Merge pull request #257 from KomputeProject/fixed_formatting
Updated formatting
This commit is contained in:
Alejandro Saucedo
GitHub
commit
ecf05b2c76
17 changed files with 162 additions and 117 deletions
1
.github/workflows/cpp_tests.yml
vendored
1
.github/workflows/cpp_tests.yml
vendored
|
|
@ -31,4 +31,3 @@ jobs:
|
|||
run: |
|
||||
export VK_ICD_FILENAMES=/swiftshader/vk_swiftshader_icd.json
|
||||
make mk_run_tests
|
||||
|
||||
|
|
|
|||
2
.github/workflows/python_tests.yml
vendored
2
.github/workflows/python_tests.yml
vendored
|
|
@ -26,5 +26,3 @@ jobs:
|
|||
run: |
|
||||
export VK_ICD_FILENAMES=/swiftshader/vk_swiftshader_icd.json
|
||||
make test_python
|
||||
|
||||
|
||||
|
|
|
|||
|
|
@ -16,4 +16,3 @@ build_vulkan_sdk:
|
|||
|
||||
push_vulkan_sdk: build_vulkan_sdk
|
||||
docker push axsauze/vulkan-sdk:0.1
|
||||
|
||||
|
|
|
|||
|
|
@ -57,4 +57,3 @@ Documentation Index (as per sidebar)
|
|||
Variable Types for Tensors, and Push/Spec Constants <overview/variable-types>
|
||||
Asynchronous & Parallel Operations <overview/async-parallel>
|
||||
Code Index <genindex>
|
||||
|
||||
|
|
|
|||
|
|
@ -1,3 +1,4 @@
|
|||
# Python Package for Kompute
|
||||
|
||||
Read the documentation under `/docs/overview` for details on the python package.
|
||||
|
||||
|
|
|
|||
|
|
@ -23,8 +23,8 @@ Algorithm::isInit()
|
|||
void
|
||||
Algorithm::destroy()
|
||||
{
|
||||
// We don't have to free memory on destroy as it's freed by the commandBuffer destructor
|
||||
// if (this->mPushConstantsData) {
|
||||
// We don't have to free memory on destroy as it's freed by the
|
||||
// commandBuffer destructor if (this->mPushConstantsData) {
|
||||
// free(this->mPushConstantsData);
|
||||
// }
|
||||
// if (this->mSpecializationConstantsData) {
|
||||
|
|
@ -238,7 +238,8 @@ Algorithm::createPipeline()
|
|||
if (this->mPushConstantsSize) {
|
||||
pushConstantRange.setStageFlags(vk::ShaderStageFlagBits::eCompute);
|
||||
pushConstantRange.setOffset(0);
|
||||
pushConstantRange.setSize(this->mPushConstantsDataTypeMemorySize * this->mPushConstantsSize);
|
||||
pushConstantRange.setSize(this->mPushConstantsDataTypeMemorySize *
|
||||
this->mPushConstantsSize);
|
||||
|
||||
pipelineLayoutInfo.setPushConstantRangeCount(1);
|
||||
pipelineLayoutInfo.setPPushConstantRanges(&pushConstantRange);
|
||||
|
|
@ -254,7 +255,8 @@ Algorithm::createPipeline()
|
|||
for (uint32_t i = 0; i < this->mSpecializationConstantsSize; i++) {
|
||||
vk::SpecializationMapEntry specializationEntry(
|
||||
static_cast<uint32_t>(i),
|
||||
static_cast<uint32_t>(this->mSpecializationConstantsDataTypeMemorySize * i),
|
||||
static_cast<uint32_t>(
|
||||
this->mSpecializationConstantsDataTypeMemorySize * i),
|
||||
this->mSpecializationConstantsDataTypeMemorySize);
|
||||
|
||||
specializationEntries.push_back(specializationEntry);
|
||||
|
|
@ -265,7 +267,8 @@ Algorithm::createPipeline()
|
|||
vk::SpecializationInfo specializationInfo(
|
||||
static_cast<uint32_t>(specializationEntries.size()),
|
||||
specializationEntries.data(),
|
||||
this->mSpecializationConstantsDataTypeMemorySize * this->mSpecializationConstantsSize,
|
||||
this->mSpecializationConstantsDataTypeMemorySize *
|
||||
this->mSpecializationConstantsSize,
|
||||
this->mSpecializationConstantsData);
|
||||
|
||||
vk::PipelineShaderStageCreateInfo shaderStage(
|
||||
|
|
@ -339,12 +342,14 @@ Algorithm::recordBindPush(const vk::CommandBuffer& commandBuffer)
|
|||
{
|
||||
if (this->mPushConstantsSize) {
|
||||
KP_LOG_DEBUG("Kompute Algorithm binding push constants memory size: {}",
|
||||
this->mPushConstantsSize * this->mPushConstantsDataTypeMemorySize);
|
||||
this->mPushConstantsSize *
|
||||
this->mPushConstantsDataTypeMemorySize);
|
||||
|
||||
commandBuffer.pushConstants(*this->mPipelineLayout,
|
||||
vk::ShaderStageFlagBits::eCompute,
|
||||
0,
|
||||
this->mPushConstantsSize * this->mPushConstantsDataTypeMemorySize,
|
||||
this->mPushConstantsSize *
|
||||
this->mPushConstantsDataTypeMemorySize,
|
||||
this->mPushConstantsData);
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -288,22 +288,22 @@ Manager::createDevice(const std::vector<uint32_t>& familyQueueIndices,
|
|||
}
|
||||
|
||||
this->mFreeDevice = true;
|
||||
|
||||
|
||||
// Getting an integer that says how many vuklan devices we have
|
||||
uint32_t deviceCount = 0;
|
||||
this->mInstance->enumeratePhysicalDevices(&deviceCount, nullptr);
|
||||
|
||||
|
||||
// This means there are no devices at all
|
||||
if (deviceCount == 0) {
|
||||
throw std::runtime_error("Failed to find GPUs with Vulkan support! "
|
||||
"Maybe you haven't installed vulkan drivers?");
|
||||
}
|
||||
|
||||
|
||||
// This means that we're exceeding our device limit, for
|
||||
// example if we have 2 devices, just physicalDeviceIndex
|
||||
// 0 and 1 are acceptable. Hence, physicalDeviceIndex should
|
||||
// always be less than deviceCount, else we raise an error
|
||||
if ( !(deviceCount > physicalDeviceIndex) ) {
|
||||
if (!(deviceCount > physicalDeviceIndex)) {
|
||||
throw std::runtime_error("There is no such physical index or device, "
|
||||
"please use your existing device");
|
||||
}
|
||||
|
|
|
|||
|
|
@ -33,9 +33,9 @@ OpAlgoDispatch::record(const vk::CommandBuffer& commandBuffer)
|
|||
|
||||
if (this->mPushConstantsSize) {
|
||||
this->mAlgorithm->setPushConstants(
|
||||
this->mPushConstantsData,
|
||||
this->mPushConstantsSize,
|
||||
this->mPushConstantsDataTypeMemorySize);
|
||||
this->mPushConstantsData,
|
||||
this->mPushConstantsSize,
|
||||
this->mPushConstantsDataTypeMemorySize);
|
||||
}
|
||||
|
||||
this->mAlgorithm->recordBindCore(commandBuffer);
|
||||
|
|
|
|||
|
|
@ -24,12 +24,13 @@ class Algorithm
|
|||
* @param spirv (optional) The spirv code to use to create the algorithm
|
||||
* @param workgroup (optional) The kp::Workgroup to use for the dispatch
|
||||
* which defaults to kp::Workgroup(tensor[0].size(), 1, 1) if not set.
|
||||
* @param specializationConstants (optional) The templatable param is to be used to
|
||||
* initialize the specialization constants which cannot be changed once set.
|
||||
* @param pushConstants (optional) This templatable param is to be used when
|
||||
* initializing the pipeline, which set the size of the push constants -
|
||||
* these can be modified but all new values must have the same data type and length
|
||||
* as otherwise it will result in errors.
|
||||
* @param specializationConstants (optional) The templatable param is to be
|
||||
* used to initialize the specialization constants which cannot be changed
|
||||
* once set.
|
||||
* @param pushConstants (optional) This templatable param is to be used
|
||||
* when initializing the pipeline, which set the size of the push constants
|
||||
* - these can be modified but all new values must have the same data type
|
||||
* and length as otherwise it will result in errors.
|
||||
*/
|
||||
template<typename S = float, typename P = float>
|
||||
Algorithm(std::shared_ptr<vk::Device> device,
|
||||
|
|
@ -44,15 +45,20 @@ class Algorithm
|
|||
this->mDevice = device;
|
||||
|
||||
if (tensors.size() && spirv.size()) {
|
||||
KP_LOG_INFO("Kompute Algorithm initialising with tensor size: {} and "
|
||||
"spirv size: {}",
|
||||
tensors.size(),
|
||||
spirv.size());
|
||||
this->rebuild(
|
||||
tensors, spirv, workgroup, specializationConstants, pushConstants);
|
||||
KP_LOG_INFO(
|
||||
"Kompute Algorithm initialising with tensor size: {} and "
|
||||
"spirv size: {}",
|
||||
tensors.size(),
|
||||
spirv.size());
|
||||
this->rebuild(tensors,
|
||||
spirv,
|
||||
workgroup,
|
||||
specializationConstants,
|
||||
pushConstants);
|
||||
} else {
|
||||
KP_LOG_INFO("Kompute Algorithm constructor with empty tensors and or "
|
||||
"spirv so not rebuilding vulkan components");
|
||||
KP_LOG_INFO(
|
||||
"Kompute Algorithm constructor with empty tensors and or "
|
||||
"spirv so not rebuilding vulkan components");
|
||||
}
|
||||
}
|
||||
|
||||
|
|
@ -64,8 +70,9 @@ class Algorithm
|
|||
* @param spirv The spirv code to use to create the algorithm
|
||||
* @param workgroup (optional) The kp::Workgroup to use for the dispatch
|
||||
* which defaults to kp::Workgroup(tensor[0].size(), 1, 1) if not set.
|
||||
* @param specializationConstants (optional) The std::vector<float> to use to
|
||||
* initialize the specialization constants which cannot be changed once set.
|
||||
* @param specializationConstants (optional) The std::vector<float> to use
|
||||
* to initialize the specialization constants which cannot be changed once
|
||||
* set.
|
||||
* @param pushConstants (optional) The std::vector<float> to use when
|
||||
* initializing the pipeline, which set the size of the push constants -
|
||||
* these can be modified but all new values must have the same vector size
|
||||
|
|
@ -87,11 +94,14 @@ class Algorithm
|
|||
if (this->mSpecializationConstantsData) {
|
||||
free(this->mSpecializationConstantsData);
|
||||
}
|
||||
uint32_t memorySize = sizeof(decltype(specializationConstants.back()));
|
||||
uint32_t memorySize =
|
||||
sizeof(decltype(specializationConstants.back()));
|
||||
uint32_t size = specializationConstants.size();
|
||||
uint32_t totalSize = size * memorySize;
|
||||
this->mSpecializationConstantsData = malloc(totalSize);
|
||||
memcpy(this->mSpecializationConstantsData, specializationConstants.data(), totalSize);
|
||||
memcpy(this->mSpecializationConstantsData,
|
||||
specializationConstants.data(),
|
||||
totalSize);
|
||||
this->mSpecializationConstantsDataTypeMemorySize = memorySize;
|
||||
this->mSpecializationConstantsSize = size;
|
||||
}
|
||||
|
|
@ -109,11 +119,11 @@ class Algorithm
|
|||
this->mPushConstantsSize = size;
|
||||
}
|
||||
|
||||
this->setWorkgroup(workgroup,
|
||||
this->mTensors.size() ? this->mTensors[0]->size() : 1);
|
||||
this->setWorkgroup(
|
||||
workgroup, this->mTensors.size() ? this->mTensors[0]->size() : 1);
|
||||
|
||||
// Descriptor pool is created first so if available then destroy all before
|
||||
// rebuild
|
||||
// Descriptor pool is created first so if available then destroy all
|
||||
// before rebuild
|
||||
if (this->isInit()) {
|
||||
this->destroy();
|
||||
}
|
||||
|
|
@ -176,9 +186,9 @@ class Algorithm
|
|||
* Sets the push constants to the new value provided to use in the next
|
||||
* bindPush()
|
||||
*
|
||||
* @param pushConstants The templatable vector is to be used to set the push constants to use in the
|
||||
* next bindPush(...) calls. The constants provided must be of the same size
|
||||
* as the ones created during initialization.
|
||||
* @param pushConstants The templatable vector is to be used to set the push
|
||||
* constants to use in the next bindPush(...) calls. The constants provided
|
||||
* must be of the same size as the ones created during initialization.
|
||||
*/
|
||||
template<typename T>
|
||||
void setPushConstants(const std::vector<T>& pushConstants)
|
||||
|
|
@ -193,21 +203,24 @@ class Algorithm
|
|||
* Sets the push constants to the new value provided to use in the next
|
||||
* bindPush() with the raw memory block location and memory size to be used.
|
||||
*
|
||||
* @param data The raw data point to copy the data from, without modifying the pointer.
|
||||
* @param data The raw data point to copy the data from, without modifying
|
||||
* the pointer.
|
||||
* @param size The number of data elements provided in the data
|
||||
* @param memorySize The memory size of each of the data elements in bytes.
|
||||
*/
|
||||
void setPushConstants(void* data, uint32_t size, uint32_t memorySize) {
|
||||
void setPushConstants(void* data, uint32_t size, uint32_t memorySize)
|
||||
{
|
||||
|
||||
uint32_t totalSize = memorySize * size;
|
||||
uint32_t previousTotalSize = this->mPushConstantsDataTypeMemorySize * this->mPushConstantsSize;
|
||||
uint32_t previousTotalSize =
|
||||
this->mPushConstantsDataTypeMemorySize * this->mPushConstantsSize;
|
||||
|
||||
if (totalSize != previousTotalSize) {
|
||||
throw std::runtime_error(
|
||||
fmt::format("Kompute Algorithm push "
|
||||
"constant total memory size provided is {} but expected {} bytes",
|
||||
totalSize,
|
||||
previousTotalSize));
|
||||
throw std::runtime_error(fmt::format(
|
||||
"Kompute Algorithm push "
|
||||
"constant total memory size provided is {} but expected {} bytes",
|
||||
totalSize,
|
||||
previousTotalSize));
|
||||
}
|
||||
if (this->mPushConstantsData) {
|
||||
free(this->mPushConstantsData);
|
||||
|
|
@ -230,13 +243,15 @@ class Algorithm
|
|||
/**
|
||||
* Gets the specialization constants of the current algorithm.
|
||||
*
|
||||
* @returns The std::vector<float> currently set for specialization constants
|
||||
* @returns The std::vector<float> currently set for specialization
|
||||
* constants
|
||||
*/
|
||||
template<typename T>
|
||||
const std::vector<T> getSpecializationConstants()
|
||||
{
|
||||
return { (T*)this->mSpecializationConstantsData,
|
||||
((T*)this->mSpecializationConstantsData) + this->mSpecializationConstantsSize };
|
||||
((T*)this->mSpecializationConstantsData) +
|
||||
this->mSpecializationConstantsSize };
|
||||
}
|
||||
/**
|
||||
* Gets the specialization constants of the current algorithm.
|
||||
|
|
@ -247,7 +262,7 @@ class Algorithm
|
|||
const std::vector<T> getPushConstants()
|
||||
{
|
||||
return { (T*)this->mPushConstantsData,
|
||||
((T*)this->mPushConstantsData) + this->mPushConstantsSize };
|
||||
((T*)this->mPushConstantsData) + this->mPushConstantsSize };
|
||||
}
|
||||
/**
|
||||
* Gets the current tensors that are used in the algorithm.
|
||||
|
|
|
|||
|
|
@ -124,8 +124,9 @@ class Manager
|
|||
}
|
||||
|
||||
/**
|
||||
* Default non-template function that can be used to create algorithm objects
|
||||
* which provides default types to the push and spec constants as floats.
|
||||
* Default non-template function that can be used to create algorithm
|
||||
* objects which provides default types to the push and spec constants as
|
||||
* floats.
|
||||
*
|
||||
* @param tensors (optional) The tensors to initialise the algorithm with
|
||||
* @param spirv (optional) The SPIRV bytes for the algorithm to dispatch
|
||||
|
|
@ -144,7 +145,8 @@ class Manager
|
|||
const std::vector<float>& specializationConstants = {},
|
||||
const std::vector<float>& pushConstants = {})
|
||||
{
|
||||
return this->algorithm<>(tensors, spirv, workgroup, specializationConstants, pushConstants);
|
||||
return this->algorithm<>(
|
||||
tensors, spirv, workgroup, specializationConstants, pushConstants);
|
||||
}
|
||||
|
||||
/**
|
||||
|
|
@ -155,10 +157,10 @@ class Manager
|
|||
* @param spirv (optional) The SPIRV bytes for the algorithm to dispatch
|
||||
* @param workgroup (optional) kp::Workgroup for algorithm to use, and
|
||||
* defaults to (tensor[0].size(), 1, 1)
|
||||
* @param specializationConstants (optional) templatable vector parameter to use for
|
||||
* specialization constants, and defaults to an empty constant
|
||||
* @param pushConstants (optional) templatable vector parameter to use for push constants,
|
||||
* and defaults to an empty constant
|
||||
* @param specializationConstants (optional) templatable vector parameter to
|
||||
* use for specialization constants, and defaults to an empty constant
|
||||
* @param pushConstants (optional) templatable vector parameter to use for
|
||||
* push constants, and defaults to an empty constant
|
||||
* @returns Shared pointer with initialised algorithm
|
||||
*/
|
||||
template<typename S = float, typename P = float>
|
||||
|
|
@ -200,7 +202,8 @@ class Manager
|
|||
/**
|
||||
* Information about the current device.
|
||||
*
|
||||
* @return vk::PhysicalDeviceProperties containing information about the device
|
||||
* @return vk::PhysicalDeviceProperties containing information about the
|
||||
*device
|
||||
**/
|
||||
vk::PhysicalDeviceProperties getDeviceProperties() const;
|
||||
|
||||
|
|
@ -211,7 +214,6 @@ class Manager
|
|||
**/
|
||||
std::vector<vk::PhysicalDevice> listDevices() const;
|
||||
|
||||
|
||||
private:
|
||||
// -------------- OPTIONALLY OWNED RESOURCES
|
||||
std::shared_ptr<vk::Instance> mInstance = nullptr;
|
||||
|
|
|
|||
|
|
@ -168,8 +168,10 @@ TEST(TestAsyncOperations, TestManagerAsyncExecution)
|
|||
|
||||
// AMD Drivers in Windows may see an error in this line due to timeout.
|
||||
// In order to fix this, it requires a change on Windows registries.
|
||||
// More details on this can be found here: https://docs.substance3d.com/spdoc/gpu-drivers-crash-with-long-computations-128745489.html
|
||||
// Context on solution discussed in github: https://github.com/KomputeProject/kompute/issues/196#issuecomment-808866505
|
||||
// More details on this can be found here:
|
||||
// https://docs.substance3d.com/spdoc/gpu-drivers-crash-with-long-computations-128745489.html
|
||||
// Context on solution discussed in github:
|
||||
// https://github.com/KomputeProject/kompute/issues/196#issuecomment-808866505
|
||||
sq1->evalAsync<kp::OpAlgoDispatch>(algo1);
|
||||
sq2->evalAsync<kp::OpAlgoDispatch>(algo2);
|
||||
|
||||
|
|
@ -232,8 +234,10 @@ TEST(TestAsyncOperations, TestManagerAsyncExecutionTimeout)
|
|||
|
||||
// AMD Drivers in Windows may see an error in this line due to timeout.
|
||||
// In order to fix this, it requires a change on Windows registries.
|
||||
// More details on this can be found here: https://docs.substance3d.com/spdoc/gpu-drivers-crash-with-long-computations-128745489.html
|
||||
// Context on solution discussed in github: https://github.com/KomputeProject/kompute/issues/196#issuecomment-808866505
|
||||
// More details on this can be found here:
|
||||
// https://docs.substance3d.com/spdoc/gpu-drivers-crash-with-long-computations-128745489.html
|
||||
// Context on solution discussed in github:
|
||||
// https://github.com/KomputeProject/kompute/issues/196#issuecomment-808866505
|
||||
sq1->evalAsync<kp::OpAlgoDispatch>(algo1);
|
||||
sq2->evalAsync<kp::OpAlgoDispatch>(algo2);
|
||||
|
||||
|
|
@ -245,7 +249,8 @@ TEST(TestAsyncOperations, TestManagerAsyncExecutionTimeout)
|
|||
std::chrono::duration_cast<std::chrono::microseconds>(endSync - startSync)
|
||||
.count();
|
||||
|
||||
// The time should several orders of magnitude smaller (in this 100k instead of 1m ns)
|
||||
// The time should several orders of magnitude smaller (in this 100k instead
|
||||
// of 1m ns)
|
||||
EXPECT_LT(duration, 100000);
|
||||
|
||||
sq1->evalAsync<kp::OpTensorSyncLocal>({ tensorA, tensorB });
|
||||
|
|
|
|||
|
|
@ -126,8 +126,8 @@ TEST(TestLogisticRegression, TestMainLogisticRegressionManualCopy)
|
|||
kp::shader_data::
|
||||
shaders_glsl_logisticregression_comp_spv_len));
|
||||
|
||||
std::shared_ptr<kp::Algorithm> algorithm =
|
||||
mgr.algorithm(params, spirv, kp::Workgroup(), std::vector<float>({ 5.0 }));
|
||||
std::shared_ptr<kp::Algorithm> algorithm = mgr.algorithm(
|
||||
params, spirv, kp::Workgroup(), std::vector<float>({ 5.0 }));
|
||||
|
||||
std::shared_ptr<kp::Sequence> sq =
|
||||
mgr.sequence()
|
||||
|
|
|
|||
|
|
@ -87,7 +87,8 @@ TEST(TestManager, TestClearDestroy)
|
|||
{
|
||||
std::shared_ptr<kp::TensorT<float>> tensorLHS = mgr.tensor({ 0, 1, 2 });
|
||||
std::shared_ptr<kp::TensorT<float>> tensorRHS = mgr.tensor({ 2, 4, 6 });
|
||||
std::shared_ptr<kp::TensorT<float>> tensorOutput = mgr.tensor({ 0, 0, 0 });
|
||||
std::shared_ptr<kp::TensorT<float>> tensorOutput =
|
||||
mgr.tensor({ 0, 0, 0 });
|
||||
|
||||
std::vector<std::shared_ptr<kp::Tensor>> params = { tensorLHS,
|
||||
tensorRHS,
|
||||
|
|
|
|||
|
|
@ -53,11 +53,8 @@ TEST(TestMultipleAlgoExecutions, TestEndToEndFunctionality)
|
|||
std::vector<float> pushConstsA({ 2.0 });
|
||||
std::vector<float> pushConstsB({ 3.0 });
|
||||
|
||||
auto algorithm = mgr.algorithm(params,
|
||||
compileSource(shader),
|
||||
workgroup,
|
||||
specConsts,
|
||||
pushConstsA);
|
||||
auto algorithm = mgr.algorithm(
|
||||
params, compileSource(shader), workgroup, specConsts, pushConstsA);
|
||||
|
||||
// 3. Run operation with string shader synchronously
|
||||
mgr.sequence()
|
||||
|
|
@ -97,11 +94,11 @@ TEST(TestMultipleAlgoExecutions, SingleSequenceRecord)
|
|||
{
|
||||
// A sharedMemoryBarrier is required as the shader is not thread-safe:w
|
||||
std::shared_ptr<kp::OpMemoryBarrier> shaderBarrier{
|
||||
new kp::OpMemoryBarrier({ tensorA },
|
||||
vk::AccessFlagBits::eTransferRead,
|
||||
vk::AccessFlagBits::eShaderWrite,
|
||||
vk::PipelineStageFlagBits::eComputeShader,
|
||||
vk::PipelineStageFlagBits::eComputeShader)
|
||||
new kp::OpMemoryBarrier({ tensorA },
|
||||
vk::AccessFlagBits::eTransferRead,
|
||||
vk::AccessFlagBits::eShaderWrite,
|
||||
vk::PipelineStageFlagBits::eComputeShader,
|
||||
vk::PipelineStageFlagBits::eComputeShader)
|
||||
};
|
||||
|
||||
mgr.sequence()
|
||||
|
|
@ -266,11 +263,8 @@ TEST(TestMultipleAlgoExecutions, TestAlgorithmUtilFunctions)
|
|||
std::vector<float> specConsts({ 2 });
|
||||
std::vector<float> pushConsts({ 2.0 });
|
||||
|
||||
auto algorithm = mgr.algorithm(params,
|
||||
compileSource(shader),
|
||||
workgroup,
|
||||
specConsts,
|
||||
pushConsts);
|
||||
auto algorithm = mgr.algorithm(
|
||||
params, compileSource(shader), workgroup, specConsts, pushConsts);
|
||||
|
||||
EXPECT_EQ(algorithm->getWorkgroup(), workgroup);
|
||||
EXPECT_EQ(algorithm->getPushConstants<float>(), pushConsts);
|
||||
|
|
|
|||
|
|
@ -44,8 +44,10 @@ TEST(TestPushConstants, TestConstantsAlgoDispatchOverride)
|
|||
// We need to run this in sequence to avoid race condition
|
||||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<float>{ 0.1, 0.2, 0.3 });
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<float>{ 0.3, 0.2, 0.1 });
|
||||
sq->eval<kp::OpAlgoDispatch>(algo,
|
||||
std::vector<float>{ 0.1, 0.2, 0.3 });
|
||||
sq->eval<kp::OpAlgoDispatch>(algo,
|
||||
std::vector<float>{ 0.3, 0.2, 0.1 });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<float>({ 0.4, 0.4, 0.4 }));
|
||||
|
|
@ -90,7 +92,8 @@ TEST(TestPushConstants, TestConstantsAlgoDispatchNoOverride)
|
|||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo);
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<float>{ 0.3, 0.2, 0.1 });
|
||||
sq->eval<kp::OpAlgoDispatch>(algo,
|
||||
std::vector<float>{ 0.3, 0.2, 0.1 });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<float>({ 0.4, 0.4, 0.4 }));
|
||||
|
|
@ -156,22 +159,22 @@ TEST(TestPushConstants, TestConstantsWrongSize)
|
|||
// pa[1] += pcs.y;
|
||||
// pa[2] += pcs.z;
|
||||
// })");
|
||||
//
|
||||
//
|
||||
// std::vector<uint32_t> spirv = compileSource(shader);
|
||||
//
|
||||
//
|
||||
// std::shared_ptr<kp::Sequence> sq = nullptr;
|
||||
//
|
||||
//
|
||||
// {
|
||||
// kp::Manager mgr;
|
||||
//
|
||||
//
|
||||
// std::shared_ptr<kp::TensorT<float>> tensor =
|
||||
// mgr.tensor({ 0, 0, 0 });
|
||||
//
|
||||
//
|
||||
// std::shared_ptr<kp::Algorithm> algo = mgr.algorithm(
|
||||
// { tensor }, spirv, kp::Workgroup({ 1 }), {}, { 0.0 });
|
||||
//
|
||||
//
|
||||
// sq = mgr.sequence()->record<kp::OpTensorSyncDevice>({ tensor });
|
||||
//
|
||||
//
|
||||
// EXPECT_THROW(sq->record<kp::OpAlgoDispatch>(
|
||||
// algo, std::vector<uint32_t>{ 1, 2, 3 }),
|
||||
// std::runtime_error);
|
||||
|
|
@ -197,7 +200,8 @@ TEST(TestPushConstants, TestConstantsMixedTypes)
|
|||
pa[2] += pcs.z;
|
||||
})");
|
||||
|
||||
struct TestConsts{
|
||||
struct TestConsts
|
||||
{
|
||||
float x;
|
||||
uint32_t y;
|
||||
int32_t z;
|
||||
|
|
@ -213,16 +217,19 @@ TEST(TestPushConstants, TestConstantsMixedTypes)
|
|||
std::shared_ptr<kp::TensorT<float>> tensor =
|
||||
mgr.tensorT<float>({ 0, 0, 0 });
|
||||
|
||||
std::shared_ptr<kp::Algorithm> algo = mgr.algorithm<float, TestConsts>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, {{ 0, 0, 0 }});
|
||||
std::shared_ptr<kp::Algorithm> algo =
|
||||
mgr.algorithm<float, TestConsts>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, { { 0, 0, 0 } });
|
||||
|
||||
sq = mgr.sequence()->eval<kp::OpTensorSyncDevice>({ tensor });
|
||||
|
||||
// We need to run this in sequence to avoid race condition
|
||||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<TestConsts>{{ 15.32, 2147483650, 10 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<TestConsts>{{ 30.32, 2147483650, -3 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo, std::vector<TestConsts>{ { 15.32, 2147483650, 10 } });
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo, std::vector<TestConsts>{ { 30.32, 2147483650, -3 } });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<float>({ 45.64, 1300, 7 }));
|
||||
|
|
@ -258,16 +265,19 @@ TEST(TestPushConstants, TestConstantsInt)
|
|||
std::shared_ptr<kp::TensorT<int32_t>> tensor =
|
||||
mgr.tensorT<int32_t>({ -1, -1, -1 });
|
||||
|
||||
std::shared_ptr<kp::Algorithm> algo = mgr.algorithm<int32_t , int32_t>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, {{ 0, 0, 0 }});
|
||||
std::shared_ptr<kp::Algorithm> algo =
|
||||
mgr.algorithm<int32_t, int32_t>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, { { 0, 0, 0 } });
|
||||
|
||||
sq = mgr.sequence()->eval<kp::OpTensorSyncDevice>({ tensor });
|
||||
|
||||
// We need to run this in sequence to avoid race condition
|
||||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<int32_t>{{ -1, -1, -1 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<int32_t>{{ -1, -1, -1 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo, std::vector<int32_t>{ { -1, -1, -1 } });
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo, std::vector<int32_t>{ { -1, -1, -1 } });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<int32_t>({ -3, -3, -3 }));
|
||||
|
|
@ -303,19 +313,25 @@ TEST(TestPushConstants, TestConstantsUnsignedInt)
|
|||
std::shared_ptr<kp::TensorT<uint32_t>> tensor =
|
||||
mgr.tensorT<uint32_t>({ 0, 0, 0 });
|
||||
|
||||
std::shared_ptr<kp::Algorithm> algo = mgr.algorithm<uint32_t , uint32_t>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, {{ 0, 0, 0 }});
|
||||
std::shared_ptr<kp::Algorithm> algo =
|
||||
mgr.algorithm<uint32_t, uint32_t>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, { { 0, 0, 0 } });
|
||||
|
||||
sq = mgr.sequence()->eval<kp::OpTensorSyncDevice>({ tensor });
|
||||
|
||||
// We need to run this in sequence to avoid race condition
|
||||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<uint32_t>{{ 2147483650, 2147483650, 2147483650 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<uint32_t>{{ 5, 5, 5 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo,
|
||||
std::vector<uint32_t>{ { 2147483650, 2147483650, 2147483650 } });
|
||||
sq->eval<kp::OpAlgoDispatch>(algo,
|
||||
std::vector<uint32_t>{ { 5, 5, 5 } });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<uint32_t>({ 2147483655, 2147483655, 2147483655 }));
|
||||
EXPECT_EQ(
|
||||
tensor->vector(),
|
||||
std::vector<uint32_t>({ 2147483655, 2147483655, 2147483655 }));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
@ -349,18 +365,29 @@ TEST(TestPushConstants, TestConstantsDouble)
|
|||
mgr.tensorT<double>({ 0, 0, 0 });
|
||||
|
||||
std::shared_ptr<kp::Algorithm> algo = mgr.algorithm<double, double>(
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, {{ 0, 0, 0 }});
|
||||
{ tensor }, spirv, kp::Workgroup({ 1 }), {}, { { 0, 0, 0 } });
|
||||
|
||||
sq = mgr.sequence()->eval<kp::OpTensorSyncDevice>({ tensor });
|
||||
|
||||
// We need to run this in sequence to avoid race condition
|
||||
// We can't use atomicAdd as swiftshader doesn't support it for
|
||||
// float
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<double>{{ 1.1111222233334444, 2.1111222233334444, 3.1111222233334444 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(algo, std::vector<double>{{ 1.1111222233334444, 2.1111222233334444, 3.1111222233334444 }});
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo,
|
||||
std::vector<double>{ { 1.1111222233334444,
|
||||
2.1111222233334444,
|
||||
3.1111222233334444 } });
|
||||
sq->eval<kp::OpAlgoDispatch>(
|
||||
algo,
|
||||
std::vector<double>{ { 1.1111222233334444,
|
||||
2.1111222233334444,
|
||||
3.1111222233334444 } });
|
||||
sq->eval<kp::OpTensorSyncLocal>({ tensor });
|
||||
|
||||
EXPECT_EQ(tensor->vector(), std::vector<double>({ 2.2222444466668888, 4.2222444466668888, 6.2222444466668888 }));
|
||||
EXPECT_EQ(tensor->vector(),
|
||||
std::vector<double>({ 2.2222444466668888,
|
||||
4.2222444466668888,
|
||||
6.2222444466668888 }));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
|
|
|||
|
|
@ -101,4 +101,3 @@ TEST(TestSpecializationConstants, TestConstantsInt)
|
|||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
|
|
|||
|
|
@ -33,7 +33,8 @@ TEST(TestTensor, DataTypes)
|
|||
{
|
||||
std::vector<uint32_t> vec{ 0, 1, 2 };
|
||||
std::shared_ptr<kp::TensorT<uint32_t>> tensor = mgr.tensorT(vec);
|
||||
EXPECT_EQ(tensor->dataType(), kp::Tensor::TensorDataTypes::eUnsignedInt);
|
||||
EXPECT_EQ(tensor->dataType(),
|
||||
kp::Tensor::TensorDataTypes::eUnsignedInt);
|
||||
}
|
||||
|
||||
{
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue