Rendering speed is a direct function on how well done is the code. Using the same algorithm the speed can be very fast or very slow depending only on how you code this algorithm.

Today with modern CPUs with SSE2 you can do four things at the same time and execute two instructions in the same CPU clock, you can add multicores that allow you to execute two, four, six,... tasks at the same time, so the overall multiplying factor can become very big and you even can add the GPU processing power.

But to achieve this formidable CPU potential is not an easy task, you must have total control over the code you create and this only can be done with Assembly language and here comes the problem. Nobody beside video card and firmware makers use Assembler. You have very little control over the code created by a C compiler that also nobody uses it again, all is done with C++ where the generated code is even worst.

Stupidity also evolves!