About the performance of Intel Celeron and PIII indicated in the two-day edition I noticed an inaccuracy probably generated by the same press releases and rumors that intel puts around. If true that the Celeron has an internal BUS blocked at 66MHz also true that the 128kB of caches of which, compared to a PIII, work at double the speed (of the INTEL family the only ones to have more than 512kB of cache are only the PentiumIII Xeons). Furthermore the structure from which PIII Xeon, PIII, PII and Celeron derive the same inherited no less than from the PentiumPRO and apart from being available with separate half-speed cache (Slot1) or integrated at full speed (PPGA, FC-PGA, Soket370) the three families of processors are exactly alike also arising from the same production process. On the other hand, we must do for the set of SSE instructions on the latest generation PIIIs that are the Intel equivalent of the AltiVEC functions. So the applications written for SSE, and only those, make use of an acceleration that the Celerons do not provide (even if rumors gave the new Celerons equipped with SSE). We can say that the Celeron is to the PIII as the G3 is to the G4. In practice, on most of the entire computational tasks a Celeron and a PIII of equal clock are equivalent, on the floating point calculation instead the PIII can make use of the SSE and earn a good 30%. However, we must not forget that for the WinTEL world there are graphic cards equipped with the geometric / graphic processors (such as the Ge-Force and the Qudro by nVidia or GLINT by 3DLabs so as to mention a few) that are able to process independently from the CPU also calculations of transformation and illumination, that is precisely those that most benefit from SSE instructions. So when opening the wallet to buy a CPU, for 90% of the cases and maybe more, a Celeron that costs 1/3 of a PIII of equal clock is worthwhile. And last but not least: the performance growth curve of CPUs with CISC architecture, as are all INTELs, it flattens more and more as the architecture approaches its limit and that of the Celerons and PIIIs is now quite close. So between 566 and 600 MHz you will practically not notice any differences in performance, unlike what happens for RISC-type CPUs that have steeper curves.Logical, at this point, the conclusion. In many cases it is advisable to be wary of what is being offered to us by the marketing sector. There are many companies that know how to sell their products very well and not always what is read in the newspapers and on the websites corresponds to the reality of the facts.