P2P Parallel Computing Using MCMS Architecture

1
P2P Parallel Computing Using MCMS Architecture
2
MCMS Architecture
Client
Servers
Job distribution node (client node)
Computational nodes (servers)
3
Equipment Resources

• Total 14 Legacy Macs with G3 processors
• Dynamic IPs as Server nodes (13 computation node)
• 128.255.163.92 (MacOSX, G3, 350 MHz/1 MB
  cache, 128 MB, SDRAM 12 GB HD)
• 128.255.163.135 (MacOSX, G3, 350 MHz/1 MB cache,
  64 MB, SDRAM 6 GB HD)
• 128.255.163.106 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.219 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.227 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.228 (MacOSX, G3, 350 MHz/1 MB cache,
  256 MB, SDRAM 6 GB HD)
• 128.255.163.237 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.16 (MacOSX, G3, 350 MHz/1 MB
  cache, 128 MB, SDRAM 12 GB HD)
• 128.255.163.242 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.96 (MacOSX, G3, 350 MHz/1 MB cache,
  128 MB, SDRAM 12 GB HD)
• 128.255.163.133 (MacOSX, G3, 350 MHz/1 MB
  cache, 128 MB, SDRAM 12 GB HD)
• 128.255.163.136 (MacOSX, G3, 350 MHz/1 MB
  cache, 128 MB, SDRAM 12 GB HD)
• 128.255.163.141 (MacOSX, G3, 350 MHz/1 MB
  cache, 128 MB, SDRAM 12 GB HD)
• Dynamic IPs as 1 client node (job distribution
  node, or service node, which performs
  decomposition and job distribution)
• 218.255.163.101 (MacOSX, G3, 350MHz/1 MB cache,
  64MB, SDRAM 6GB HD)

4
(No Transcript)
5
Numerical Integration

• Newton-Cotes Method
• Decomposition

6
Numerical Integration

• Global assembling
• Total Global intervals
• 1 billion (light intensive computation performed
  on each computational node)
• 10 billions (intensive computation performed on
  each computational node)
• 100 billions (very intensive computation
  performed on each computational node)

7
(No Transcript)