OrdinaryMagician/vkdoom_m
OrdinaryMagician/vkdoom_m
1
Fork 0
Code Activity

Use LLVM to JIT the code for one of the drawer functions

Browse source
This commit is contained in:
Magnus Norddahl 2016-09-26 09:00:19 +02:00
commit 3dd8b593b6
52 changed files with 4705 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

106
src/r_compiler/ssa/ssa_ubyte_ptr.cpp Normal file
View file

@ -0,0 +1,106 @@
#include "ssa_ubyte_ptr.h"
#include "ssa_scope.h"
#include "r_compiler/llvm_include.h"
SSAUBytePtr::SSAUBytePtr()
: v(0)
{
}
SSAUBytePtr::SSAUBytePtr(llvm::Value *v)
: v(v)
{
}
llvm::Type *SSAUBytePtr::llvm_type()
{
return llvm::Type::getInt8PtrTy(SSAScope::context());
}
SSAUBytePtr SSAUBytePtr::operator[](SSAInt index) const
{
return SSAUBytePtr::from_llvm(SSAScope::builder().CreateGEP(v, index.v, SSAScope::hint()));
}
SSAUByte SSAUBytePtr::load() const
{
return SSAUByte::from_llvm(SSAScope::builder().CreateLoad(v, false, SSAScope::hint()));
}
SSAVec4i SSAUBytePtr::load_vec4ub() const
{
// _mm_cvtsi32_si128 as implemented by clang:
SSAInt i32 = SSAInt::from_llvm(SSAScope::builder().CreateLoad(SSAScope::builder().CreateBitCast(v, llvm::Type::getInt32PtrTy(SSAScope::context()), SSAScope::hint()), false, SSAScope::hint()));
llvm::Value *v = SSAScope::builder().CreateInsertElement(llvm::UndefValue::get(SSAVec4i::llvm_type()), i32.v, SSAInt(0).v, SSAScope::hint());
v = SSAScope::builder().CreateInsertElement(v, SSAInt(0).v, SSAInt(1).v, SSAScope::hint());
v = SSAScope::builder().CreateInsertElement(v, SSAInt(0).v, SSAInt(2).v, SSAScope::hint());
v = SSAScope::builder().CreateInsertElement(v, SSAInt(0).v, SSAInt(3).v, SSAScope::hint());
SSAVec4i v4i = SSAVec4i::from_llvm(v);
SSAVec8s low = SSAVec8s::bitcast(SSAVec16ub::shuffle(SSAVec16ub::bitcast(v4i), 0, 0, 16+0, 1, 16+1, 2, 16+2, 3, 16+3, 4, 16+4, 5, 16+5, 6, 16+6, 7, 16+7)); // _mm_unpacklo_epi8
return SSAVec4i::extendlo(low); // _mm_unpacklo_epi16
/*
llvm::PointerType *m4xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 4)->getPointerTo();
llvm::Type *m4xint32type = llvm::VectorType::get(llvm::Type::getInt32Ty(SSAScope::context()), 4);
llvm::Value *v4ub = SSAScope::builder().CreateLoad(SSAScope::builder().CreateBitCast(v, m4xint8typeptr, SSAScope::hint()), false, SSAScope::hint());
return SSAVec4i::from_llvm(SSAScope::builder().CreateZExt(v4ub, m4xint32type));
*/
}
SSAVec16ub SSAUBytePtr::load_vec16ub() const
{
llvm::PointerType *m16xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 16)->getPointerTo();
return SSAVec16ub::from_llvm(SSAScope::builder().CreateLoad(SSAScope::builder().CreateBitCast(v, m16xint8typeptr, SSAScope::hint()), false, SSAScope::hint()));
}
SSAVec16ub SSAUBytePtr::load_unaligned_vec16ub() const
{
llvm::PointerType *m16xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 16)->getPointerTo();
return SSAVec16ub::from_llvm(SSAScope::builder().Insert(new llvm::LoadInst(SSAScope::builder().CreateBitCast(v, m16xint8typeptr, SSAScope::hint()), SSAScope::hint(), false, 4), SSAScope::hint()));
}
void SSAUBytePtr::store(const SSAUByte &new_value)
{
SSAScope::builder().CreateStore(new_value.v, v, false);
}
void SSAUBytePtr::store_vec4ub(const SSAVec4i &new_value)
{
// Store using saturate:
SSAVec8s v8s(new_value, new_value);
SSAVec16ub v16ub(v8s, v8s);
llvm::Type *m16xint8type = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 16);
llvm::PointerType *m4xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 4)->getPointerTo();
std::vector<llvm::Constant*> constants;
constants.push_back(llvm::ConstantInt::get(SSAScope::context(), llvm::APInt(32, 0)));
constants.push_back(llvm::ConstantInt::get(SSAScope::context(), llvm::APInt(32, 1)));
constants.push_back(llvm::ConstantInt::get(SSAScope::context(), llvm::APInt(32, 2)));
constants.push_back(llvm::ConstantInt::get(SSAScope::context(), llvm::APInt(32, 3)));
llvm::Value *mask = llvm::ConstantVector::get(constants);
llvm::Value *val_vector = SSAScope::builder().CreateShuffleVector(v16ub.v, llvm::UndefValue::get(m16xint8type), mask, SSAScope::hint());
SSAScope::builder().CreateStore(val_vector, SSAScope::builder().CreateBitCast(v, m4xint8typeptr, SSAScope::hint()), false);
}
void SSAUBytePtr::store_vec16ub(const SSAVec16ub &new_value)
{
llvm::PointerType *m16xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 16)->getPointerTo();
llvm::StoreInst *inst = SSAScope::builder().CreateAlignedStore(new_value.v, SSAScope::builder().CreateBitCast(v, m16xint8typeptr, SSAScope::hint()), 16);
// The following generates _mm_stream_si128, maybe!
// llvm::MDNode *node = llvm::MDNode::get(SSAScope::context(), SSAScope::builder().getInt32(1));
// inst->setMetadata(SSAScope::module()->getMDKindID("nontemporal"), node);
}
void SSAUBytePtr::store_unaligned_vec16ub(const SSAVec16ub &new_value)
{
/*llvm::Value *values[2] =
{
SSAScope::builder().CreateBitCast(v, llvm::Type::getInt8PtrTy(SSAScope::context())),
new_value.v
};
SSAScope::builder().CreateCall(SSAScope::intrinsic(llvm::Intrinsic::x86_sse2_storeu_dq), values);*/
llvm::PointerType *m16xint8typeptr = llvm::VectorType::get(llvm::Type::getInt8Ty(SSAScope::context()), 16)->getPointerTo();
llvm::StoreInst *inst = SSAScope::builder().CreateStore(new_value.v, SSAScope::builder().CreateBitCast(v, m16xint8typeptr, SSAScope::hint()));
}