated bit of processing technology to help you do them - your brain. It would, for instance, be very difficult to program a computer to do the same things: computer scientists have made a start on this, but at the turn of the millennium, artificial seeing and hearing systems are no match for the brain of even a simple animal. This is because we don't know much about how the brain is doing it. Psychophysicists aim to find out how exactly the brain works.
Computers have finite resources just like human beings so ultimately computers can only compute (and eventually understand) what their finite capacity allows even if it eventually is much more then human beings, it is still finite and limited to the degree that nature allows.