The second half of this year is going to be very interesting. I think Intel management has FINALLY seen that they have to get there "stuff" together and produce a truely competitive desktop CPU. Intel was really hammered by the 90nm process. T
hey designed the P4 to scale to super high clock frequencies (and it had to in order to have good performance) and then they tried to build it on a process that had a gate leakage problem that was never fully solved. High clock + high gate leakage = Nuclear Power Plant heat levels! And lower clock frequencies than they want.
Intel has two things going for them coming out in the second half that (should) (hopefully) relevel the performance field: They claim they have "fixed" the gate leakage issue in the 65nm process. And Conroe is very much getting away from the super high clock speed design and getting back to designing for best PERFORMANCE instead of best clock frequency. And, to top it off, they have actually REDESIGNED THE FREAKING FPU!!!! The base FPU has been largely untouched since the days of the PIII. They just kept adding on new SIMD instructions (MMX, SSE, etc.) and assumed (what were they thinking???) that everyone would rewrite all the exiting code to use these new super instructions. Of course, only the apps that REALLY got a big advantage with SIMD rewrote stuff (video...). Everyone else was left wondering why the P4 sucked so bad in things like games and every day applications.
So, if they bring out Conroe at the 3+ Ghz I have heard (online, not at work) and the reworked FPU is all that it is cracked up to be, then this might be a VERY nice CPU that really brings performance back to the Intel camp. And the beautiful thing about that is it will then force AMD to jump through a few hoops to try to beat it. And WE win.
hey designed the P4 to scale to super high clock frequencies (and it had to in order to have good performance) and then they tried to build it on a process that had a gate leakage problem that was never fully solved. High clock + high gate leakage = Nuclear Power Plant heat levels! And lower clock frequencies than they want.
Intel has two things going for them coming out in the second half that (should) (hopefully) relevel the performance field: They claim they have "fixed" the gate leakage issue in the 65nm process. And Conroe is very much getting away from the super high clock speed design and getting back to designing for best PERFORMANCE instead of best clock frequency. And, to top it off, they have actually REDESIGNED THE FREAKING FPU!!!! The base FPU has been largely untouched since the days of the PIII. They just kept adding on new SIMD instructions (MMX, SSE, etc.) and assumed (what were they thinking???) that everyone would rewrite all the exiting code to use these new super instructions. Of course, only the apps that REALLY got a big advantage with SIMD rewrote stuff (video...). Everyone else was left wondering why the P4 sucked so bad in things like games and every day applications.
So, if they bring out Conroe at the 3+ Ghz I have heard (online, not at work) and the reworked FPU is all that it is cracked up to be, then this might be a VERY nice CPU that really brings performance back to the Intel camp. And the beautiful thing about that is it will then force AMD to jump through a few hoops to try to beat it. And WE win.