Your more than welcome for the input, of course FWIW. :)
I agree on the relative _theoretical_ overclocking potential of the Celeron--it's a .25 micron part, like the >300 mhz PIIs, and since it lacks cache chips, cache performance isn't an issue (which can be a problem with highly overclocked PIIs.) Still, I can't imagaine that any _reasonable_ overclocking would come close to compensating for the lack of cache. And IMHO, no one who's at all concerned about reliability would do more than modest overclocking, if that. Also, Celeron is a newer part, which means the fabs producing it may be less likely to have achieved yields that would in practice allow much greater overclocking.
Networth's 6 month rule of thumb seems reasonable enough to me on new software. Only other point I'd add is that one should avoid piecemeal upgrades unless one has a substantial reason for doing so--something that's becoming less true as hardware power has zipped ahead of software power in recent years, and 32-bit software platforms have begun to grow stable (esp NT and some non-windows platforms.)
I think that dual processors almost never make sense for non-speciality folks. The performance gains are just too modest for the investment and compromises required. Money and tweaking can typically be better spent elsewhere.
Regards to you, and thanks for all your good work. You pointed me to the Matrox users' group, among other spots, and this thread is full of helpful resources.
Dave |
