Like other people, I remain unconvinced by his arguments. The main problem is that no one can say for certain what will actually happen, as the 64-bit version d

Like other people, I remain unconvinced by his arguments. The main problem is that no one can say for certain what will actually happen, as the 64-bit version does not exist. I also care about performance (for a living!) and I've been humbled enough times by "for sure this is what will happen" type thinking that I know it's just too hard to predict with certainty what will happen with such a complex interaction of parts. Specifically, I'm skeptical of the argument that more registers will not help, but no one can test the arguments empirically. Which is always dangerous ground. I do, however, buy people's arguments that they are feeling real pain by the memory limitations. Even if 32-bit is faster at the moment, I think it's a short-sighted decision to not port to 64-bit. First, the memory issue that everyone has brought up. Two, I find it likely that over time, using the now-standard bitedness (I know that's not a word, forgive me, I have a cold and mind is foggy) will yield benefits because it's the common path in the processor. (ygra, I know you're not necessarily in agreement with everything Rico said, so I'm not really arguing with you.)