Re: Extremely poor performance crunching random numbers under PIV-FC5

[Chris Jones <jonesc hep phy cam ac uk>]

Hi,

I'm no expert, but I think there are different random number 
implementations around. For instance, one "truly" random one basis its 
numbers on events such as network traffic or "user inputs". This is 
truly random, but does require some time to gain the "entropy" needed to 
create a random number.

I don't know why, but maybe the -static option somehow changes the 
underlying random number generator used ?

A couple things you could try -

I see you use the time to initialize the random generator - Maybe you 
could try hard coding this, so you get the same set of random numbers 
each run - Then, see if you get different numbers with the dynamic and 
static versions.

If you run "top" whilst running the apps, is the cpu usage 100% each 
time ? If with the dynamic it is much less, it would  suggest the time 
is not lost in the cpu but elsewhere (like waiting for random 
network/user events ?).

cheers Chris