Skip to content

[AMDGPU] Fix GCNUpwardRPTracker. (WIP) #71186

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Nov 10, 2023
Merged

[AMDGPU] Fix GCNUpwardRPTracker. (WIP) #71186

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
e2254cb
[AMDGPU] Fix GCNUpwardRPTracker.
vpykhtin Nov 2, 2023
bfc74ff
comments
vpykhtin Nov 7, 2023
c0e353f
comments
vpykhtin Nov 10, 2023
c49bb0f
Merge branch 'main' into fix_upward_rp
vpykhtin Nov 10, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions llvm/lib/Target/AMDGPU/GCNIterativeScheduler.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -251,7 +251,7 @@ GCNIterativeScheduler::getRegionPressure(MachineBasicBlock::iterator Begin,
assert(UPTracker.isValid() ||
(dbgs() << "Tracked region ",
printRegion(dbgs(), Begin, End, LIS), false));
return UPTracker.moveMaxPressure();
return UPTracker.getMaxPressureAndReset();
}

// returns max pressure for a tentative schedule
Expand All @@ -272,7 +272,7 @@ GCNIterativeScheduler::getSchedulePressure(const Region &R,
for (auto I = Schedule.end(), B = Schedule.begin(); I != B;) {
RPTracker.recede(*getMachineInstr(*--I));
}
return RPTracker.moveMaxPressure();
return RPTracker.getMaxPressureAndReset();
}

void GCNIterativeScheduler::enterRegion(MachineBasicBlock *BB, // overridden
Expand Down
137 changes: 72 additions & 65 deletions llvm/lib/Target/AMDGPU/GCNRegPressure.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -166,66 +166,60 @@ static LaneBitmask getDefRegMask(const MachineOperand &MO,
MRI.getTargetRegisterInfo()->getSubRegIndexLaneMask(MO.getSubReg());
}

static LaneBitmask getUsedRegMask(const MachineOperand &MO,
const MachineRegisterInfo &MRI,
const LiveIntervals &LIS) {
assert(MO.isUse() && MO.isReg() && MO.getReg().isVirtual());

if (auto SubReg = MO.getSubReg())
return MRI.getTargetRegisterInfo()->getSubRegIndexLaneMask(SubReg);

auto MaxMask = MRI.getMaxLaneMaskForVReg(MO.getReg());
if (SIRegisterInfo::getNumCoveredRegs(MaxMask) > 1) // cannot have subregs
return MaxMask;

// For a tentative schedule LIS isn't updated yet but livemask should remain
// the same on any schedule. Subreg defs can be reordered but they all must
// dominate uses anyway.
auto SI = LIS.getInstructionIndex(*MO.getParent()).getBaseIndex();
return getLiveLaneMask(MO.getReg(), SI, LIS, MRI);
}

static SmallVector<RegisterMaskPair, 8>
collectVirtualRegUses(const MachineInstr &MI, const LiveIntervals &LIS,
static void
collectVirtualRegUses(SmallVectorImpl<RegisterMaskPair> &RegMaskPairs,
const MachineInstr &MI, const LiveIntervals &LIS,
const MachineRegisterInfo &MRI) {
SmallVector<RegisterMaskPair, 8> Res;
SlotIndex InstrSI;
for (const auto &MO : MI.operands()) {
if (!MO.isReg() || !MO.getReg().isVirtual())
continue;
if (!MO.isUse() || !MO.readsReg())
continue;

auto const UsedMask = getUsedRegMask(MO, MRI, LIS);
Register Reg = MO.getReg();
if (llvm::any_of(RegMaskPairs, [Reg](const RegisterMaskPair &RM) {
return RM.RegUnit == Reg;
}))
continue;

LaneBitmask UseMask;
auto &LI = LIS.getInterval(Reg);
if (!LI.hasSubRanges())
UseMask = MRI.getMaxLaneMaskForVReg(Reg);
else {
// For a tentative schedule LIS isn't updated yet but livemask should
// remain the same on any schedule. Subreg defs can be reordered but they
// all must dominate uses anyway.
if (!InstrSI)
InstrSI = LIS.getInstructionIndex(*MO.getParent()).getBaseIndex();
UseMask = getLiveLaneMask(LI, InstrSI, MRI);
}

auto Reg = MO.getReg();
auto I = llvm::find_if(
Res, [Reg](const RegisterMaskPair &RM) { return RM.RegUnit == Reg; });
if (I != Res.end())
I->LaneMask |= UsedMask;
else
Res.push_back(RegisterMaskPair(Reg, UsedMask));
RegMaskPairs.emplace_back(Reg, UseMask);
}
return Res;
}

///////////////////////////////////////////////////////////////////////////////
// GCNRPTracker

LaneBitmask llvm::getLiveLaneMask(unsigned Reg,
SlotIndex SI,
LaneBitmask llvm::getLiveLaneMask(unsigned Reg, SlotIndex SI,
const LiveIntervals &LIS,
const MachineRegisterInfo &MRI) {
return getLiveLaneMask(LIS.getInterval(Reg), SI, MRI);
}

LaneBitmask llvm::getLiveLaneMask(const LiveInterval &LI, SlotIndex SI,
const MachineRegisterInfo &MRI) {
LaneBitmask LiveMask;
const auto &LI = LIS.getInterval(Reg);
if (LI.hasSubRanges()) {
for (const auto &S : LI.subranges())
if (S.liveAt(SI)) {
LiveMask |= S.LaneMask;
assert(LiveMask < MRI.getMaxLaneMaskForVReg(Reg) ||
LiveMask == MRI.getMaxLaneMaskForVReg(Reg));
assert(LiveMask == (LiveMask & MRI.getMaxLaneMaskForVReg(LI.reg())));
}
} else if (LI.liveAt(SI)) {
LiveMask = MRI.getMaxLaneMaskForVReg(Reg);
LiveMask = MRI.getMaxLaneMaskForVReg(LI.reg());
}
return LiveMask;
}
Expand Down Expand Up @@ -261,15 +255,14 @@ void GCNRPTracker::reset(const MachineInstr &MI,
MaxPressure = CurPressure = getRegPressure(*MRI, LiveRegs);
}

void GCNUpwardRPTracker::reset(const MachineInstr &MI,
const LiveRegSet *LiveRegsCopy) {
GCNRPTracker::reset(MI, LiveRegsCopy, true);
}
////////////////////////////////////////////////////////////////////////////////
// GCNUpwardRPTracker

void GCNUpwardRPTracker::reset(const MachineRegisterInfo &MRI_,
const LiveRegSet &LiveRegs_) {
MRI = &MRI_;
LiveRegs = LiveRegs_;
LastTrackedMI = nullptr;
MaxPressure = CurPressure = getRegPressure(MRI_, LiveRegs_);
}

Expand All @@ -281,41 +274,55 @@ void GCNUpwardRPTracker::recede(const MachineInstr &MI) {
if (MI.isDebugInstr())
return;

auto const RegUses = collectVirtualRegUses(MI, LIS, *MRI);

// calc pressure at the MI (defs + uses)
auto AtMIPressure = CurPressure;
for (const auto &U : RegUses) {
auto LiveMask = LiveRegs[U.RegUnit];
AtMIPressure.inc(U.RegUnit, LiveMask, LiveMask | U.LaneMask, *MRI);
}
// update max pressure
MaxPressure = max(AtMIPressure, MaxPressure);

for (const auto &MO : MI.all_defs()) {
if (!MO.getReg().isVirtual() || MO.isDead())
continue;

auto Reg = MO.getReg();
auto DecrementDef = [this](const MachineOperand &MO) {
Register Reg = MO.getReg();
auto I = LiveRegs.find(Reg);
if (I == LiveRegs.end())
continue;
auto &LiveMask = I->second;
auto PrevMask = LiveMask;
return;

LaneBitmask &LiveMask = I->second;
LaneBitmask PrevMask = LiveMask;
LiveMask &= ~getDefRegMask(MO, *MRI);
CurPressure.inc(Reg, PrevMask, LiveMask, *MRI);
if (LiveMask.none())
LiveRegs.erase(I);
};

// Decrement non-early-clobber defs.
SmallVector<const MachineOperand *, 2> EarlyClobberDefs;
for (const MachineOperand &MO : MI.all_defs()) {
if (!MO.getReg().isVirtual())
continue;
if (!MO.isEarlyClobber())
DecrementDef(MO);
else
EarlyClobberDefs.push_back(&MO);
}
for (const auto &U : RegUses) {
auto &LiveMask = LiveRegs[U.RegUnit];
auto PrevMask = LiveMask;

// Increment uses.
SmallVector<RegisterMaskPair, 8> RegUses;
collectVirtualRegUses(RegUses, MI, LIS, *MRI);
for (const RegisterMaskPair &U : RegUses) {
LaneBitmask &LiveMask = LiveRegs[U.RegUnit];
LaneBitmask PrevMask = LiveMask;
LiveMask |= U.LaneMask;
CurPressure.inc(U.RegUnit, PrevMask, LiveMask, *MRI);
}

// Point of maximum pressure: non-early-clobber defs are decremented and uses
// are incremented.
MaxPressure = max(CurPressure, MaxPressure);

// Now decrement early clobber defs.
for (const MachineOperand *MO : EarlyClobberDefs)
DecrementDef(*MO);

assert(CurPressure == getRegPressure(*MRI, LiveRegs));
}

////////////////////////////////////////////////////////////////////////////////
// GCNDownwardRPTracker

bool GCNDownwardRPTracker::reset(const MachineInstr &MI,
const LiveRegSet *LiveRegsCopy) {
MRI = &MI.getParent()->getParent()->getRegInfo();
Expand Down Expand Up @@ -562,15 +569,15 @@ bool GCNRegPressurePrinter::runOnMachineFunction(MachineFunction &MF) {
} else {
GCNUpwardRPTracker RPT(LIS);
RPT.reset(MRI, MBBEndSlot);
RPT.moveMaxPressure(); // Clear max pressure.

LiveOut = RPT.getLiveRegs();
RPAtMBBEnd = RPT.getPressure();

for (auto &MI : reverse(MBB)) {
RPT.resetMaxPressure();
RPT.recede(MI);
if (!MI.isDebugInstr())
RP.emplace_back(RPT.getPressure(), RPT.moveMaxPressure());
RP.emplace_back(RPT.getPressure(), RPT.getMaxPressure());
}

LiveIn = RPT.getLiveRegs();
Expand Down
53 changes: 38 additions & 15 deletions llvm/lib/Target/AMDGPU/GCNRegPressure.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -130,13 +130,6 @@ class GCNRPTracker {

GCNRegPressure getPressure() const { return CurPressure; }

// returns MaxPressure, resetting it
decltype(MaxPressure) moveMaxPressure() {
auto Res = MaxPressure;
MaxPressure.clear();
return Res;
}

decltype(LiveRegs) moveLiveRegs() {
return std::move(LiveRegs);
}
Expand All @@ -149,24 +142,41 @@ class GCNUpwardRPTracker : public GCNRPTracker {
public:
GCNUpwardRPTracker(const LiveIntervals &LIS_) : GCNRPTracker(LIS_) {}

// reset tracker to the point just below MI
// filling live regs upon this point using LIS
void reset(const MachineInstr &MI, const LiveRegSet *LiveRegs = nullptr);

// reset tracker and set live register set to the specified value.
void reset(const MachineRegisterInfo &MRI_, const LiveRegSet &LiveRegs_);

// reset tracker at the specified slot index.
void reset(const MachineRegisterInfo &MRI_, SlotIndex SI) {
reset(MRI_, llvm::getLiveRegs(SI, LIS, MRI_));
void reset(const MachineRegisterInfo &MRI, SlotIndex SI) {
reset(MRI, llvm::getLiveRegs(SI, LIS, MRI));
}

// move to the state just above the MI
// reset tracker to the end of the MBB.
void reset(const MachineBasicBlock &MBB) {
reset(MBB.getParent()->getRegInfo(),
LIS.getSlotIndexes()->getMBBEndIdx(&MBB));
}

// reset tracker to the point just after MI (in program order).
void reset(const MachineInstr &MI) {
reset(MI.getMF()->getRegInfo(), LIS.getInstructionIndex(MI).getDeadSlot());
}

// move to the state just before the MI (in program order).
void recede(const MachineInstr &MI);

// checks whether the tracker's state after receding MI corresponds
// to reported by LIS
// to reported by LIS.
bool isValid() const;

const GCNRegPressure &getMaxPressure() const { return MaxPressure; }

void resetMaxPressure() { MaxPressure = CurPressure; }

GCNRegPressure getMaxPressureAndReset() {
GCNRegPressure RP = MaxPressure;
resetMaxPressure();
return RP;
}
};

class GCNDownwardRPTracker : public GCNRPTracker {
Expand All @@ -180,6 +190,13 @@ class GCNDownwardRPTracker : public GCNRPTracker {

MachineBasicBlock::const_iterator getNext() const { return NextMI; }

// Return MaxPressure and clear it.
GCNRegPressure moveMaxPressure() {
auto Res = MaxPressure;
MaxPressure.clear();
return Res;
}

// Reset tracker to the point before the MI
// filling live regs upon this point using LIS.
// Returns false if block is empty except debug values.
Expand Down Expand Up @@ -209,6 +226,12 @@ LaneBitmask getLiveLaneMask(unsigned Reg,
const LiveIntervals &LIS,
const MachineRegisterInfo &MRI);

LaneBitmask getLiveLaneMask(const LiveInterval &LI, SlotIndex SI,
const MachineRegisterInfo &MRI);

GCNRPTracker::LiveRegSet getLiveRegs(SlotIndex SI, const LiveIntervals &LIS,
const MachineRegisterInfo &MRI);

/// creates a map MachineInstr -> LiveRegSet
/// R - range of iterators on instructions
/// After - upon entry or exit of every instruction
Expand Down
Loading