Lines Matching full:scheduling

67   // Scheduling stages for this strategy.
75   // track register pressure for actual scheduling heuristics.
121 /// The goal of this scheduling strategy is to maximize kernel occupancy (i.e.
128 /// The goal of this scheduling strategy is to maximize ILP for a single wave
178   // Occupancy target at the beginning of function scheduling cycle.
244 // GCNSchedStrategy applies multiple scheduling stages to a function.
265   // Record the original order of instructions before scheduling.
268   // RP before scheduling the current region.
271   // RP after scheduling the current region.
279   // Initialize state for a scheduling stage. Returns false if the current stage
283   // Finalize state after finishing a scheduling pass on the function.
286   // Setup for scheduling a region. Returns false if the current region should
296   // Check result of scheduling.
306   // Returns true if scheduling should be reverted.
317   // Attempt to revert scheduling for this region.
351 // Retry function scheduling if we found resulting occupancy and it is
352 // lower than used for other scheduling passes. This will give more freedom