SDF3
Scenario-Aware Dataflow
 
 
  • SDF3/SADF
  • Documentation
  • Examples
  • Download
  • Support
  • Publications
  • Sitemap
 
 

Examples

The following tables include some synthetic and real-life SADF graphs that can be used for benchmarking purposes. The colouring of the graph name indicates feasibility of exact state-space based analysis on a P4 with 3GB of usable memory. Exact analysis of green graphs is feasible for all performance metrics. For orange graphs, exact analysis of some performance metrics is feasible, while for other metrics, exact analysis suffers too much from state-space explosion. These orange graphs can still be analysed using simulation.

Synthetic SADF Graphs

Graph Name Download Description
example1 xml / jpg Example with 4 kernels and 1 detector operating in 2 scenarios with (real-valued) execution time distributions
example2 xml / jpg Example with 2 kernels and 1 detector operating in 2 scenarios with (integer-valued) execution time distributions
example3 xml / jpg Example with 1 kernel and 2 detectors operating in 2 scenarios and 2 hierarchically controlled scenarios with fixed (integer-valued) execution times
sdf xml / jpg Example with 3 kernels operating in a default scenario with fixed (integer-valued) execution times (Resembles in fact an SDF graph)
extended_sdf xml / jpg Example with 3 kernels operating in a default scenario with (integer-valued) execution time distributions
problem13a xml / jpg Example with 2 kernels operating in a default scenario with (integer-valued) execution time distributions. This is problem 13 (case 1) for the MoCC Workshop in 2005
problem18 xml / jpg Example with 7 kernels operating in a default scenario with (integer-valued) execution time distributions. This is problem 18 for the MoCC Workshop in 2005
problem19 xml / jpg Example with 3 kernels and 1 detector operating in 3 scenarios with (real-valued) execution time distributions. This is problem 19 for the MoCC Workshop in 2005

Real-Life SADF Graphs

The underlined examples indicate the original SADF graph of the modelled application. The other listed examples are derived from those original models.

Graph Name Download Description
mpeg4_avc xml / jpg MPEG-4 AVC Decoder with 4 kernels operating in default scenarios with execution time distributions
mpeg4_sp xml / jpg MPEG-4 SP Decoder with 4 kernels and 1 detector operating in 9 scenarios with execution time distributions for each scenario and full pipelining
mpeg4_sp_averaged xml / jpg MPEG-4 SP Decoder with 4 kernels and 1 detector operating in 9 scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mpeg4_sp) and full pipelining
mpeg4_sp_maximized xml / jpg MPEG-4 SP Decoder with 4 kernels and 1 detector operating in 9 scenarios with fixed execution times for each scenario (equal to the maximum value of the distributions in example mpeg4_sp) and full pipelining
mpeg4_sp_averaged_pd2 xml / jpg MPEG-4 SP Decoder with 4 kernels and 1 detector operating in 9 scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mpeg4_sp) and pipelining degree of 2
mpeg4_sp_averaged_pd1 xml / jpg MPEG-4 SP Decoder with 4 kernels and 1 detector operating in 9 scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mpeg4_sp) and pipelining degree of 1
mpeg4_sp_converted_sdf xml / jpg MPEG-4 SP Decoder with 5 kernels operating in the worst-case scenario (P99) with fixed execution times (equal to the mean value of the distributions in example mpeg4_sp) and full pipelining (Resembles in fact an SDF graph)
mpeg4_sp_worst_case_sdf xml / jpg MPEG-4 SP Decoder with 5 kernels operating in the worst-case scenario (P99) with fixed execution times (equal to the maximum value of the distributions in example mpeg4_sp) and full pipelining (Resembles in fact an SDF graph)
ce_converted_csdf xml / jpg Channel Equalizer with 9 kernels and 1 detector operating in 8 scenarios with fixed execution times for each scenario and full pipelining (Resembles in fact a CSDF graph)
ce_converted_csdf_strongly_consistent xml / jpg Channel Equalizer with 9 kernels and 1 detector operating in 8 scenarios with fixed execution times for each scenario and full pipelining (Resembles in fact a CSDF graph). This SADF graph is a strongly consistent variant of the original example
mp3 xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with execution time distributions for each scenario and full pipelining
mp3_averaged xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mp3) and full pipelining
mp3_maximized xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with fixed execution times for each scenario (equal to the maximum value of the distributions in example mp3) and full pipelining
mp3_averaged_pd1 xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mp3) and pipelining degree of 1
mp3_averaged_pd2 xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mp3) and pipelining degree of 2
mp3_averaged_pd3 xml / jpg MP3 Decoder with 15 kernels and 3 detectors operating in 5 scenarios and 2 hierarchically controlled scenarios with fixed execution times for each scenario (equal to the mean value of the distributions in example mp3) and pipelining degree of 3
mp3_converted_sdf xml / jpg MP3 Decoder with 23 kernels operating in scenario M with fixed execution times (equal to the mean value of the distributions in example mp3) and full pipelining (Resembles in fact an SDF graph)
mp3_worst_case_sdf xml / jpg MP3 Decoder with 15 kernels operating in the worst-case scenario with fixed execution times (equal to the maximum value of the distributions in example mp3) and full pipelining (Resembles in fact an SDF graph)
Copyright © 2008 Electronic Systems Group. All rights reserved.
Webmaster : Content