LLVM: llvm::ARMOverrideBypasses Class Reference (original) (raw)

Post-process the DAG to create cluster edges between instrs that may be fused by the processor into a single operation. More...

#include "[Target/ARM/ARMLatencyMutations.h](ARMLatencyMutations%5F8h%5Fsource.html)"

Post-process the DAG to create cluster edges between instrs that may be fused by the processor into a single operation.

Definition at line 27 of file ARMLatencyMutations.h.

◆ apply()

void llvm::ARMOverrideBypasses::apply ( ScheduleDAGInstrs * DAGInstrs) overridevirtual

◆ makeBundleAssumptions()

unsigned llvm::ARMOverrideBypasses::makeBundleAssumptions ( SUnit & ISU, SDep & Dep ) protected

◆ memoryRAWHazard()

◆ setBidirLatencies()

void llvm::ARMOverrideBypasses::setBidirLatencies ( SUnit & SrcSU, SDep & SrcDep, unsigned latency ) staticprotected

◆ zeroOutputDependences()

bool llvm::ARMOverrideBypasses::zeroOutputDependences ( SUnit & ISU, SDep & Dep ) staticprotected

◆ AA

◆ DAG

◆ TII

The documentation for this class was generated from the following files: