Okay, I'm lazy too, so out of curiosity... Where's ">1.5GHz" angle against the current state of x86, anyone got particular benchmarks?
I've been duallie with this old Pentium II machine long enough to notice what it does and doesn't buy you, and I've been keeping some small track of Apple's issues there (still undecided, but I gather the current state of the Xnu arch takes the classic approach of spreading computation across processors without worrying so much about kernel tasks... on the other hand, DF work is raising the issue of cache locality, so maybe that's not a wrong idea, while the current state of Apple does sound less-than-granular -- once a process or thread is scheduled to a CPU, it never migrates??) ...
I do get the sense that whatever the next new-and-interesting thing everyone will want to do (probably media-wise, but maybe I'm biased, MP3 playback being the last sudden new computational demand everyone everywhere now demands of any system) will be one of those things that miiight thread, but still place a huge load on a single CPU -- realtime 1080i encoding or transcoding, say, and whatever becomes the de-facto 'standard' for the new groovy thing will naturally expand to the abilities of currently-cheap x86, because that's where the "acceptable tradeoffs" live for most people. Equal chance that something structured for Cell or massive multithreading (personal search is getting hot, and the arduous process of extracting hashes or however that's going to work threads pretty well above the serialization storage implies) or whatever comes around to blow all our minds, but that's the sort of thing you might kind of solve for by dropping in a quad or 8-core when they cough them up, while assumptions of clock speed and bus bandwidth and things are what send you scrabbling for a whole new computer. (Never mind that the hottest PowerPC products aren't exactly upgradeable like that anyway..)
Or in other words, no matter how many cores you have, you'll still hit those bumps where you really wish each individual one was faster, so how're things doing there now that AMD and Intel can't ramp so easily either?
