RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-23 19:23:23 +01:00
Code Issues Projects Releases Wiki Activity

Reformat the loop that does AddRecExpr * AddRecExpr reduction.

Browse Source 
No functionality.

llvm-svn: 157672
This commit is contained in:
Andrew Trick 2012-05-30 03:35:17 +00:00
parent a46443d6db
commit 0fda877fa8
1 changed files with 56 additions and 55 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

111
lib/Analysis/ScalarEvolution.cpp
View File

@ -2037,63 +2037,64 @@ const SCEV *ScalarEvolution::getMulExpr(SmallVectorImpl<const SCEV *> &Ops,
for (unsigned OtherIdx = Idx+1;
OtherIdx < Ops.size() && isa<SCEVAddRecExpr>(Ops[OtherIdx]);
++OtherIdx) {
if (AddRecLoop == cast<SCEVAddRecExpr>(Ops[OtherIdx])->getLoop()) {
// {A1,+,A2,+,...,+,An}<L> * {B1,+,B2,+,...,+,Bn}<L>
// = {x=1 in [ sum y=x..2x [ sum z=max(y-x, y-n)..min(x,n) [
// choose(x, 2x)*choose(2x-y, x-z)*A_{y-z}*B_z
// ]]],+,...up to x=2n}.
// Note that the arguments to choose() are always integers with values
// known at compile time, never SCEV objects.
//
// The implementation avoids pointless extra computations when the two
// addrec's are of different length (mathematically, it's equivalent to
// an infinite stream of zeros on the right).
bool OpsModified = false;
for (; OtherIdx != Ops.size() && isa<SCEVAddRecExpr>(Ops[OtherIdx]);
++OtherIdx)
if (const SCEVAddRecExpr *OtherAddRec =
dyn_cast<SCEVAddRecExpr>(Ops[OtherIdx]))
if (OtherAddRec->getLoop() == AddRecLoop) {
bool Overflow = false;
Type *Ty = AddRec->getType();
bool LargerThan64Bits = getTypeSizeInBits(Ty) > 64;
SmallVector<const SCEV*, 7> AddRecOps;
for (int x = 0, xe = AddRec->getNumOperands() +
OtherAddRec->getNumOperands() - 1;
x != xe && !Overflow; ++x) {
const SCEV *Term = getConstant(Ty, 0);
for (int y = x, ye = 2*x+1; y != ye && !Overflow; ++y) {
uint64_t Coeff1 = Choose(x, 2*x - y, Overflow);
for (int z = std::max(y-x, y-(int)AddRec->getNumOperands()+1),
ze = std::min(x+1, (int)OtherAddRec->getNumOperands());
z < ze && !Overflow; ++z) {
uint64_t Coeff2 = Choose(2*x - y, x-z, Overflow);
uint64_t Coeff;
if (LargerThan64Bits)
Coeff = umul_ov(Coeff1, Coeff2, Overflow);
else
Coeff = Coeff1*Coeff2;
const SCEV *CoeffTerm = getConstant(Ty, Coeff);
const SCEV *Term1 = AddRec->getOperand(y-z);
const SCEV *Term2 = OtherAddRec->getOperand(z);
Term = getAddExpr(Term, getMulExpr(CoeffTerm, Term1,Term2));
}
}
AddRecOps.push_back(Term);
}
if (!Overflow) {
const SCEV *NewAddRec = getAddRecExpr(AddRecOps,
AddRec->getLoop(),
SCEV::FlagAnyWrap);
if (Ops.size() == 2) return NewAddRec;
Ops[Idx] = AddRec = cast<SCEVAddRecExpr>(NewAddRec);
Ops.erase(Ops.begin() + OtherIdx); --OtherIdx;
OpsModified = true;
}
if (AddRecLoop != cast<SCEVAddRecExpr>(Ops[OtherIdx])->getLoop())
continue;
// {A1,+,A2,+,...,+,An}<L> * {B1,+,B2,+,...,+,Bn}<L>
// = {x=1 in [ sum y=x..2x [ sum z=max(y-x, y-n)..min(x,n) [
// choose(x, 2x)*choose(2x-y, x-z)*A_{y-z}*B_z
// ]]],+,...up to x=2n}.
// Note that the arguments to choose() are always integers with values
// known at compile time, never SCEV objects.
//
// The implementation avoids pointless extra computations when the two
// addrec's are of different length (mathematically, it's equivalent to
// an infinite stream of zeros on the right).
bool OpsModified = false;
for (; OtherIdx != Ops.size() && isa<SCEVAddRecExpr>(Ops[OtherIdx]);
++OtherIdx) {
const SCEVAddRecExpr *OtherAddRec =
dyn_cast<SCEVAddRecExpr>(Ops[OtherIdx]);
if (!OtherAddRec || OtherAddRec->getLoop() != AddRecLoop)
continue;
bool Overflow = false;
Type *Ty = AddRec->getType();
bool LargerThan64Bits = getTypeSizeInBits(Ty) > 64;
SmallVector<const SCEV*, 7> AddRecOps;
for (int x = 0, xe = AddRec->getNumOperands() +
OtherAddRec->getNumOperands() - 1; x != xe && !Overflow; ++x) {
const SCEV *Term = getConstant(Ty, 0);
for (int y = x, ye = 2*x+1; y != ye && !Overflow; ++y) {
uint64_t Coeff1 = Choose(x, 2*x - y, Overflow);
for (int z = std::max(y-x, y-(int)AddRec->getNumOperands()+1),
ze = std::min(x+1, (int)OtherAddRec->getNumOperands());
z < ze && !Overflow; ++z) {
uint64_t Coeff2 = Choose(2*x - y, x-z, Overflow);
uint64_t Coeff;
if (LargerThan64Bits)
Coeff = umul_ov(Coeff1, Coeff2, Overflow);
else
Coeff = Coeff1*Coeff2;
const SCEV *CoeffTerm = getConstant(Ty, Coeff);
const SCEV *Term1 = AddRec->getOperand(y-z);
const SCEV *Term2 = OtherAddRec->getOperand(z);
Term = getAddExpr(Term, getMulExpr(CoeffTerm, Term1,Term2));
}
if (OpsModified)
return getMulExpr(Ops);
}
AddRecOps.push_back(Term);
}
if (!Overflow) {
const SCEV *NewAddRec = getAddRecExpr(AddRecOps, AddRec->getLoop(),
SCEV::FlagAnyWrap);
if (Ops.size() == 2) return NewAddRec;
Ops[Idx] = AddRec = cast<SCEVAddRecExpr>(NewAddRec);
Ops.erase(Ops.begin() + OtherIdx); --OtherIdx;
OpsModified = true;
}
}
if (OpsModified)
return getMulExpr(Ops);
}
// Otherwise couldn't fold anything into this recurrence. Move onto the