- Nov 22, 2019
-
-
Alexander Diewald authored
Feature dependencies See merge request !18
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Nov 19, 2019
-
-
Alexander Diewald authored
* Previous implementation was broken and only worked for the mapping + scheduling configuration. * Produced types are no longer explicitly exposed by D3SEModules as they would bind to the inputs of these modules. Instead a D3SEInjector wrapper is responsible for obtaining the Feature-annotated artifacts. * Inputs of D3SEModules don't need to specify @Genotyped or @InputArtifact annotations anymore that cause a fix type binding and conflict with the dynamic feature selection. Instead, these annotations are bound by the configuration phase. * Creates and Decodes annotations are removed as they are no longer needed. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Nov 01, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Oct 18, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
D3SE: Enable dependencies between exploration features See merge request !17
-
- Oct 15, 2019
-
-
Alexander Diewald authored
* IExplorationFeatures can now declare their predecessors and successors by priorTo() and after() statements. * All creator and decoder modules are now d3se modules that are private guice modules. * The d3se modules mark every generated output with their containing features as an annotation. This allows the same type to be provided by different modules. The coupling of feature-annotated artifacts is done by a feature dependency graph and the module config phase from guice. * Within d3se modules, only the artifacts are visible (no need to reason about their origin --> handled when declaring the feature). * IExplorationFeatures are pluggable and can be en-/disabled subject to a system's needs. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 19, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 18, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 17, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
* Drop the use of entryXYZ types. * Mapping tables now support m:n mappings, but no more multiple instances per allocation. This will be handled in a separate step that will be performed before the mapping. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 15, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 13, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 12, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
* NOTE: These constraints are disabled by default. They are too strict since they enforce equal start/end times between memory fetching and execution tasks. * Proper constraints for this execution model would require considering the size of caches and the size of fetched data to avoid cache over- flows. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
If a task hosts strongly causal components we may not define causality on the task itself, but its write task. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 11, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
* Import the AllocationToSchedule generator for Schedule skeletons and adjust it for the needs of the SMTStrictTTDecoder. * Fix the signal & route handling in the SMTStrictTTDecoder. Wrong artifacts have been passed around method calls. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 10, 2019
-
-
Alexander Diewald authored
* Instead a zero duration is assumed for this degenerate case. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
* Remove the unneeded setAssignedResources method (avoid code duplicates). * Move the task replica validation to the encoding. * Fix the crossover operation. * Fix the removal of Mapping Entries in the MappingEncoding. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 09, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 08, 2019
-
-
Alexander Diewald authored
* Create an intersects method for N sets. * Respect allocation constraints in the allocation of the first member of a coupled set. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 07, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
* Adjust the slot number validation (all tasks are scheduled) for the new IScheduledTask representation. * Switch the StrictTTSchedule to operate on ISchedulableTasks. * Rename the schedulable entities: *Schedule* -> *Schedulable*. * Add a debug method to the EDF scheduler that prints unscheduled tasks whose predecessors are scheduled. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 06, 2019
-
-
Alexander Diewald authored
* Correctly adjust the ready time of tasks to the finish time of their predecessors. Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
- Sep 05, 2019
-
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-
Alexander Diewald authored
Signed-off-by:
Alexander Diewald <diewald@fortiss.org>
-