Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache Linköping University, IDA, ESLAB 1 System-Level Design Process Informal specification Architecture selection Modelling Functional simulation System model Formal verification System architecture Mapping Estimation Scheduling ok ok Mapped and scheduled model P1 P2 Simulation ok Formal verification Analysis Lower levels of design Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 2 Real-Time Systems Soft RTS Hard RTS Occasionally missing a deadline Missing a is not desired but accepted deadline is Analysis unaccaptable Can be based on Analysis other execution timeon models Is based the WCET Provides a feasibility Provides degree yes-no answers Focus of this Established thesis methods Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 3 Outline Stochastic execution time model Contribution Problem formulation Exact solution Approximate solution Extensions Conclusions and future work Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 4 Task Execution Time Variability Application characteristics (data dependent loops and branches) Architectural factors (pipeline hazards, cache misses) External factors (network load) Insufficient knowledge Alternative Models: Average Interval Stochastic Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 5 Why not WCET? Soft real-time applications (missing a deadline is acceptable) WCET becomes pessimistic probability density Leads to processor under-utilization fast WCET computation time slow WCET computation time Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 6 Related Work L. Abeni and G. Butazzo, “Integrating Multimedia Applications in Hard Real-Time Systems”, 1998 J. Kim and K.G. Shin, “Execution Time Analysis of Communicating Tasks in Distributed Systems”, 1996 A. Kalavade, P. Moghe, “A Tool for Performance Estimation for Networked Embedded Systems”, 1998 J. Lehoczky, “Real Time Queueing Systems”, 1996 T. Tia et al., “Probabilistic Performance Guarantee for Real-Time Tasks with Varying Computation Times”, 1995 T. Zhou et al., “A Probabilistic Performance Metric for Real-Time System Design”, 1999 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 7 Limitation of Previous Work Monoprocessor systems Particular classes of task execution time probability density functions (exponential) Discrete sets of possible execution times Particular scheduling policies (FIFO, fixed priority) Restricted application classes (independent tasks) Analysis applicable under particular circumstances (heavy traffic) Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 8 Contribution An exact method for schedulability analysis, efficiently applicable but no limited to monoprocessor systems An approximate method for schedulability analysis, trading analysis efficiency for result accuracy Both methods are applicable on systems with as unrestricted assumptions as possible Experiments and method-specific extensions Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 9 Problem Formulation (1) Input: Set of task graphs, periodic tasks, deadlines less than or equal to the periods, statically mapped Set of execution times probability density functions (continuous) Scheduling policy probab probab Designer controlled discarding (rejection) execution time Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB execution time 10 Problem Formulation (2) Output: Ratio of missed deadlines per task graph Limitations: Non-preemption 15% Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 3% 11 Outline Stochastic execution time model Contribution Problem formulation Exact solution Approximate solution Extensions Conclusions and future work Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 12 Analysis Method Relies on the analysis of the underlying stochastic process A state of the process should capture enough information to be able to generate the next states and to compute the corresponding transition probabilities Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 13 Naive Stochastic Process 0 3 5 A, 0, {B} B, t0, {} B, t1, {} B, tk, {A} B, tk+1, {A} Number of next states equals the number of possible execution times (infinitely many) Group as many states as possible in equivalent states Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 14 PMIs A PMI is delimited by the arrival times and deadlines The sorting of the tasks according to their priorities is unique inside of a PMI 0 t0 t1 3 tk tk+1 5 6 9 10 12 15 A, 0, {B} B, t0, {} B, [0, 3), B, t{} 1, {} B, B, tk, [3, {A}5), {A} B, tk+1, {A} Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 15 Stochastic Process s1 A, [0, 3), {B} s2 s4 B, [0, 3), {} -, [0, 3), {} s5 s3 0 3 execA B, [3, 5), {A} s6 A, [3, 5), {} A, [5, 6), {B} Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 0 3 5 execB 16 Analysis (1) [0, 3) [3, 5) [5, 6) [6, 9) [9, 10) [10, 12) [12, 15) Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 17 Influence of the Number of Tasks Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 18 Sliding Window Size Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 19 Influence of the Data Dependencies Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 20 Influence of the Instantiations Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 21 Outline Stochastic execution time model Contribution Problem formulation Exact solution Approximate solution Extensions Conclusions and future work Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 22 Limitations of the Exact Solution The number of states increases dramatically in the case of multiprocessor systems It has to perform as many convolutions as there exist states in the stochastic process (time) It has to store as many probability distributions as there exist states in the sliding window (memory) Approximate the ETPDFs by functions of exponential distributions A much larger Markov chain is obtained, but it requires less resources to solve Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 23 Approach Outline (2) Task graphs Modelling Approximation GSPN Coxian distribs CTMC constr. CTMC Results Analysis Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 24 Application Modelling (1) Task graphs Modelling Approximation GSPN Coxian distribs CTMC constr. CTMC Results Analysis Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 25 Application Modelling (2) A E B C F D Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 26 Application Modelling (3) A B C D F C F probab A E D B E Firing delay equals execution time firing delay Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 27 Approximation (1) Task graphs Modelling Approximation GSPN Coxian distribs CTMC constr. CTMC Results Analysis Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 28 Approximation (2) 11 (11)1 22 33 (12)2 j i i 1 G ( s) i (1 j ) s i j 1 s j i 1 r Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 29 CTMC Construction (1) Task graphs Modelling Approximation GSPN Coxian distribs CTMC constr. CTMC Results Analysis Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 30 CTMC Construction (2) X, Y X, Y X SMP Approximation of the SMP X Approximation of X Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 31 Construction of the CTMC The global generator of the Markov chain becomes then M ( Aj ) I jen ( I j ) Bi ( I j ) D j ien j i , jen j i , jen M is expressed in terms of small matrices and can be generated on the fly – memory savings Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 32 Analysis Time vs. Number of Tasks Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 33 Analysis Time vs. Number of Procs Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 34 Growth with Number of Stages Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 35 Accuracy Accuracy vs analysis complexity compared to an exact approach presented in previous work Stages 2 3 4 5 Relative error 8.7% 4.1% 1.04% 0.4% Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 36 Individual Task Periods (1) 360 120 A 2 15 9 B C 4 6 G H 3 5 J D E 60 12 I F 9 15 24 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 37 Individual Task Periods (2) Tasks 12 15 18 21 24 27 Stochastic process size Identical Individual 922.14 1440.42 2385.85 3153.47 2034.00 4059.42 14590.66 17012.80 19840.19 35362.85 42486.28 64800.19 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB Increase 56.20% 32.17% 99.57% 16.60% 78.23% 52.52% 38 Deadlines < Periods (1) Deadlines shorter than periods lead to an increase in the number of PMIs 0 0 3 2 3 4 5 6 5 6 8 9 10 12 15 9 10 11 12 14 15 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 39 Deadlines < Periods (2) Tasks 12 15 18 21 24 27 Stochastic process size d=p d<p 1440.42 3153.37 4059.42 11636.29 35142.60 35044.30 2361.38 3851.90 5794.33 24024.35 48964.80 40218.60 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB Increase 63.93% 22.14% 42.73% 106.46% 39.33% 14.76% 40 Rejection vs. Discarding (1) Tasks 12 15 18 21 24 27 Stochastic process size Discarding Rejection 2223.52 95789.23 7541.00 924548.19 4864.60 364146.60 18425.43 1855073.00 14876.16 1207253.83 55609.54 5340827.45 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB Increase 42.07 121.60 73.85 99.68 80.15 95.04 41 Rejection vs. Discarding (2) Tasks 12 15 18 21 24 27 Stochastic process size Discarding Rejection 8437.85 18291.23 27815.28 90092.47 24089.19 194300.66 158859.21 816296.36 163593.31 845778.31 223088.90 1182925.81 Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB Increase 1.16 2.23 7.06 4.13 4.17 4.30 42 Conclusions Exact solution for the schedulability analysis. Mainly applicable to monoprocessors systems Approximation approach to performance analysis of multiprocessor real-time applications Larger scale applications can be analysed due to the PMI and sliding window approache (exact solution) and due to an efficient scheme to store the underlying stochastic process (approximate solution) Provides the possibility to trade-off analysis speed and memory demand with analysis accuracy Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 43 Future Work Better support for design space exploration (more performance indicators for diagnosis) More efficient extraction of the performance indicators (exploiting symetries at the application and modelling level) Relaxation of the assumptions (inspecting different mapping possibilities) Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 44 Analysis (2) [0, 2) [2, 4) [4, 6) [6, 8) Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB 45 Stochastic Process 0 3 5 0 3 execA 0 3 5 execB 0 3 s1 A, [0, 3), {B} 0 0 z2 3 s2 3 5 z2*execB B, [0, 3), {} s3 8 s4 -, [0, 3), {} s5 B, [3, 5), {A} z3 s6 A, [3, 5), {} Schedulability Analysis of Real-Time Systems with Stochastic Task Execution Times Sorin Manolache, Linköping University, IDA, ESLAB A, [5, 6), {B} 46