the uniqueness of Golomb
Unlike RC5-64, each packet
is given out twice to two unique users. When the packets are returned, the server
compares both packets to ensure the validity of the tests. Also different from
RC5-64 and most other distributed computing projects is that the size of the
packet given to each client is not proportional to the client's computer speed.
Someone running a regular Pentium could get a larger stub to process, while
someone with a Pentium III processor could get a smaller stub. Distributed.net
does, however, plan to make the size of the packet proportional to the computer
speed, but it is difficult to judge what size stub will take a longer time to
process as there are no accurate predictors. Distributed.net is currently compiling
a database of nodes and times to create the predictor needed.
OGR is unique from other
distributed computing projects in other ways as well. Distributed.net's OGR
project may have end results similar to the first OGR project: that the existing
known rulers are the most optimal. Because, unlike the Great Internet Mersenne
Prime Search and Pihex, the OGR project isn't about searching for larger numbers
(rulers). It is about trying to discover new smaller, optimal rulers while finding
the most optimal 24-mark ruler.
The OGR project will terminate
when the 24-mark ruler is found. Whether new optimal rulers of smaller marks
will be found along the way, or if a new 24-mark ruler will be the same as the
old one, remains to be seen. If this search is successful and the 24-mark optimal
Golomb is different, Distributed.net might plan an OGR-25 search. If not, they
will at least know that the existing 24-mark ruler is correct.
©2000 Team DC (Thinkquest Team C007645). Hosted by ThinkQuest.