Too few actually get my true point. Perhaps I was writing too long posts. I am sorry if this happend.
To try to analyze what is going on, I am taking the Athlon 64 as an example for some technical considerations: http://www.hardware.fr/articles/478/page1.html

My conclusions (some might be right only in theory) on what we have at this moment are (regarding more or less the Athlon 64 situation):

1) CPU's rely heavily on Mhz for boosting performance, more than a dozen other things like Cache, FPU, extensions. Raising the frequency is easyer than doing other changes, so it is mostly used to increase performance. Athlon was the faster than P3 when using FPU operations, but overall the difference was not "amazing". Of course, extensions help, some are very important (SSE), but usually you can't get more than 5-10% increase in performance.

Eg: AMD 3DNow! Quake is boosting nearly 5fps more than the normal Quake (when you get on a 450 AMD K6-2 Mhz system with a Voodoo Banshee board some 22 fps in 1024x768x16 running in OpenGL (glide in AMD Quake didn't seem to work properly)). Other games (Warzone) show the same situation: 5fps. Actually you won't get more than a 5fps increase when on 640x480x16, in both situations. You can conclude that it is a fixed performance increase, somewhere in the range of 10-20%. And this is very good situation compared to average performance gains!

2) At the moment you can't adapt the 32 bit addressing space to fit 2x16 bit wide data or instructions, or 4x8 bit. If this would be possible the performance would be greatly enhanced, as a good percentage of programs use rarely 32 bit. Anyway, implementing this arhitecture WOULD COST QUITE A LOT.

3) As a series of tests (if you can understand french is even better): http://www.hardware.fr/articles/478/page1.html
... concluded that the actual difference is very, very small with 32 oposing 64 bit. Some might say that Athlon 64 is not purely 64 bit (at least for now), but anyway...

Solely on the 64 bit future:

It will help a lot, it could bring serious benefits, however, for the moment we can't judge correctly what we will obtain in the future (as it was the same situation in 1985 when the 32 bit 386 was released by Intel), around 1996 we were using properly and efficiently enough the 32 bit arhitecture.

I would say that until 2005 we would not see anything surprising, but, overall, the technology is VERY PROMISING.

Saying this, the 1-5% increase in performance is not so bad, given the fact that 100% 64 bit environment don't exist at the moment. Even for 32 bit, the newest platform even now is not 100% 32 Bit, even with Windows XP (i would say 90-95% 32 bit).

But, overall, the Athlon 64 is a big step forward, also including the very good HT technology, doesn't matter if Athlon 64 would be remembered as the 32 bit 386 was, it would still be a big step forward.

Note: the first part (with conclusions) points to what we have at the moment, as opposed to what it can be, thus it is clearly in contrast with what it could be in the future (like this solely 64 bit future section).


I appreciate very much the persons that would have the time to read this, and again I will thank the persons that will let this thread undeleted, even if it is quite long, and a bit confussing.

Thank you.