Files
TheRock/rocm-libraries/shared/rocroller/lib/source/ExpressionTransformations/CombineShifts.cpp
T
derek.holloway 7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
Flatten subprojects into a single base monorepo
2026-09-02 16:34:47 -07:00

208 lines
6.8 KiB
C++

// Copyright Advanced Micro Devices, Inc., or its affiliates.
// SPDX-License-Identifier: MIT
#include <rocRoller/Expression.hpp>
#include <bit>
template <typename T>
constexpr auto cast_to_unsigned(T val)
{
return static_cast<typename std::make_unsigned<T>::type>(val);
}
namespace rocRoller
{
namespace Expression
{
struct GetShiftInfo
{
struct Result
{
bool isShift = false;
bool isSigned = false;
bool shiftLeft = false;
ExpressionPtr value = nullptr;
ExpressionPtr shiftAmount = nullptr;
};
Result operator()(auto const& expr) const
{
return {};
}
template <CShift Shift>
Result operator()(Shift const& expr) const
{
bool isSigned = std::same_as<Shift, ArithmeticShiftR> //
&& DataTypeInfo::Get(resultVariableType(expr.lhs)).isSigned;
return {true, isSigned, std::same_as<Shift, ShiftL>, expr.lhs, expr.rhs};
}
Result call(Expression const& expr) const
{
return std::visit(*this, expr);
}
};
struct CombineShiftsExpressionVisitor
{
template <CUnary Expr>
ExpressionPtr operator()(Expr const& expr) const
{
Expr cpy = expr;
cpy.arg = call(expr.arg);
return std::make_shared<Expression>(cpy);
}
template <CBinary Expr>
requires(!CShift<Expr>) ExpressionPtr operator()(Expr const& expr) const
{
Expr cpy = expr;
cpy.lhs = call(expr.lhs);
cpy.rhs = call(expr.rhs);
return std::make_shared<Expression>(cpy);
}
template <CTernary Expr>
ExpressionPtr operator()(Expr const& expr) const
{
Expr cpy = expr;
cpy.lhs = call(expr.lhs);
cpy.r1hs = call(expr.r1hs);
cpy.r2hs = call(expr.r2hs);
return std::make_shared<Expression>(cpy);
}
template <CNary Expr>
ExpressionPtr operator()(Expr const& expr) const
{
auto cpy = expr;
std::ranges::for_each(cpy.operands, [this](auto& op) { op = call(op); });
return std::make_shared<Expression>(std::move(cpy));
}
ExpressionPtr operator()(ScaledMatrixMultiply const& expr) const
{
auto cpy = expr;
cpy.matA = call(expr.matA);
cpy.matB = call(expr.matB);
cpy.matC = call(expr.matC);
cpy.scaleA = call(expr.scaleA);
cpy.scaleB = call(expr.scaleB);
return std::make_shared<Expression>(cpy);
}
template <CShift Expr>
ExpressionPtr operator()(Expr const& expr) const
{
ExpressionPtr rv;
auto lhs = call(expr.lhs);
auto rhs = call(expr.rhs);
auto [_, isSigned, isLeft, __, shiftAmount] = GetShiftInfo().call(Expr{lhs, rhs});
auto [lhsIsShift, lhsIsSigned, lhsIsLeft, lhsValue, lhsAmount]
= GetShiftInfo().call(*lhs);
if(lhsIsShift)
{
if(!isSigned && (isLeft != lhsIsLeft))
{
// Left then right. AND-off MSBs, if the shifts are the same.
// Or, Right then left. AND-off LSBs, if the shifts are the same.
auto sameShift = (shiftAmount == lhsAmount);
if(canEvaluateTo(true, sameShift))
{
// If we get here then:
// - Both shifts can be evaluated at translate time.
// - Both shifts are equal to each other.
int bits = getUnsignedInt(evaluate(shiftAmount));
auto lhsValueType = resultVariableType(lhsValue);
if(lhsValueType.getElementSize() == 4)
{
auto mask = (~(0u));
if(isLeft)
mask <<= bits;
else
mask >>= bits;
static_assert(sizeof(mask) == 4);
rv = lhsValue & literal(mask, lhsValueType);
}
else if(lhsValueType.getElementSize() == 8)
{
auto mask = (~(0ul));
if(isLeft)
mask <<= bits;
else
mask >>= bits;
static_assert(sizeof(mask) == 8);
rv = lhsValue & literal(mask, lhsValueType);
}
// Else: leave the original expression.
}
}
}
if(rv != nullptr)
{
copyComment(rv, lhs);
copyComment(rv, expr);
return rv;
}
return std::make_shared<Expression>(Expr{lhs, rhs, expr.comment});
}
template <CValue Value>
ExpressionPtr operator()(Value const& expr) const
{
return std::make_shared<Expression>(expr);
}
ExpressionPtr call(ExpressionPtr expr) const
{
if(!expr)
return expr;
auto resultVarType = resultVariableType(expr);
auto rv = std::visit(*this, *expr);
if(resultVariableType(rv) != resultVarType)
return convert(resultVarType, rv);
return rv;
}
};
/**
* Attempts to use combineShifts for all of the shifts within an Expression.
*/
ExpressionPtr combineShifts(ExpressionPtr expr)
{
auto visitor = CombineShiftsExpressionVisitor();
auto rv = visitor.call(expr);
AssertFatal(resultVariableType(expr) == resultVariableType(rv),
"Transformation should not alter result type.",
ShowValue(expr),
ShowValue(rv));
return rv;
}
}
}