Well unfortunately at a certain point floating point gets pushed aside by bus speed. Bus speed is often much much much lower than your processors speed and thus floating point operation. It's tough trying to think about how a whole system works and not just focusing on one component. Sure your processor can handle pushing Xmb/sec. but if your system bus can't deliver that much then your gong to see your physical limitation. So once you've got enough processing power and enough ram then you just have to worry about getting information from ram to processor and back again. Think of your system bus like a hose running between your ram(pipes under the house) and your processor(your swimming pool your trying to fill). The bigger the hose the faster your going to fill the pool. That's a real simplistic way to think about it. And then you also have to figure it's a loop between your processor and your ram so your talking effectively half the data rate. Wonderful your system bus can transfere 4.2Gb/sec but you have half of that heading towards your processor and half heading away. Your bus speed is then 2.1Gb/sec. Thus a pentium4 with the new 4x ddr is out performing athlons in 2x and 3x ddrs in rendering benchmarks. Not so much that the new p4's are better they simply have more bus bandwidth. And with the mhz speeds meaning less and less why do we still measure by these retarded numbers? Why don't we go by processes/sec or some other more meaningful number?