Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

adjust output and parameters in gpu autotuner #5779

Open
wants to merge 1 commit into
base: master
Choose a base branch
from
+11 −11
Open

adjust output and parameters in gpu autotuner #5779

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 5 additions & 5 deletions opm/simulators/linalg/gpuistl/GpuDILU.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -270,12 +270,12 @@ GpuDILU<M, X, Y, l>::tuneThreadBlockSizes()
auto tuneMoveThreadBlockSizeInUpdate = [this](int moveThreadBlockSize){
this->update(moveThreadBlockSize, m_DILUFactorizationThreadBlockSize);
};
m_moveThreadBlockSize = detail::tuneThreadBlockSize(tuneMoveThreadBlockSizeInUpdate, "Kernel moving data to reordered matrix");
m_moveThreadBlockSize = detail::tuneThreadBlockSize(tuneMoveThreadBlockSizeInUpdate, "(in DILU update) Move data to reordered matrix");

auto tuneFactorizationThreadBlockSizeInUpdate = [this](int factorizationThreadBlockSize){
this->update(m_moveThreadBlockSize, factorizationThreadBlockSize);
};
m_DILUFactorizationThreadBlockSize = detail::tuneThreadBlockSize(tuneFactorizationThreadBlockSizeInUpdate, "Kernel computing DILU factorization");
m_DILUFactorizationThreadBlockSize = detail::tuneThreadBlockSize(tuneFactorizationThreadBlockSizeInUpdate, "(in DILU update) DILU factorization");

// tune the thread-block size of the apply
GpuVector<field_type> tmpV(m_gpuMatrix.N() * m_gpuMatrix.blockSize());
Expand All @@ -285,12 +285,12 @@ GpuDILU<M, X, Y, l>::tuneThreadBlockSizes()
auto tuneLowerSolveThreadBlockSizeInApply = [this, &tmpV, &tmpD](int lowerSolveThreadBlockSize){
this->apply(tmpV, tmpD, lowerSolveThreadBlockSize, m_DILUFactorizationThreadBlockSize);
};
m_lowerSolveThreadBlockSize = detail::tuneThreadBlockSize(tuneLowerSolveThreadBlockSizeInApply, "Kernel computing a lower triangular solve for a level set");
m_lowerSolveThreadBlockSize = detail::tuneThreadBlockSize(tuneLowerSolveThreadBlockSizeInApply, "(in DILU apply) Triangular lower solve");

auto tuneUpperSolveThreadBlockSizeInApply = [this, &tmpV, &tmpD](int upperSolveThreadBlockSize){
this->apply(tmpV, tmpD, m_moveThreadBlockSize, upperSolveThreadBlockSize);
this->apply(tmpV, tmpD, m_lowerSolveThreadBlockSize, upperSolveThreadBlockSize);
};
m_upperSolveThreadBlockSize = detail::tuneThreadBlockSize(tuneUpperSolveThreadBlockSizeInApply, "Kernel computing an upper triangular solve for a level set");
m_upperSolveThreadBlockSize = detail::tuneThreadBlockSize(tuneUpperSolveThreadBlockSizeInApply, "(in DILU apply) Triangular upper solve");
}

} // namespace Opm::gpuistl
Expand Down
10 changes: 5 additions & 5 deletions opm/simulators/linalg/gpuistl/OpmGpuILU0.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -331,13 +331,13 @@ OpmGpuILU0<M, X, Y, l>::tuneThreadBlockSizes()
auto tuneMoveThreadBlockSizeInUpdate
= [this](int moveThreadBlockSize) { this->update(moveThreadBlockSize, m_ILU0FactorizationThreadBlockSize); };
m_moveThreadBlockSize
= detail::tuneThreadBlockSize(tuneMoveThreadBlockSizeInUpdate, "Kernel moving data to reordered matrix");
= detail::tuneThreadBlockSize(tuneMoveThreadBlockSizeInUpdate, "(in ILU update) Move data to reordered matrix");

auto tuneFactorizationThreadBlockSizeInUpdate = [this](int factorizationThreadBlockSize) {
this->update(m_moveThreadBlockSize, factorizationThreadBlockSize);
};
m_ILU0FactorizationThreadBlockSize
= detail::tuneThreadBlockSize(tuneFactorizationThreadBlockSizeInUpdate, "Kernel computing ILU0 factorization");
= detail::tuneThreadBlockSize(tuneFactorizationThreadBlockSizeInUpdate, "(in ILU update) ILU factorization");

// tune the thread-block size of the apply
GpuVector<field_type> tmpV(m_gpuMatrix.N() * m_gpuMatrix.blockSize());
Expand All @@ -348,13 +348,13 @@ OpmGpuILU0<M, X, Y, l>::tuneThreadBlockSizes()
this->apply(tmpV, tmpD, lowerSolveThreadBlockSize, m_ILU0FactorizationThreadBlockSize);
};
m_lowerSolveThreadBlockSize = detail::tuneThreadBlockSize(
tuneLowerSolveThreadBlockSizeInApply, "Kernel computing a lower triangular solve for a level set");
tuneLowerSolveThreadBlockSizeInApply, "(in ILU apply) Triangular lower solve");

auto tuneUpperSolveThreadBlockSizeInApply = [this, &tmpV, &tmpD](int upperSolveThreadBlockSize) {
this->apply(tmpV, tmpD, m_moveThreadBlockSize, upperSolveThreadBlockSize);
this->apply(tmpV, tmpD, m_lowerSolveThreadBlockSize, upperSolveThreadBlockSize);
};
m_upperSolveThreadBlockSize = detail::tuneThreadBlockSize(
tuneUpperSolveThreadBlockSizeInApply, "Kernel computing an upper triangular solve for a level set");
tuneUpperSolveThreadBlockSizeInApply, "(in ILU apply) Triangular upper solve");
}

} // namespace Opm::gpuistl
Expand Down
2 changes: 1 addition & 1 deletion opm/simulators/linalg/gpuistl/detail/autotuner.hpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,7 +83,7 @@ tuneThreadBlockSize(func& f, std::string descriptionOfFunction)
}

OpmLog::info(
fmt::format("{}: Tuned Blocksize: {} (fastest runtime: {}).", descriptionOfFunction, bestBlockSize, bestTime));
fmt::format("[Kernel tuning completed] {}: Tuned Blocksize = {}, Fastest Runtime = {}ms.", descriptionOfFunction, bestBlockSize, bestTime));

return bestBlockSize;
}
Expand Down