You have to know assembly for it to mean too much, but it helps to understand the history, which it gives a bit of on the page. Originally x86 had 8 16-bit registers, which are like variables. So if you wanted to have nine variables, you'd have to push one to the stack in RAM. When they switched to 64-bit, they also doubled the number to 16 64-bit registers. Now intel is doubling it again to 32 64-bit registers.
They're also adding a lot more functions to their assembly, but I haven't looked at them yet and those would be harder to explain. But if the CPU implements a function directly, that saves on cycles because you don't have to call multiple other functions to do the same thing.