Linpack: Benchmarking the Jedi Cluster

1
Linpack Benchmarking the Jedi Cluster

• This presentation will probably involve audience
  discussion, which will create action items. Use
  PowerPoint to keep track of these action items
  during your presentation
• In Slide Show, click on the right mouse button
• Select Meeting Minder
• Select the Action Items tab
• Type in action items as they come up
• Click OK to dismiss this box
• This will automatically create an Action Item
  slide at the end of your presentation with your
  points entered.

• Steven Moy

2
Midterm Presentation

• Project Goals / Progress
• Linpack Benchmark
• Top500 List
• Uniprocessor Performance
• Multiprocessor Performance
• Overview and Expectations for final presentation

3
Goals

• Run Linpack benchmarks on a single processor
• Scale up to a cluster of computers
• Perform Speedup analysis

4
Progress

• Completed Uniprocessor Benchmarking
• Working on compiling and configuring the
  High-Performance Linpack Benchmark

5
Linpack Benchmark

• The benchmark used in the LINPACK Benchmark is to
  solve a dense system of linear equations.
• Since the problem is very regular, the
  performance achieved is quite high, and the
  performance numbers give a good correction of
  peak performance.
• http//www.netlib.org/benchmark/top500/top500.list
  .html

6
Top500 High Performance Computers

• 1 ASCI White, SP Power3 375 MHz, 8192
  Processors -- 4938 RMax Lawrence Livermore
  National Laboratory Energy
• 5 SP Power3 375 MHz 1417 RMax, 1336
  Processors Naval Oceanographic Office
  (NAVOCEANO) Aerospace
• 331 SP Power3 375 MHz -- 75.9 RMax, 72
  Processors Georgia Institute of Technology
  Research
• Jedi Pentium III Xeon 550 Mhz, 136 Processors

7
Uniprocessor Performance Danish
8
Uniprocessor Performance - Jedi
9
Ideal Multiprocessor Performance for Danish
Cluster
10
Milestones

• March 19 Midterm Presentation
• April 30 End of Semester Presentation

11
Goals for Next Review

• End of April 2001
• Uniprocessor vs. Multiprocessor Benchmark
• Speedup Analysis showing scale from a single
  processor to an entire cluster of computers.