Used to make a VdfSchedule.
More...
#include <scheduler.h>
|
| static VDF_API void | _SetScheduleValid (VdfSchedule *schedule, const VdfNetwork *network) |
| |
| static VDF_API void | _InitializeRequestMasks (const VdfRequest &request, VdfSchedule *schedule, PoolPriorityVector *poolOutputs) |
| |
| static VDF_API void | _MarkSmallSchedule (VdfSchedule *schedule) |
| | Marks the schedule as small if it is indeed small. More...
|
| |
| static VDF_API void | _ScheduleBufferPasses (const VdfRequest &request, VdfSchedule *schedule) |
| |
| static VDF_API void | _ScheduleForPassThroughs (const VdfRequest &request, VdfSchedule *schedule, const PoolPriorityVector &sortedPoolOutputs) |
| |
| static VDF_API void | _GenerateTasks (VdfSchedule *schedule, const PoolPriorityVector &sortedPoolOutputs) |
| |
| static VDF_API void | _ScheduleTaskGraph (VdfSchedule *schedule, const PoolPriorityVector &sortedPoolOutputs) |
| |
| static VDF_API void | _ComputeLockMasks (const VdfRequest &request, VdfSchedule *schedule, const PoolPriorityVector &sortedPoolOutputs) |
| |
| static VDF_API void | _ApplyAffectsMasks (VdfSchedule *schedule) |
| |
| static VDF_API bool | _ApplyAffectsMasksForNode (VdfScheduleNode *node) |
| |
| static VDF_API void | _UpdateAffectsMaskForInvocation (VdfSchedule *schedule, VdfScheduleNode *node) |
| |
| static VDF_API bool | _UpdateAffectsMasksForNode (VdfSchedule *schedule, VdfScheduleNode *node) |
| |
| static VDF_API void | _UpdateLockMaskForNode (VdfSchedule *schedule, VdfScheduleNode *node) |
| |
Used to make a VdfSchedule.
Definition at line 34 of file scheduler.h.
Applies the affects mask to the schedule.
Applies the affects masks to the scheduled node.
This mask is the ANDing of the request mask and the affects mask (if any). It is used by iterators to quickly skip only to the elements that are affected by the node. It is an optional mask, and if it doesn't exist, the scheduled affects mask will simply be request mask (a super set of the truly affected elements).
Setup the lock masks require for sparse mung buffer locking for all the outputs in the pool chain.
sortedPoolOutputs contains all the pool outputs in the schedule. Callers are responsible for ensuring that sortedPoolOutputs are in descending order of their respective pool chain index, i.e. the pool output furthest downstream will be at the front of the vector.
Generate tasks for the scheduled task graph. The task graph is used by the parallel evaluation engine.
Initializes the request masks for all the outputs that will be computed as a result of request.
poolOutputs is an output parameter. The vector will contain all the pool outputs scheduled with request masks, sorted in reverse order of pool chain index, i.e. the pool output furthest downstream will be at the front of the vector.
Marks the schedule as small if it is indeed small.
Method to schedule the buffer passes and the "keep" masks for the scheduled nodes.
Schedulers that care about performance will want to call this after all the outputs have gone through the _ScheduleOutput method above.
Schedule the outputs from which buffers should be passed. This helps ensure that potentially large portions of the network that won't have any effect in this schedule are skipped when passing buffers.
sortedPoolOutputs contains all the pool outputs in the schedule. Callers are responsible for ensuring that sortedPoolOutputs are in descending order of their respective pool chain index, i.e. the pool output furthest downstream will be at the front of the vector.
Schedule the task graph for multi-threaded munging. This will generate tasks and invocations, as well as dependencies between them.
Method to signal that a schedule is done being built and that it is now valid for the given network.
Updates the affects mask on an existing scheduled node invocation.
Updates schedule for node if affects mask changed.
Updates schedule for node if the lock mask changed.
Update schedule after the affects mask changed on output.
The documentation for this class was generated from the following file: