P4 to Git Change 1076493 by yaxunl@yaxunl_stg_win50 on 2014/09/12 14:45:24

ECR #377625 - AMDIL Function support: Calculate total private memory usage by a kernel including memory used by called functions.

	This cannot be done by IPA since stack size is known only after register allocation due to potential register spill, but MachineFunctionAnalysis cannot persist after CGSCC pass with current LLVM version.

	This change adds private memory usage metadata for non-kernel functions. The total private memory usage by a kernel is calculated when AMDIL is split for different kernels. BIF will contain total private memory size.

Affected files ...

... //depot/stg/opencl/drivers/opencl/compiler/lib/utils/amdilUtils.cpp#1 add
... //depot/stg/opencl/drivers/opencl/compiler/lib/utils/amdilUtils.hpp#1 add
... //depot/stg/opencl/drivers/opencl/compiler/llvm/lib/Target/AMDIL/AMDILKernelManager.cpp#451 edit
... //depot/stg/opencl/drivers/opencl/compiler/llvm/lib/Target/AMDIL/AMDILKernelManager.h#51 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprogram.cpp#175 edit
... //depot/stg/opencl/drivers/opencl/runtime/device/gpu/gpuprogram.hpp#54 edit
Этот коммит содержится в:
foreman
2014-09-12 15:02:32 -04:00
родитель fea6100aa9
Коммит 42f4b2af97
4 изменённых файлов: 53 добавлений и 4 удалений
+22 -2
Просмотреть файл
@@ -5,6 +5,7 @@
#include "os/os.hpp"
#include "utils/flags.hpp"
#include "include/aclTypes.h"
#include "utils/amdilUtils.hpp"
#include "utils/bif_section_labels.hpp"
#include "device/gpu/gpuprogram.hpp"
#include "device/gpu/gpublit.hpp"
@@ -232,6 +233,21 @@ NullProgram::isCalled(const ILFunc* base, const ILFunc* func)
return false;
}
uint
ILFunc::totalHwPrivateUsage() {
if (totalHwPrivateSize_ >= 0)
return totalHwPrivateSize_;
uint maxChildUsage = 0;
for (size_t i = 0; i < calls_.size(); ++i) {
uint childUsage = calls_[i]->totalHwPrivateUsage();
if (childUsage > maxChildUsage)
maxChildUsage = childUsage;
}
totalHwPrivateSize_ = hwPrivateSize_ + maxChildUsage;
return totalHwPrivateSize_;
}
void
NullProgram::patchMain(std::string& kernel, uint index)
{
@@ -540,7 +556,6 @@ NullProgram::linkImpl(amd::option::Options* options)
// Accumulate all emulated local and private sizes,
// necessary for the kernel execution
initData.localSize_ += func->localSize_;
initData.privateSize_ += func->privateSize_;
// Accumulate all HW local and private sizes,
// necessary for the kernel execution
@@ -548,6 +563,9 @@ NullProgram::linkImpl(amd::option::Options* options)
initData.hwPrivateSize_ += func->hwPrivateSize_;
initData.flags_ |= func->flags_;
}
initData.privateSize_ = baseFunc->totalHwPrivateUsage();
amdilUtils::changePrivateUAVLength(kernel,
initData.privateSize_);
// Create a GPU kernel
bool created;
@@ -913,7 +931,6 @@ NullProgram::linkImpl(const std::vector<device::Program*>& inputPrograms,
// Accumulate all emulated local and private sizes,
// necessary for the kernel execution
initData.localSize_ += func->localSize_;
initData.privateSize_ += func->privateSize_;
// Accumulate all HW local and private sizes,
// necessary for the kernel execution
@@ -921,6 +938,9 @@ NullProgram::linkImpl(const std::vector<device::Program*>& inputPrograms,
initData.hwPrivateSize_ += func->hwPrivateSize_;
initData.flags_ |= func->flags_;
}
initData.privateSize_ = baseFunc->totalHwPrivateUsage();
amdilUtils::changePrivateUAVLength(kernel,
initData.privateSize_);
// Create a GPU kernel
bool created;