450MHz-PentiumII(Xeon, dual)
onlxeon[105]_% time spread
processed 100 events 
0.370u 0.020s 0:00.38 102.6%    0+0k 0+0io 100pf+0w
->3.7msec/event

1GHz-PentiumIII(dual)
elecref[104]_% time spread
processed 100 events 
0.120u 0.000s 0:00.12 100.0%    0+0k 0+0io 94pf+0w
->1.2msec/event

733MHz-PentiumIII(single)
onlpcnf.kek.jp[111]_% time ./spread
processed 100 events 
0.220u 0.020s 0:00.23 104.3%    0+0k 0+0io 99pf+0w
->2.2msec/sevent

2GHz-Pentium4(single)
[yasu@harry sp]$ time  ./spread   
processed 100 events 
real    0m0.097s
user    0m0.090s
sys     0m0.000s
->0.9msec/event

-----------
Compileオプションを変えて2GHz-Pentium4(single)の上で走らせる
1. gcc -O -o spread spread.c -lm
real    0m0.051s
user    0m0.050s
sys     0m0.000s
->0.5msec/event

2. gcc  -finline-functions -O  -o spread spread.c -lm
real    0m0.052s
user    0m0.050s
sys     0m0.000s

3. gcc  -funroll-all-loops -O  -o spread spread.c -lm
real    0m0.046s
user    0m0.050s
sys     0m0.000s

4. gcc  -O3  -o spread spread.c -lm
real    0m0.053s
user    0m0.050s
sys     0m0.000s

結局のところ、極めてオーソドックスな-Oがよく利きました。

ちなみに、1GHz-PentiumIII(dual)で-Oを付けた場合は
elecref[106]_% time ./spread
processed 100 events 
0.070u 0.000s 0:00.07 100.0%    0+0k 0+0io 91pf+0w
->0.7msec/event
であるので、
Pentium4(2GHz)とPentiumIII(1GHz)は
P4 : PIII = 0.5 :0.7 = 5 : 7
P4はやはり速くないですね。