Performance Evaluation of P2P and Cloud Computing Applications - - - PowerPoint PPT Presentation

▶
performance evaluation of p2p and cloud computing
SMART_READER_LITE
LIVE PREVIEW

Performance Evaluation of P2P and Cloud Computing Applications - - - PowerPoint PPT Presentation

Nov 17, 2022 38 likes •291 views

Performance Evaluation of P2P and Cloud Computing Applications - A Module for SimGRID Bogdan Cornea, Julien Bourgeois, Vaidy Sunderam 14 June 2012, Valpr, Ecully Origins Performance prediction dPerf dPerf and P2P apps.

slide-1
SLIDE 1

Performance Evaluation

  • f

P2P and Cloud Computing Applications

  • A Module for SimGRID

Bogdan Cornea, Julien Bourgeois, Vaidy Sunderam 14 June 2012, Valpré, Ecully

slide-2
SLIDE 2

2/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-3
SLIDE 3

3/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-4
SLIDE 4

4/25

Origins of an idea

  • Alternative to grids and clusters

– Computing

  • Infinite ressources of P2P
  • Lower cost
  • Existence of

– ChronosMix tool – Real computing applications – Budget

  • P2P computing frame
  • P2P performance prediction tool
slide-5
SLIDE 5

5/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-6
SLIDE 6

6/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-7
SLIDE 7

7/25

dPerf for performance prediction

Input code C, C++, Fortran / P2P-SAP, MPI Prediction ROSE compiler HW counters PAPI SimGrid MSG Trace replay Network scaling Scaling the N° nodes Scaling

  • f

N° nodes Automatic code translator

slide-8
SLIDE 8

8/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-9
SLIDE 9

9/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-10
SLIDE 10

10/25

dPerf and P2P apps.

  • Target: P2P computing

– Heterogeneous – P2P communication protocol

  • Adapted to P2P performance computing
  • Developed at LAAS (France)

– Real code ported to P2P computing

  • By LAAS (France)

– Scaling

slide-11
SLIDE 11

11/25

dPerf and P2P apps.

  • How ?

– Automatic instrumentation + execution

  • Simple block benchmarking
  • Take in account compiler optimization

Procs.

Time [s] NAS IS ; Gcc optimization level : 0.

slide-12
SLIDE 12

12/25

dPerf and P2P apps.

  • How ?
  • optimized block benchmarking***

Procs.

Time [s] NAS IS ; Gcc optimization level : 0.

slide-13
SLIDE 13

13/25

dPerf and P2P apps.

  • How ?

– Scaling of results

  • Trace-based simulation with SimGrid MSG

– Network configurations

Procs.

Time [s]

Obstacle problem ; P2P implementation Gcc optimization level : 0.

slide-14
SLIDE 14

14/25

dPerf and P2P apps.

  • How ?

– Scaling of results

  • Trace-based simulation with SimGrid MSG

– Number of nodes

slide-15
SLIDE 15

15/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-16
SLIDE 16

16/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-17
SLIDE 17

17/25

Module for SimGrid

  • Initially
  • After integration

Input code dPerf SimGrid Input code SimGrid dPerf MSG SMPI

slide-18
SLIDE 18

18/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-19
SLIDE 19

19/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-20
SLIDE 20

20/25

dPerf for Cloud apps.

  • Ongoing
  • Many perspectives
slide-21
SLIDE 21

21/25

dPerf for Cloud apps.

  • Phase 1

MPI HPC code dPerf adapter Static analysis, Static communication topology identification Cloud Emory University MPI to any Cloud

slide-22
SLIDE 22

22/25

dPerf for Cloud apps.

  • Phase 2

– Performance

prediction of applications on Cloud

– Fine tune dPerf – Compare original

performance to adapted performance

Cloud performance prediction MPI HPC code SimGrid dPerf MSG SMPI adapter tuning

slide-23
SLIDE 23

23/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-24
SLIDE 24

24/25

  • Origins
  • Performance prediction

– dPerf

  • dPerf and P2P apps.
  • Module for SimGrid
  • dPerf for Cloud apps.
  • Future work
slide-25
SLIDE 25

25/25

Future work

  • Support for C++, Fortran
  • Multi-core
  • Memory
  • SMPI ? dPerf ? MSG

– Helps the integration process

  • Compare to other tools based on MSG