Re: Why AMD chose low-end Intel platform for the test?
Look again at what was being tested.
They were comparing efficiency of the chips/compilers, not raw performance.
It's easier to show high efficiency on a single low clocked chip than it is on multiple, high clocked chips.
For example, Opteron can perform a maximum of 3 ADD/SUB operations per clock, while P4 can do 2.
So, at equal 50% "performance" numbers from those graphs, the 1.6ghz 3200+ would be executing 2.4 Billion OP/Sec while the 1.7ghz P4 would be executinng 1.7 Billion OP/Sec.
The quartet system (with 4 1.8ghz CPUs) would be performing 10.8 Billion OP/Sec....