I would just get the Q6600 and make sure you get a G0 stepping. Getting a server part to use in a normal PC seems.. wired.
I would be surprised at 25%, AFAIK, all Conroe parts are built on the same masks, and process on the line. They are binned differently based on testing much like how speed is selected, and I believe the rest of the differences are down to microcode. I can believe the 3% cited by sjhujh though.
Exactly. ^^ inverted, what is weird about using a server part in a normal PC? The server part is the better choice, because it can take the higher temps while overclocking just as well.
I think what the article is trying to point out is that the two CPU may be exactly the same in terms of architecture, but they have been tuned differently. This means that they are suited to different tasks... So for the average home user the Q6600 would be the better choice over the Xenon because the chip has been tuned to perform the functions that a desktop user would want better...
well, the tuning would have to be after manufacturing. They can't change transistors on a manufacturing scale, so that means software, like microcode. The BIOS checks the CPUID info and selects the microcode patch accordingly, so the 2 CPUs could be the same base CPUIDs, but different extended IDs indicating so the Xeon can get the right tweaks at boot time. The chip could be tweaked even more later via updated microcode in a BIOS update.
the article said it's different prefetchers....... i've never trusted CustomPC to give real, accurate technical information. i'd say they meant the way cache is configured. cache prefetching is a hardware optimisation, and is done during design phrase. perhaps the have different algorithms for generating TLB and some other optimisations. Q6600 G0 is still the best one to get, IMHO