CPU Benchmarks.
June 24, 2003, 05:06 pmI can recall back in the early nineties when the Intel Pentium processors first came out. I would think to myself, “I wonder which is faster: a 486 100MHz or a Pentium 75MHz?” Of course, I was way off. Clock speed does not directly mean performance. The re-engineered processor core of a Pentium would likely outperform a faster clock speed previous generation processor in any practical evaluation.
One of my friends in high school would talk about how he wanted to get some Sun box (whose processors at the time had clock speeds of 233 - 300 MHz (I think?) – more than twice that of available x86 processors), install Windows NT on it, and use it to play id software's Doom. I told him that if he did that, he wouldn't get better performance. I seriously doubt that Windows NT at the time ran better on Sun hardware than on x86 hardware; I don't even want to talk about how much Windows NT sucked for games. In addition, I reiterated that comparing clock speeds – especially when comparing CISC & RISC processors – is worthless.
And now Apple is making some pretty big claims about their newest G5 processor. People are providing evidence that Apple skewed the benchmark results in their favor, pulling tricks like compiling the benchmark with custom, highly optimized versions of gcc while disabling performance enhancing features like hyperthreading for their Intel P4/Xeon comparisons. Lame.
I don't think people use macs because they think they're the fastest, most powerful computers. They have nice (overly priced) hardware with a nice operating system. I wouldn't mind trying one. But this whole PC-supercomputer thing is just lame and bullshit.
And don't get me started on this whole 64-bit crap. Unfortunately, we can blame SEGA for starting all that with their Genesis advertisements claiming 16-bit superiority over the less 8-bit NES.