llvm/llvm/lib/Target/AMDGPU/GCNIterativeScheduler.cpp

//===- GCNIterativeScheduler.cpp ------------------------------------------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
///
/// \file
/// This file implements the class GCNIterativeScheduler.
///
//===----------------------------------------------------------------------===//

#include "GCNIterativeScheduler.h"
#include "GCNSchedStrategy.h"
#include "SIMachineFunctionInfo.h"

usingnamespacellvm;

#define DEBUG_TYPE …

namespace llvm {

std::vector<const SUnit *> makeMinRegSchedule(ArrayRef<const SUnit *> TopRoots,
                                              const ScheduleDAG &DAG);

std::vector<const SUnit *> makeGCNILPScheduler(ArrayRef<const SUnit *> BotRoots,
                                               const ScheduleDAG &DAG);
} // namespace llvm

// shim accessors for different order containers
static inline MachineInstr *getMachineInstr(MachineInstr *MI) { … }
static inline MachineInstr *getMachineInstr(const SUnit *SU) { … }
static inline MachineInstr *getMachineInstr(const SUnit &SU) { … }

#if !defined(NDEBUG) || defined(LLVM_ENABLE_DUMP)
LLVM_DUMP_METHOD
static void printRegion(raw_ostream &OS,
                        MachineBasicBlock::iterator Begin,
                        MachineBasicBlock::iterator End,
                        const LiveIntervals *LIS,
                        unsigned MaxInstNum =
                          std::numeric_limits<unsigned>::max()) {
  auto BB = Begin->getParent();
  OS << BB->getParent()->getName() << ":" << printMBBReference(*BB) << ' '
     << BB->getName() << ":\n";
  auto I = Begin;
  MaxInstNum = std::max(MaxInstNum, 1u);
  for (; I != End && MaxInstNum; ++I, --MaxInstNum) {
    if (!I->isDebugInstr() && LIS)
      OS << LIS->getInstructionIndex(*I);
    OS << '\t' << *I;
  }
  if (I != End) {
    OS << "\t...\n";
    I = std::prev(End);
    if (!I->isDebugInstr() && LIS)
      OS << LIS->getInstructionIndex(*I);
    OS << '\t' << *I;
  }
  if (End != BB->end()) { // print boundary inst if present
    OS << "----\n";
    if (LIS) OS << LIS->getInstructionIndex(*End) << '\t';
    OS << *End;
  }
}

LLVM_DUMP_METHOD
static void printLivenessInfo(raw_ostream &OS,
                              MachineBasicBlock::iterator Begin,
                              MachineBasicBlock::iterator End,
                              const LiveIntervals *LIS) {
  const auto BB = Begin->getParent();
  const auto &MRI = BB->getParent()->getRegInfo();

  const auto LiveIns = getLiveRegsBefore(*Begin, *LIS);
  OS << "LIn RP: " << print(getRegPressure(MRI, LiveIns));

  const auto BottomMI = End == BB->end() ? std::prev(End) : End;
  const auto LiveOuts = getLiveRegsAfter(*BottomMI, *LIS);
  OS << "LOt RP: " << print(getRegPressure(MRI, LiveOuts));
}

LLVM_DUMP_METHOD
void GCNIterativeScheduler::printRegions(raw_ostream &OS) const {
  const auto &ST = MF.getSubtarget<GCNSubtarget>();
  for (const auto R : Regions) {
    OS << "Region to schedule ";
    printRegion(OS, R->Begin, R->End, LIS, 1);
    printLivenessInfo(OS, R->Begin, R->End, LIS);
    OS << "Max RP: " << print(R->MaxPressure, &ST);
  }
}

LLVM_DUMP_METHOD
void GCNIterativeScheduler::printSchedResult(raw_ostream &OS,
                                             const Region *R,
                                             const GCNRegPressure &RP) const {
  OS << "\nAfter scheduling ";
  printRegion(OS, R->Begin, R->End, LIS);
  printSchedRP(OS, R->MaxPressure, RP);
  OS << '\n';
}

LLVM_DUMP_METHOD
void GCNIterativeScheduler::printSchedRP(raw_ostream &OS,
                                         const GCNRegPressure &Before,
                                         const GCNRegPressure &After) const {
  const auto &ST = MF.getSubtarget<GCNSubtarget>();
  OS << "RP before: " << print(Before, &ST)
     << "RP after:  " << print(After, &ST);
}
#endif

// DAG builder helper
class GCNIterativeScheduler::BuildDAG { … };

class GCNIterativeScheduler::OverrideLegacyStrategy { … };

namespace {

// just a stub to make base class happy
class SchedStrategyStub : public MachineSchedStrategy { … };

} // end anonymous namespace

GCNIterativeScheduler::GCNIterativeScheduler(MachineSchedContext *C,
                                             StrategyKind S)
  : … { … }

// returns max pressure for a region
GCNRegPressure
GCNIterativeScheduler::getRegionPressure(MachineBasicBlock::iterator Begin,
                                         MachineBasicBlock::iterator End)
  const { … }

// returns max pressure for a tentative schedule
template <typename Range> GCNRegPressure
GCNIterativeScheduler::getSchedulePressure(const Region &R,
                                           Range &&Schedule) const { … }

void GCNIterativeScheduler::enterRegion(MachineBasicBlock *BB, // overridden
                                        MachineBasicBlock::iterator Begin,
                                        MachineBasicBlock::iterator End,
                                        unsigned NumRegionInstrs) { … }

void GCNIterativeScheduler::schedule() { … }

void GCNIterativeScheduler::finalizeSchedule() { … }

// Detach schedule from SUnits and interleave it with debug values.
// Returned schedule becomes independent of DAG state.
std::vector<MachineInstr*>
GCNIterativeScheduler::detachSchedule(ScheduleRef Schedule) const { … }

void GCNIterativeScheduler::setBestSchedule(Region &R,
                                            ScheduleRef Schedule,
                                            const GCNRegPressure &MaxRP) { … }

void GCNIterativeScheduler::scheduleBest(Region &R) { … }

// minimal required region scheduler, works for ranges of SUnits*,
// SUnits or MachineIntrs*
template <typename Range>
void GCNIterativeScheduler::scheduleRegion(Region &R, Range &&Schedule,
                                           const GCNRegPressure &MaxRP) { … }

// Sort recorded regions by pressure - highest at the front
void GCNIterativeScheduler::sortRegionsByPressure(unsigned TargetOcc) { … }

///////////////////////////////////////////////////////////////////////////////
// Legacy MaxOccupancy Strategy

// Tries to increase occupancy applying minreg scheduler for a sequence of
// most demanding regions. Obtained schedules are saved as BestSchedule for a
// region.
// TargetOcc is the best achievable occupancy for a kernel.
// Returns better occupancy on success or current occupancy on fail.
// BestSchedules aren't deleted on fail.
unsigned GCNIterativeScheduler::tryMaximizeOccupancy(unsigned TargetOcc) { … }

void GCNIterativeScheduler::scheduleLegacyMaxOccupancy(
  bool TryMaximizeOccupancy) { … }

///////////////////////////////////////////////////////////////////////////////
// Minimal Register Strategy

void GCNIterativeScheduler::scheduleMinReg(bool force) { … }

///////////////////////////////////////////////////////////////////////////////
// ILP scheduler port

void GCNIterativeScheduler::scheduleILP(
  bool TryMaximizeOccupancy) { … }