1 Homework Assignment 1
1.1 Task
The goal is to determine the number of dual CPU compute nodes necessary to accom-
plish an analysis task in the span of one month.
1.2 Background
The SETI@Home project sends a block of data approximately 600K in size to individual
users for processing. Upon completing the computation (with no communication with
any other system or server) about 4.5 hours later using a Pentium4 Xeon 2.4GHz
processor, the (extremely small) results are then sent back to the SETI@Home servers.
Upon successful upload of results, a new block of data is retrieved by the client.
1.3 Problem
You’ve been awarded a grant to buy hardware to support the SETI@Home project.
The grant requires that you process 100,000 blocks of SETI@Home data every month.
Unlike traditional grants, this grant will allow you to spend as much money as you
need to meet that computational goal. You must also make sure that your network
connection is good enough to support this data load.
1.4 Questions to Answer
1. How long would it take one machine to process 100,000 blocks of data?
2. Using our standard x335, dual processor, Pentium4 Xeon 2.4GHz systems, how many
nodes (minimum) would you need to process all the data in one month?
3. Using the worst case scenario approach1, how much bandwidth (MegaBytes/second) is
the network required to provide in order to support this analysis? Assume a download
rate of 600K/second.
4. Assuming a factor of 10 conversion between MegaBytes/second and Megabits/second,
is gigabit ethernet connectivity to the SETI@Home servers good enough?
1All machines request a block of data at the same time
1