Skip to content

Commit

Permalink
Fix most type conversions, disable GCC ABI change warning
Browse files Browse the repository at this point in the history 
See doe300/VC4CL#41
  • Loading branch information
@doe300
doe300 committed Jul 11, 2018
1 parent df53ebc commit 10641d4
Show file tree
Hide file tree
Showing 4 changed files with 153 additions and 22 deletions.
2 changes: 1 addition & 1 deletion src/CMakeLists.txt
View file
Original file line number Diff line number Diff line change
Expand Up @@ -16,7 +16,7 @@ set(VC4C_ENABLED_WARNINGS -Wall -Wextra -Wold-style-cast -Wno-unused-parameter -
if ("${CMAKE_CXX_COMPILER_ID}" STREQUAL "Clang")
SET(VC4C_ENABLED_WARNINGS ${VC4C_ENABLED_WARNINGS} -Weverything -Wno-c++98-compat -Wno-c++98-compat-pedantic -Wno-shadow -Wno-padded -Wno-shadow-field-in-constructor -Wno-global-constructors -Wno-exit-time-destructors -Wno-missing-prototypes -Wno-gnu-anonymous-struct -Wno-nested-anon-types -Wno-documentation -Wno-unused-command-line-argument -Wno-unused-member-function -Wno-gnu-zero-variadic-macro-arguments -Wno-covered-switch-default -Wno-switch-enum -Wswitch)
elseif("${CMAKE_CXX_COMPILER_ID}" STREQUAL "GNU")
SET(VC4C_ENABLED_WARNINGS ${VC4C_ENABLED_WARNINGS} -Wdouble-promotion -fdelete-null-pointer-checks -Wnull-dereference -Wuninitialized -Wsuggest-attribute=pure -Wsuggest-attribute=const -Wsuggest-attribute=format -Wsuggest-override -Wconversion -Wzero-as-null-pointer-constant)
SET(VC4C_ENABLED_WARNINGS ${VC4C_ENABLED_WARNINGS} -Wdouble-promotion -fdelete-null-pointer-checks -Wnull-dereference -Wuninitialized -Wsuggest-attribute=pure -Wsuggest-attribute=const -Wsuggest-attribute=format -Wsuggest-override -Wconversion -Wzero-as-null-pointer-constant -Wno-psabi)
endif()
target_compile_options(${VC4C_LIBRARY_NAME} PRIVATE ${VC4C_ENABLED_WARNINGS})
target_compile_options(${VC4C_PROGRAM_NAME} PRIVATE ${VC4C_ENABLED_WARNINGS})
Expand Down
12 changes: 9 additions & 3 deletions src/asm/OpCodes.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -492,9 +492,15 @@ Optional<Value> OpCode::calculate(const Optional<Value>& firstOperand, const Opt
Value res(ContainerValue(numElements), resultType);
for(unsigned char i = 0; i < numElements; ++i)
{
auto tmp = calculate(
firstVal->hasType(ValueType::CONTAINER) ? firstVal->container.elements.at(i) : firstVal.value(),
secondVal->hasType(ValueType::CONTAINER) ? secondVal->container.elements.at(i) : secondVal.value());
Optional<Value> tmp = NO_VALUE;
if(numOperands == 1)
tmp = calculate(
firstVal->hasType(ValueType::CONTAINER) ? firstVal->container.elements.at(i) : firstVal.value(),
NO_VALUE);
else
tmp = calculate(
firstVal->hasType(ValueType::CONTAINER) ? firstVal->container.elements.at(i) : firstVal.value(),
secondVal->hasType(ValueType::CONTAINER) ? secondVal->container.elements.at(i) : secondVal.value());
if(!tmp)
// result could not be calculated for a single component of the vector, abort
return NO_VALUE;
Expand Down
138 changes: 125 additions & 13 deletions test/TestArithmetic.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -10,17 +10,18 @@
#include "VC4C.h"
#include "tools.h"

#include <cstring>
#include <limits>
#include <random>
#include <sstream>

template <typename T, std::size_t N>
template <typename T, std::size_t N, T min = std::numeric_limits<T>::min(), T max = std::numeric_limits<T>::max()>
static std::array<T, N> generateInput(bool allowNull)
{
std::array<T, N> arr;
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_int_distribution<T> dis(std::numeric_limits<T>::min(), std::numeric_limits<T>::max());
std::uniform_int_distribution<T> dis(min, max);

for(std::size_t i = 0; i < N; ++i)
{
Expand All @@ -36,13 +37,14 @@ static std::array<T, N> generateInput(bool allowNull)
return arr;
}

template <std::size_t N>
template <std::size_t N, typename T = float>
static std::array<float, N> generateInput(bool allowNull)
{
std::array<float, N> arr;
std::random_device rd;
std::mt19937 gen(rd());
std::uniform_real_distribution<float> dis(std::numeric_limits<float>::min(), std::numeric_limits<float>::max());
std::uniform_real_distribution<float> dis(
static_cast<float>(std::numeric_limits<T>::min()), static_cast<float>(std::numeric_limits<T>::max()));

for(std::size_t i = 0; i < N; ++i)
{
Expand All @@ -64,6 +66,13 @@ __kernel void test(__global TYPE* out, const __global TYPE* in0, const __global
}
)";

static const std::string SELECTION_OPERATION = R"(
__kernel void test(__global TYPE* out, const __global TYPE* in0, const __global TYPE* in1) {
size_t gid = get_global_id(0);
out[gid] = select(in0[gid], in1[gid], in0[gid]);
}
)";

static const std::string RELATIONAL_OPERATION = R"(
__kernel void test(__global int16* out, const __global TYPE* in0, const __global TYPE* in1) {
size_t gid = get_global_id(0);
Expand Down Expand Up @@ -190,6 +199,12 @@ TestArithmetic::TestArithmetic(const vc4c::Configuration& config) : config(confi
TEST_ADD(TestArithmetic::testUnsignedCharLessEquals);
TEST_ADD(TestArithmetic::testFloatLessEquals);

TEST_ADD(TestArithmetic::testSignedIntSelection);
TEST_ADD(TestArithmetic::testSignedShortSelection);
TEST_ADD(TestArithmetic::testSignedCharSelection);
TEST_ADD(TestArithmetic::testUnsignedIntSelection);
TEST_ADD(TestArithmetic::testUnsignedShortSelection);
TEST_ADD(TestArithmetic::testUnsignedCharSelection);
TEST_ADD(TestArithmetic::testSignedIntAnd);
TEST_ADD(TestArithmetic::testSignedShortAnd);
TEST_ADD(TestArithmetic::testSignedCharAnd);
Expand All @@ -209,10 +224,18 @@ TestArithmetic::TestArithmetic(const vc4c::Configuration& config) : config(confi
TEST_ADD(TestArithmetic::testUnsignedTruncation);
TEST_ADD(TestArithmetic::testSignExtension);
TEST_ADD(TestArithmetic::testZeroExtension);
TEST_ADD(TestArithmetic::testSignedToFloat);
TEST_ADD(TestArithmetic::testUnsignedToFloat);
TEST_ADD(TestArithmetic::testFloatToSigned);
TEST_ADD(TestArithmetic::testFloatToUnsigned);
TEST_ADD(TestArithmetic::testSignedIntToFloat);
TEST_ADD(TestArithmetic::testSignedShortToFloat);
TEST_ADD(TestArithmetic::testSignedCharToFloat);
TEST_ADD(TestArithmetic::testUnsignedIntToFloat);
TEST_ADD(TestArithmetic::testUnsignedShortToFloat);
TEST_ADD(TestArithmetic::testUnsignedCharToFloat);
TEST_ADD(TestArithmetic::testFloatToSignedInt);
TEST_ADD(TestArithmetic::testFloatToSignedShort);
TEST_ADD(TestArithmetic::testFloatToSignedChar);
TEST_ADD(TestArithmetic::testFloatToUnsignedInt);
TEST_ADD(TestArithmetic::testFloatToUnsignedShort);
TEST_ADD(TestArithmetic::testFloatToUnsignedChar);
}

void TestArithmetic::onMismatch(const std::string& expected, const std::string& result)
Expand Down Expand Up @@ -332,6 +355,25 @@ void testRelationalOperation<float>(vc4c::Configuration& config, const std::stri
in0, in1, out, op, options.substr(pos, options.find(' ', pos) - pos), onError);
}

template <typename T>
static void testSelectionOperation(vc4c::Configuration& config, const std::string& options,
const std::function<void(const std::string&, const std::string&)>& onError)
{
std::stringstream code;
compileBuffer(config, code, SELECTION_OPERATION, options);

auto in0 = generateInput<T, 16 * 12>(true);
auto in1 = generateInput<T, 16 * 12>(true);

auto out = runEmulation<T, T, 16, 12>(code, {in0, in1});
auto op = [](T a, T b) -> T {
typename std::make_signed<T>::type signedVal;
std::memcpy(&signedVal, &a, sizeof(T));
return signedVal < 0 ? b : a;
};
checkBinaryResults<T, T>(in0, in1, out, op, "select", onError);
}

template <typename I, typename O>
static void testConversionOperation(vc4c::Configuration& config, const std::string& options,
const std::function<O(I)>& op, const std::function<void(const std::string&, const std::string&)>& onError)
Expand All @@ -352,7 +394,8 @@ static void testConversionOperation(vc4c::Configuration& config, const std::stri
std::stringstream code;
compileBuffer(config, code, CONVERSION_OPERATION, options);

auto in = generateInput<16 * 12>(true);
// values out of the range of O (the integer type) are implementation-defined
auto in = generateInput<16 * 12, O>(true);

auto out = runEmulation<float, O, 16, 12>(code, {in});
checkUnaryResults<O, float, 16 * 12>(in, out, op, "convert", onError);
Expand Down Expand Up @@ -719,6 +762,42 @@ void TestArithmetic::testFloatLessEquals()
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedIntSelection()
{
testSelectionOperation<int>(config, "-DTYPE=int16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedShortSelection()
{
testSelectionOperation<short>(config, "-DTYPE=short16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedCharSelection()
{
testSelectionOperation<char>(config, "-DTYPE=char16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedIntSelection()
{
testSelectionOperation<unsigned int>(config, "-DTYPE=uint16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedShortSelection()
{
testSelectionOperation<unsigned short>(config, "-DTYPE=ushort16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedCharSelection()
{
testSelectionOperation<unsigned char>(config, "-DTYPE=uchar16",
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedIntAnd()
{
testRelationalOperation<int>(config, "-DTYPE=int16 -DOP=&&", checkAnd<int>,
Expand Down Expand Up @@ -849,42 +928,75 @@ void TestArithmetic::testZeroExtension()
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedToFloat()
void TestArithmetic::testSignedIntToFloat()
{
testConversionOperation<int, float>(config, "-DIN=int16 -DOUT=float16", convert<int, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedShortToFloat()
{
testConversionOperation<short, float>(config, "-DIN=short16 -DOUT=float16", convert<short, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testSignedCharToFloat()
{
testConversionOperation<char, float>(config, "-DIN=char16 -DOUT=float16", convert<char, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedToFloat()
void TestArithmetic::testUnsignedIntToFloat()
{
testConversionOperation<unsigned int, float>(config, "-DIN=uint16 -DOUT=float16", convert<unsigned int, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedShortToFloat()
{
testConversionOperation<unsigned short, float>(config, "-DIN=ushort16 -DOUT=float16",
convert<unsigned short, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testUnsignedCharToFloat()
{
testConversionOperation<unsigned char, float>(config, "-DIN=uchar16 -DOUT=float16", convert<unsigned char, float>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testFloatToSigned()
void TestArithmetic::testFloatToSignedInt()
{
testConversionOperation<int>(config, "-DIN=float16 -DOUT=int16", convert<float, int>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testFloatToSignedShort()
{
testConversionOperation<short>(config, "-DIN=float16 -DOUT=short16", convert<float, short>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testFloatToSignedChar()
{
testConversionOperation<char>(config, "-DIN=float16 -DOUT=char16", convert<float, char>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}
void TestArithmetic::testFloatToUnsigned()

void TestArithmetic::testFloatToUnsignedInt()
{
testConversionOperation<unsigned int>(config, "-DIN=float16 -DOUT=uint16", convert<float, unsigned int>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testFloatToUnsignedShort()
{
testConversionOperation<unsigned short>(config, "-DIN=float16 -DOUT=ushort16", convert<float, unsigned short>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}

void TestArithmetic::testFloatToUnsignedChar()
{
testConversionOperation<unsigned char>(config, "-DIN=float16 -DOUT=uchar16", convert<float, unsigned char>,
std::bind(&TestArithmetic::onMismatch, this, std::placeholders::_1, std::placeholders::_2));
}
23 changes: 18 additions & 5 deletions test/TestArithmetic.h
View file
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,12 @@ class TestArithmetic : public Test::Suite
void testFloatLessEquals();

// logical operators
void testSignedIntSelection();
void testSignedShortSelection();
void testSignedCharSelection();
void testUnsignedIntSelection();
void testUnsignedShortSelection();
void testUnsignedCharSelection();
void testSignedIntAnd();
void testSignedShortAnd();
void testSignedCharAnd();
Expand All @@ -96,7 +102,6 @@ class TestArithmetic : public Test::Suite
void testSignedNot();
void testUnsignedNot();
void testFloatNot();
void testSelection();

//XXX vector instructions
void testExtractElement();
Expand All @@ -108,10 +113,18 @@ class TestArithmetic : public Test::Suite
void testUnsignedTruncation();
void testSignExtension();
void testZeroExtension();
void testSignedToFloat();
void testUnsignedToFloat();
void testFloatToSigned();
void testFloatToUnsigned();
void testSignedIntToFloat();
void testSignedShortToFloat();
void testSignedCharToFloat();
void testUnsignedIntToFloat();
void testUnsignedShortToFloat();
void testUnsignedCharToFloat();
void testFloatToSignedInt();
void testFloatToSignedShort();
void testFloatToSignedChar();
void testFloatToUnsignedInt();
void testFloatToUnsignedShort();
void testFloatToUnsignedChar();
//XXX
void testVectorBitcastTruncation4To1();
void testVectorBitcastTruncation2To1();
Expand Down

0 comments on commit 10641d4

Please sign in to comment.