AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Rename device access functions

Browse Source 
Rename functions that access devices to reflect the derived device
they return. This includes the base device::Device and the derived
gpu/pal/roc device classes in both NullDevice and Device forms. Change
to use the least derived versions to clarify what operations will be
available.

Change-Id: I1abb6bfed7efa24852bc8d0d49acaea357d8b5d0
This commit is contained in:
Tony Tye
2021-01-10 01:12:54 +00:00
parent f679b05df7
commit 001fd66cac
13 changed files with 147 additions and 127 deletions
Download Patch File Download Diff File Expand all files Collapse all files
rocclr/device/pal/palkernel.cpp
+25 -27
View File
@@ -43,8 +43,8 @@ void HSAILKernel::setWorkGroupInfo(const uint32_t privateSegmentSize,
// Make sure runtime matches HW alignment, which is 256 scratch regs (DWORDs) per wave
constexpr uint32_t ScratchRegAlignment = 256;
workGroupInfo_.scratchRegs_ =
amd::alignUp((workGroupInfo_.scratchRegs_ * dev().info().wavefrontWidth_),
ScratchRegAlignment) / dev().info().wavefrontWidth_;
amd::alignUp((workGroupInfo_.scratchRegs_ * device().info().wavefrontWidth_),
ScratchRegAlignment) / device().info().wavefrontWidth_;
workGroupInfo_.privateMemSize_ = workGroupInfo_.scratchRegs_ * sizeof(uint32_t);
workGroupInfo_.localMemSize_ = workGroupInfo_.usedLDSSize_ = groupSegmentSize;
workGroupInfo_.usedSGPRs_ = numSGPRs;
@@ -52,13 +52,13 @@ void HSAILKernel::setWorkGroupInfo(const uint32_t privateSegmentSize,
workGroupInfo_.usedVGPRs_ = numVGPRs;
if (!prog().isNull()) {
workGroupInfo_.availableLDSSize_ = dev().properties().gfxipProperties.shaderCore.ldsSizePerCu;
workGroupInfo_.availableLDSSize_ = palDevice().properties().gfxipProperties.shaderCore.ldsSizePerCu;
workGroupInfo_.availableSGPRs_ =
dev().properties().gfxipProperties.shaderCore.numAvailableSgprs;
palDevice().properties().gfxipProperties.shaderCore.numAvailableSgprs;
workGroupInfo_.availableVGPRs_ =
dev().properties().gfxipProperties.shaderCore.numAvailableVgprs;
palDevice().properties().gfxipProperties.shaderCore.numAvailableVgprs;
workGroupInfo_.preferredSizeMultiple_ = workGroupInfo_.wavefrontPerSIMD_ =
dev().info().wavefrontWidth_;
device().info().wavefrontWidth_;
} else {
workGroupInfo_.availableLDSSize_ = 64 * Ki;
workGroupInfo_.availableSGPRs_ = 104;
@@ -105,7 +105,7 @@ bool HSAILKernel::aqlCreateHWInfo(amd::hsa::loader::Symbol* sym) {
}
HSAILKernel::HSAILKernel(std::string name, HSAILProgram* prog, std::string compileOptions)
: device::Kernel(prog->dev(), name, *prog),
: device::Kernel(prog->device(), name, *prog),
compileOptions_(compileOptions),
index_(0),
code_(0),
@@ -128,12 +128,12 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
options.append(openClKernelName.c_str());
// Append an option so that we can selectively enable a SCOption on CZ
// whenever IOMMUv2 is enabled.
if (dev().settings().svmFineGrainSystem_) {
if (palNullDevice().settings().svmFineGrainSystem_) {
options.append(" -sc-xnack-iommu");
}
error = aclCompile(dev().compiler(), prog().binaryElf(), options.c_str(), ACL_TYPE_CG,
error = aclCompile(palNullDevice().compiler(), prog().binaryElf(), options.c_str(), ACL_TYPE_CG,
ACL_TYPE_ISA, nullptr);
buildLog_ += aclGetCompilerLog(dev().compiler());
buildLog_ += aclGetCompilerLog(palNullDevice().compiler());
if (error != ACL_SUCCESS) {
LogError("Failed to finalize kernel");
return false;
@@ -144,7 +144,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
// Pull out metadata from the ELF
size_t sizeOfArgList;
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_ARGUMENT_ARRAY,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_ARGUMENT_ARRAY,
openClKernelName.c_str(), nullptr, &sizeOfArgList);
if (error != ACL_SUCCESS) {
return false;
@@ -154,7 +154,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
if (nullptr == aclArgList) {
return false;
}
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_ARGUMENT_ARRAY,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_ARGUMENT_ARRAY,
openClKernelName.c_str(), aclArgList, &sizeOfArgList);
if (error != ACL_SUCCESS) {
return false;
@@ -164,30 +164,30 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
delete[] aclArgList;
size_t sizeOfWorkGroupSize;
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE,
openClKernelName.c_str(), nullptr, &sizeOfWorkGroupSize);
if (error != ACL_SUCCESS) {
return false;
}
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE,
openClKernelName.c_str(), workGroupInfo_.compileSize_, &sizeOfWorkGroupSize);
if (error != ACL_SUCCESS) {
return false;
}
// Copy wavefront size
workGroupInfo_.wavefrontSize_ = dev().info().wavefrontWidth_;
workGroupInfo_.wavefrontSize_ = device().info().wavefrontWidth_;
// Find total workgroup size
if (workGroupInfo_.compileSize_[0] != 0) {
workGroupInfo_.size_ = workGroupInfo_.compileSize_[0] * workGroupInfo_.compileSize_[1] *
workGroupInfo_.compileSize_[2];
} else {
workGroupInfo_.size_ = dev().info().preferredWorkGroupSize_;
workGroupInfo_.size_ = device().info().preferredWorkGroupSize_;
}
// Pull out printf metadata from the ELF
size_t sizeOfPrintfList;
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_GPU_PRINTF_ARRAY,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_GPU_PRINTF_ARRAY,
openClKernelName.c_str(), nullptr, &sizeOfPrintfList);
if (error != ACL_SUCCESS) {
return false;
@@ -199,7 +199,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
if (nullptr == aclPrintfList) {
return false;
}
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_GPU_PRINTF_ARRAY,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_GPU_PRINTF_ARRAY,
openClKernelName.c_str(), aclPrintfList, &sizeOfPrintfList);
if (error != ACL_SUCCESS) {
return false;
@@ -213,7 +213,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
aclMetadata md;
md.enqueue_kernel = false;
size_t sizeOfDeviceEnqueue = sizeof(md.enqueue_kernel);
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_DEVICE_ENQUEUE,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_DEVICE_ENQUEUE,
openClKernelName.c_str(), &md.enqueue_kernel, &sizeOfDeviceEnqueue);
if (error != ACL_SUCCESS) {
return false;
@@ -222,7 +222,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
md.kernel_index = -1;
size_t sizeOfIndex = sizeof(md.kernel_index);
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_KERNEL_INDEX,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_KERNEL_INDEX,
openClKernelName.c_str(), &md.kernel_index, &sizeOfIndex);
if (error != ACL_SUCCESS) {
return false;
@@ -230,7 +230,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
index_ = md.kernel_index;
size_t sizeOfWavesPerSimdHint = sizeof(workGroupInfo_.wavesPerSimdHint_);
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_WAVES_PER_SIMD_HINT,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_WAVES_PER_SIMD_HINT,
openClKernelName.c_str(), &workGroupInfo_.wavesPerSimdHint_,
&sizeOfWavesPerSimdHint);
if (error != ACL_SUCCESS) {
@@ -240,7 +240,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
waveLimiter_.enable();
size_t sizeOfWorkGroupSizeHint = sizeof(workGroupInfo_.compileSizeHint_);
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE_HINT,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_WORK_GROUP_SIZE_HINT,
openClKernelName.c_str(), workGroupInfo_.compileSizeHint_,
&sizeOfWorkGroupSizeHint);
if (error != ACL_SUCCESS) {
@@ -248,7 +248,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
}
size_t sizeOfVecTypeHint;
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_VEC_TYPE_HINT,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_VEC_TYPE_HINT,
openClKernelName.c_str(), NULL, &sizeOfVecTypeHint);
if (error != ACL_SUCCESS) {
return false;
@@ -259,7 +259,7 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
if (NULL == VecTypeHint) {
return false;
}
error = aclQueryInfo(dev().compiler(), prog().binaryElf(), RT_VEC_TYPE_HINT,
error = aclQueryInfo(palNullDevice().compiler(), prog().binaryElf(), RT_VEC_TYPE_HINT,
openClKernelName.c_str(), VecTypeHint, &sizeOfVecTypeHint);
if (error != ACL_SUCCESS) {
return false;
@@ -273,8 +273,6 @@ bool HSAILKernel::init(amd::hsa::loader::Symbol* sym, bool finalize) {
return true;
}
const Device& HSAILKernel::dev() const { return reinterpret_cast<const Device&>(dev_); }
const HSAILProgram& HSAILKernel::prog() const {
return reinterpret_cast<const HSAILProgram&>(prog_);
}
@@ -476,7 +474,7 @@ bool LightningKernel::init() {
workGroupInfo()->usedSGPRs_, workGroupInfo()->usedVGPRs_);
// Copy wavefront size
workGroupInfo_.wavefrontSize_ = dev().info().wavefrontWidth_;
workGroupInfo_.wavefrontSize_ = device().info().wavefrontWidth_;
if (workGroupInfo_.size_ == 0) {
return false;