Evaluating Node Orderings for Improved Compactness Carl Albing, - - PowerPoint PPT Presentation

▶
evaluating node orderings for improved compactness
SMART_READER_LITE
LIVE PREVIEW

Evaluating Node Orderings for Improved Compactness Carl Albing, - - PowerPoint PPT Presentation

Nov 13, 2023 110 likes •363 views

6th International Workshop in Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS'15) Evaluating Node Orderings for Improved Compactness Carl Albing, Ph.D. Dep't. of Computer Science U.S. Naval

slide-1
SLIDE 1

Evaluating Node Orderings for Improved Compactness

Carl Albing, Ph.D. Dep't. of Computer Science U.S. Naval Academy. 6th International Workshop in Performance Modeling, Benchmarking and Simulation of High Performance Computer Systems (PMBS'15)

slide-2
SLIDE 2

Topics

  • Appl Placement

Node Ordering →

  • Metrics for Placement

MIND →

  • Mean Placement Calculation (MPC)
  • Graphing the MPC
  • Interpreting the Results
  • A Reality Check
slide-3
SLIDE 3

Acknowledgements

This work was supported in part by the DoD HPC Modernization Program. It also would not have been possible without the infrastructure support and help from the accommodating staff at ORNL, NERSC, and NOAA.

slide-4
SLIDE 4

Application Placement

  • Performance variations tied to placement
  • Causes related to:

– latency (distances) – bandwidth (job/job inter.)

  • Compact placement

– better for both

  • Utilization suffers
slide-5
SLIDE 5

Node Ordering

  • General Concept
  • Appl Placement
  • 3D Torus
slide-6
SLIDE 6

Example Node Orderings

  • Min dimension first vs. Max

– torodial vs. snake

1 4 7 10 8 5 2 3 6 9 11 12 15 14 13 1 6 7 12 8 5 2 3 4 9 11 10 15 14 13 1 2 3 4 5 6 7 10 8 9 11 12 15 14 13

slide-7
SLIDE 7

Example Node Orderings

  • nid #
  • 2Cube, BBB, Hilbert
slide-8
SLIDE 8

Node Ordering

  • Many orderings
  • How to choose?
  • Benchmarks are expensive
  • Convert the whole system?
  • Any theoretical or empirical help?
slide-9
SLIDE 9

Metrics

  • For a single placement:

– Diameter

  • r

– Mean Inter-Node Distance

  • r

– Hop-Bytes

slide-10
SLIDE 10

Metrics

  • Diameter

– which is more compact?

slide-11
SLIDE 11

Metrics

  • Mean Inter-Node Distance (MIND)

· Hop-Bytes – not enough data

slide-12
SLIDE 12

Mean Placement Calculation

  • For a Given Job Size – place it starting

at 1st location in list

  • Calculate MIND for job
  • Re-place, re-MIND at each loc.
  • Take the Mean
  • That's 1 value
  • Next size
slide-13
SLIDE 13

Using MPC

  • Each MPC Yields a Curve
  • Specific to a Machine Size/Shape
  • Three real systems:

site X Y Z nodes NOAA 7 12 16 1316 NERSC 17 8 24 6528 ORNL 25 32 24 18944

slide-14
SLIDE 14

Results

  • Three Systems

– “small”, “med”, large HPC

  • Use in real placement sequence
slide-15
SLIDE 15

Comparing Node Orderings

slide-16
SLIDE 16

A Closer Look

slide-17
SLIDE 17

“Medium” Size HPC System

slide-18
SLIDE 18

MPC Over the Full Range

slide-19
SLIDE 19

Large HPC System

slide-20
SLIDE 20

Large System – Full Range

slide-21
SLIDE 21

Testing the Results

  • Real World?
  • NERSC system
  • 1 month of placements
  • Mean per job size; > 5 jobs
slide-22
SLIDE 22

Theoretical - MPC

slide-23
SLIDE 23

Actual Jobs, Theoretical Placement

slide-24
SLIDE 24

Summary

  • Node Ordering, MIND
  • Mean Placement Calculation (MPC)

– System Size, “Shape” – Orderings favor Job Sizes

  • Rough predictor
  • Larger data set coming
  • Other aspects of placement