Intel is slated to release next quarter a new P4 & chipset w/ an 800mhz FSB and dual channel DDR 400 support. If you can hold off, existing P4 533mhz and 845PE chipset prices should drop by a lot. (And the new stuff will of course offer even more performance.) I mention this b/c I bought a Dell right before the 440BX chipset came out. If I had had that mobo instead of the 440LX, I could have upgraded all the way up to a P3. I know there is always reason to wait, but with such an impending release, you're looking a potentiallly huge drop in prices for the existing P4 chips.
these little things are getting hotter like little nuclear reactors. there seems no way to get them cool off physically and quietly. lots of chichi software cooling which are actually very slow.
http://www.xbitlabs.com/news/story.html?id=1046339848 http://www.extremetech.com/article2/0,3973,907059,00.asp Also, the price for the latest CPU 2.4c GHz (version is Hyperthreading CPU), 2.6C and 2.8C will be the same than the current price for these speed. The release is plan for May 11th 2003.
great tips, Hyperthreading actually works whether or not the actual software was recompiled for it. They seemed to make the actual system hyperthread capable instead of just the singular execution of an application subject to hyperthreading. So they have "cafeteria - ized" the system throughput instead of just adding more waiters for one sitting.
Intel chips are cooler than AMDs and people seem to run those okay. I'm pretty sure there a viable ways to physically cool off a P4 if you're worried.
All I can tell you is that the benchmarks I've seen for HT show it running somewhat slower on most non-multi threaded apps. The difference wasn't great. You could probably find them by doing a search on Intel Hyperthreading Review--or something along those lines. (Of course I read the first reviews and things may have changed or better ways to use the processor learned.)
Too many folks and so called experts seem to think that because the machine is hyper-thread capable that all software will get a bump in performance. Not so. If a software is designed to complete something before issuing the next instruction, it will defeat the purpose somewhat. What if the manager (software) does not understand/recognize your new capabilities to truly multi-task? Yes, you might now be able to serve all the food in the cafeteria immediately to everyone all at once. But if the bottom line instruction given is, "you can't start to eat until the guy next to you finishes..."