wrench::WfCommonsWorkflowParser
-
class WfCommonsWorkflowParser
A class that implement methods to read workflow files provided by the WfCommons project.
Public Static Functions
-
static std::shared_ptr<Workflow> createExecutableWorkflowFromJSON(const std::string &filename, const std::string &reference_flop_rate, bool redundant_dependencies = false, bool ignore_cycle_creating_dependencies = false, unsigned long min_cores_per_task = 1, unsigned long max_cores_per_task = 1, bool enforce_num_cores = false)
Create an NON-abstract workflow based on a JSON file.
- Parameters
filename – the path to the JSON file
reference_flop_rate – a reference compute speed (in flops/sec), assuming a task’s computation is purely flops. This is needed because JSON files specify task execution times in seconds, but the WRENCH simulation needs some notion of “amount of computation” to apply reasonable scaling. (Because the XML platform description specifies host compute speeds in flops/sec). The times in the JSON file are thus assumed to be obtained on an machine with flop rate reference_flop_rate.
redundant_dependencies – Workflows provided by WfCommons sometimes include control/data dependencies between tasks that are already induced by other control/data dependencies (i.e., they correspond to transitive closures or existing edges in the workflow graphs). Passing redundant_dependencies=true force these “redundant” dependencies to be added as edges in the workflow. Passing redundant_dependencies=false will ignore these “redundant” dependencies. Most users would likely pass “false”.
ignore_cycle_creating_dependencies – if true, simply ignore dependencies that would make the workflow graph acyclic. If false, throw an exception if the workflow graph would be made acyclic by adding a dependency.
min_cores_per_task – If the JSON file does not specify a number of cores for a task, the minimum number of cores on which the task can run is set to this value. (default is 1)
max_cores_per_task – If the JSON file does not specify a number of cores for a task, the maximum number of cores on which the task can run is set to this value. (default is 1)
enforce_num_cores – Use the min_cores_per_task and max_cores_per_task values even if the JSON file specifies a number of cores for a task. (default is false)
- Throws
std::invalid_argument –
- Returns
a workflow
-
static std::shared_ptr<Workflow> createWorkflowFromJSON(const std::string &filename, const std::string &reference_flop_rate, bool redundant_dependencies = false, bool ignore_cycle_creating_dependencies = false, unsigned long min_cores_per_task = 1, unsigned long max_cores_per_task = 1, bool enforce_num_cores = false)
Create an abstract workflow based on a JSON file.
- Parameters
filename – the path to the JSON file
reference_flop_rate – a reference compute speed (in flops/sec), assuming a task’s computation is purely flops. This is needed because JSON files specify task execution times in seconds, but the WRENCH simulation needs some notion of “amount of computation” to apply reasonable scaling. (Because the XML platform description specifies host compute speeds in flops/sec). The times in the JSON file are thus assumed to be obtained on an machine with flop rate reference_flop_rate. NOTE: This is only used if the JSON file does not provide information regarding the machine on which a task was executed. In this case, the machine speed information is used.
redundant_dependencies – Workflows provided by WfCommons sometimes include control/data dependencies between tasks that are already induced by other control/data dependencies (i.e., they correspond to transitive closures or existing edges in the workflow graphs). Passing redundant_dependencies=true force these “redundant” dependencies to be added as edges in the workflow. Passing redundant_dependencies=false will ignore these “redundant” dependencies. Most users would likely pass “false”.
ignore_cycle_creating_dependencies – if true, simply ignore dependencies that would make the workflow graph acyclic. If false, throw an exception if the workflow graph would be made acyclic by adding a dependency.
min_cores_per_task – If the JSON file does not specify a number of cores for a task, the minimum number of cores on which the task can run is set to this value. (default is 1)
max_cores_per_task – If the JSON file does not specify a number of cores for a task, the maximum number of cores on which the task can run is set to this value. (default is 1)
enforce_num_cores – Use the min_cores_per_task and max_cores_per_task values even if the JSON file specifies a number of cores for a task. (default is false)
- Throws
std::invalid_argument –
- Returns
a workflow
-
static std::shared_ptr<Workflow> createExecutableWorkflowFromJSON(const std::string &filename, const std::string &reference_flop_rate, bool redundant_dependencies = false, bool ignore_cycle_creating_dependencies = false, unsigned long min_cores_per_task = 1, unsigned long max_cores_per_task = 1, bool enforce_num_cores = false)