On Monday 25 September 2006 21:22, M Harris wrote:
Tomorrow when I have a little more time I will measure the entire system again, this time with each machine cranking on a 20,000 digit computation of pi--- just to see what the overall load increase will be with all of the processors "hot" so to speak--- never measured that before either! Ok, again as promised, I measured my home system/network for a contrast comparison between all processors idle vs all processors pegged. First I established a baseline for the house (systems down) including my office cd amp (forgot that before). So... I removed the cd amp from the system load numbers. The entire network up and idling consumes 625 watts (see previous post for network compliment). I then shipped this command to each processor on the network to get it cranking on a computation of pi to 50,000 digits:
time echo "scale=50000; 16*a(1/5)-4*a(1/239)" |bc -lq The scale was chosen significantly large enough so that even the fastest machine on the network could in no way complete the calculation before I could take the measurement. ( to see the pi calculation complete on your own system in a reasonable time set the scale to something smaller than 1000 ). Survey says.... With all processors pegged the entire network consumes *780 watts*. So, to put this in perspective, even with all processors pegged the entire network still consumes 5% less power than my coffee maker. Over an (8) hour day the network will consume 6.2 KwHr pegged vs 5.2 KwHr idling. The coffee maker consumes about 6.6 KwHr over that same period. Moral to the story?? Well, if I give up coffee.... I will be healthier and the savings will completely pay for running my home network. ;-))) -- Kind regards, M Harris <><