mirror of
https://github.com/ceres-solver/ceres-solver.git
synced 2026-08-29 16:40:38 +08:00
b158515089
Main focus of this change is to parallelize remaining operations (most of them are operations on vectors) in code-path utilized with iterative Schur complement. Parallelization is handled using lazy evaluation of Eigen expressions. On linux pc with intel 8176 processor parallelization of vector operations has the following effect: Running ./bin/parallel_vector_operations_benchmark Run on (112 X 3200.32 MHz CPU s) CPU Caches: L1 Data 32 KiB (x56) L1 Instruction 32 KiB (x56) L2 Unified 1024 KiB (x56) L3 Unified 39424 KiB (x2) Load Average: 3.30, 8.41, 11.82 ----------------------------------- Benchmark Time ----------------------------------- SetZero 10009532 ns SetZeroParallel/1 10024139 ns ... SetZeroParallel/16 877606 ns Negate 4978856 ns NegateParallel/1 5145413 ns ... NegateParallel/16 721823 ns Assign 10731408 ns AssignParallel/1 10749944 ns ... AssignParallel/16 1829381 ns D2X 15214399 ns D2XParallel/1 15623245 ns ... D2XParallel/16 2687060 ns DivideSqrt 8220050 ns DivideSqrtParallel/1 9088467 ns ... DivideSqrtParallel/16 905569 ns Clamp 3502010 ns ClampParallel/1 4507897 ns ... ClampParallel/16 759576 ns Norm 4426782 ns NormParallel/1 4442805 ns ... NormParallel/16 430290 ns Dot 9023276 ns DotParallel/1 9031304 ns ... DotParallel/16 1157267 ns Axpby 14608289 ns AxpbyParallel/1 14570825 ns ... AxpbyParallel/16 2672220 ns ----------------------------------- Multi-threading of vector operations in ISC and program evaluation results into the following improvement: Running ./bin/evaluation_benchmark -------------------------------------------------------------------------------------- Benchmark this2fd81de-------------------------------------------------------------------------------------- Residuals<problem-13682-4456117-pre.txt>/1 4136 ms 4292 ms Residuals<problem-13682-4456117-pre.txt>/2 2919 ms 2670 ms Residuals<problem-13682-4456117-pre.txt>/4 2065 ms 2198 ms Residuals<problem-13682-4456117-pre.txt>/8 1458 ms 1609 ms Residuals<problem-13682-4456117-pre.txt>/16 1152 ms 1227 ms ResidualsAndJacobian<problem-13682-4456117-pre.txt>/1 19759 ms 20084 ms ResidualsAndJacobian<problem-13682-4456117-pre.txt>/2 10921 ms 10977 ms ResidualsAndJacobian<problem-13682-4456117-pre.txt>/4 6220 ms 6941 ms ResidualsAndJacobian<problem-13682-4456117-pre.txt>/8 3490 ms 4398 ms ResidualsAndJacobian<problem-13682-4456117-pre.txt>/16 2277 ms 3172 ms Plus<problem-13682-4456117-pre.txt>/1 339 ms 322 ms Plus<problem-13682-4456117-pre.txt>/2 220 ms Plus<problem-13682-4456117-pre.txt>/4 128 ms Plus<problem-13682-4456117-pre.txt>/8 78.0 ms Plus<problem-13682-4456117-pre.txt>/16 49.8 ms ISCRightMultiplyAndAccumulate<problem-13682-4456117-pre.txt>/1 2434 ms 2478 ms ISCRightMultiplyAndAccumulate<problem-13682-4456117-pre.txt>/2 2706 ms 2688 ms ISCRightMultiplyAndAccumulate<problem-13682-4456117-pre.txt>/4 1430 ms 1548 ms ISCRightMultiplyAndAccumulate<problem-13682-4456117-pre.txt>/8 742 ms 883 ms ISCRightMultiplyAndAccumulate<problem-13682-4456117-pre.txt>/16 438 ms 555 ms ISCRightMultiplyAndAccumulateDiag<problem-13682-4456117-pre.txt>/1 2438 ms 2481 ms ISCRightMultiplyAndAccumulateDiag<problem-13682-4456117-pre.txt>/2 2565 ms 2790 ms ISCRightMultiplyAndAccumulateDiag<problem-13682-4456117-pre.txt>/4 1434 ms 1551 ms ISCRightMultiplyAndAccumulateDiag<problem-13682-4456117-pre.txt>/8 765 ms 892 ms ISCRightMultiplyAndAccumulateDiag<problem-13682-4456117-pre.txt>/16 435 ms 559 ms JacobianSquaredColumnNorm<problem-13682-4456117-pre.txt>/1 1278 ms JacobianSquaredColumnNorm<problem-13682-4456117-pre.txt>/2 1555 ms JacobianSquaredColumnNorm<problem-13682-4456117-pre.txt>/4 833 ms JacobianSquaredColumnNorm<problem-13682-4456117-pre.txt>/8 459 ms JacobianSquaredColumnNorm<problem-13682-4456117-pre.txt>/16 250 ms JacobianScaleColumns<problem-13682-4456117-pre.txt>/1 1468 ms JacobianScaleColumns<problem-13682-4456117-pre.txt>/2 1871 ms JacobianScaleColumns<problem-13682-4456117-pre.txt>/4 957 ms JacobianScaleColumns<problem-13682-4456117-pre.txt>/8 528 ms JacobianScaleColumns<problem-13682-4456117-pre.txt>/16 294 ms End-to-end improvements with bundle_adjuster invoked with ./bin/bundle_adjuster --num_threads 28 --num_iterations 40 \ --linear_solver iterative_schur \ --preconditioner jacobi --input --------------------------------------------- Problem this2fd81de--------------------------------------------- problem-13682-4456117-pre.txt 508.6 892.7 problem-1778-993923-pre.txt 763.8 1129.9 problem-1723-156502-pre.txt 6.3 14.4 problem-356-226730-pre.txt 76.3 116.2 problem-257-65132-pre.txt 38.6 52.0 Change-Id: Ie31cc5015f13fa479c16ffb5ce48c9b880990d49
273 lines
9.7 KiB
C++
273 lines
9.7 KiB
C++
// Ceres Solver - A fast non-linear least squares minimizer
|
|
// Copyright 2022 Google Inc. All rights reserved.
|
|
// http://ceres-solver.org/
|
|
//
|
|
// Redistribution and use in source and binary forms, with or without
|
|
// modification, are permitted provided that the following conditions are met:
|
|
//
|
|
// * Redistributions of source code must retain the above copyright notice,
|
|
// this list of conditions and the following disclaimer.
|
|
// * Redistributions in binary form must reproduce the above copyright notice,
|
|
// this list of conditions and the following disclaimer in the documentation
|
|
// and/or other materials provided with the distribution.
|
|
// * Neither the name of Google Inc. nor the names of its contributors may be
|
|
// used to endorse or promote products derived from this software without
|
|
// specific prior written permission.
|
|
//
|
|
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
|
|
// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|
// ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
|
|
// LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
|
|
// CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
|
|
// SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
|
|
// INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
|
|
// CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
|
|
// ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
|
|
// POSSIBILITY OF SUCH DAMAGE.
|
|
//
|
|
// Author: sameeragarwal@google.com (Sameer Agarwal)
|
|
|
|
#include "ceres/partitioned_matrix_view.h"
|
|
|
|
#include <memory>
|
|
#include <random>
|
|
#include <sstream>
|
|
#include <string>
|
|
#include <vector>
|
|
|
|
#include "ceres/block_structure.h"
|
|
#include "ceres/casts.h"
|
|
#include "ceres/internal/eigen.h"
|
|
#include "ceres/linear_least_squares_problems.h"
|
|
#include "ceres/sparse_matrix.h"
|
|
#include "glog/logging.h"
|
|
#include "gtest/gtest.h"
|
|
|
|
namespace ceres {
|
|
namespace internal {
|
|
|
|
const double kEpsilon = 1e-14;
|
|
|
|
// Param = <problem_id, num_threads>
|
|
using Param = ::testing::tuple<int, int>;
|
|
|
|
static std::string ParamInfoToString(testing::TestParamInfo<Param> info) {
|
|
Param param = info.param;
|
|
std::stringstream ss;
|
|
ss << ::testing::get<0>(param) << "_" << ::testing::get<1>(param);
|
|
return ss.str();
|
|
}
|
|
|
|
class PartitionedMatrixViewTest : public ::testing::TestWithParam<Param> {
|
|
protected:
|
|
void SetUp() final {
|
|
const int problem_id = ::testing::get<0>(GetParam());
|
|
const int num_threads = ::testing::get<1>(GetParam());
|
|
auto problem = CreateLinearLeastSquaresProblemFromId(problem_id);
|
|
CHECK(problem != nullptr);
|
|
A_ = std::move(problem->A);
|
|
auto block_sparse = down_cast<BlockSparseMatrix*>(A_.get());
|
|
|
|
options_.num_threads = num_threads;
|
|
options_.context = &context_;
|
|
options_.elimination_groups.push_back(problem->num_eliminate_blocks);
|
|
pmv_ = PartitionedMatrixViewBase::Create(options_, *block_sparse);
|
|
|
|
LinearSolver::Options options_single_threaded = options_;
|
|
options_single_threaded.num_threads = 1;
|
|
pmv_single_threaded_ =
|
|
PartitionedMatrixViewBase::Create(options_, *block_sparse);
|
|
|
|
EXPECT_EQ(pmv_->num_col_blocks_e(), problem->num_eliminate_blocks);
|
|
EXPECT_EQ(pmv_->num_col_blocks_f(),
|
|
block_sparse->block_structure()->cols.size() -
|
|
problem->num_eliminate_blocks);
|
|
EXPECT_EQ(pmv_->num_cols(), A_->num_cols());
|
|
EXPECT_EQ(pmv_->num_rows(), A_->num_rows());
|
|
}
|
|
|
|
double RandDouble() { return distribution_(prng_); }
|
|
|
|
LinearSolver::Options options_;
|
|
ContextImpl context_;
|
|
std::unique_ptr<LinearLeastSquaresProblem> problem_;
|
|
std::unique_ptr<SparseMatrix> A_;
|
|
std::unique_ptr<PartitionedMatrixViewBase> pmv_;
|
|
std::unique_ptr<PartitionedMatrixViewBase> pmv_single_threaded_;
|
|
std::mt19937 prng_;
|
|
std::uniform_real_distribution<double> distribution_ =
|
|
std::uniform_real_distribution<double>(0.0, 1.0);
|
|
};
|
|
|
|
TEST_P(PartitionedMatrixViewTest, RightMultiplyAndAccumulateE) {
|
|
Vector x1(pmv_->num_cols_e());
|
|
Vector x2(pmv_->num_cols());
|
|
x2.setZero();
|
|
|
|
for (int i = 0; i < pmv_->num_cols_e(); ++i) {
|
|
x1(i) = x2(i) = RandDouble();
|
|
}
|
|
|
|
Vector expected = Vector::Zero(pmv_->num_rows());
|
|
A_->RightMultiplyAndAccumulate(x2.data(), expected.data());
|
|
|
|
Vector actual = Vector::Zero(pmv_->num_rows());
|
|
pmv_->RightMultiplyAndAccumulateE(x1.data(), actual.data());
|
|
|
|
for (int i = 0; i < pmv_->num_rows(); ++i) {
|
|
EXPECT_NEAR(actual(i), expected(i), kEpsilon);
|
|
}
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, RightMultiplyAndAccumulateF) {
|
|
Vector x1(pmv_->num_cols_f());
|
|
Vector x2(pmv_->num_cols());
|
|
x2.setZero();
|
|
|
|
for (int i = 0; i < pmv_->num_cols_f(); ++i) {
|
|
x1(i) = x2(i + pmv_->num_cols_e()) = RandDouble();
|
|
}
|
|
|
|
Vector actual = Vector::Zero(pmv_->num_rows());
|
|
pmv_->RightMultiplyAndAccumulateF(x1.data(), actual.data());
|
|
|
|
Vector expected = Vector::Zero(pmv_->num_rows());
|
|
A_->RightMultiplyAndAccumulate(x2.data(), expected.data());
|
|
|
|
for (int i = 0; i < pmv_->num_rows(); ++i) {
|
|
EXPECT_NEAR(actual(i), expected(i), kEpsilon);
|
|
}
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, LeftMultiplyAndAccumulate) {
|
|
Vector x = Vector::Zero(pmv_->num_rows());
|
|
for (int i = 0; i < pmv_->num_rows(); ++i) {
|
|
x(i) = RandDouble();
|
|
}
|
|
Vector x_pre = x;
|
|
|
|
Vector expected = Vector::Zero(pmv_->num_cols());
|
|
Vector e_actual = Vector::Zero(pmv_->num_cols_e());
|
|
Vector f_actual = Vector::Zero(pmv_->num_cols_f());
|
|
|
|
A_->LeftMultiplyAndAccumulate(x.data(), expected.data());
|
|
pmv_->LeftMultiplyAndAccumulateE(x.data(), e_actual.data());
|
|
pmv_->LeftMultiplyAndAccumulateF(x.data(), f_actual.data());
|
|
|
|
for (int i = 0; i < pmv_->num_cols(); ++i) {
|
|
EXPECT_NEAR(expected(i),
|
|
(i < pmv_->num_cols_e()) ? e_actual(i)
|
|
: f_actual(i - pmv_->num_cols_e()),
|
|
kEpsilon);
|
|
}
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, BlockDiagonalFtF) {
|
|
std::unique_ptr<BlockSparseMatrix> block_diagonal_ff(
|
|
pmv_->CreateBlockDiagonalFtF());
|
|
const auto bs_diagonal = block_diagonal_ff->block_structure();
|
|
const int num_rows = pmv_->num_rows();
|
|
const int num_cols_f = pmv_->num_cols_f();
|
|
const int num_cols_e = pmv_->num_cols_e();
|
|
const int num_col_blocks_f = pmv_->num_col_blocks_f();
|
|
const int num_col_blocks_e = pmv_->num_col_blocks_e();
|
|
|
|
CHECK_EQ(block_diagonal_ff->num_rows(), num_cols_f);
|
|
CHECK_EQ(block_diagonal_ff->num_cols(), num_cols_f);
|
|
|
|
EXPECT_EQ(bs_diagonal->cols.size(), num_col_blocks_f);
|
|
EXPECT_EQ(bs_diagonal->rows.size(), num_col_blocks_f);
|
|
|
|
Matrix EF;
|
|
A_->ToDenseMatrix(&EF);
|
|
const auto F = EF.topRightCorner(num_rows, num_cols_f);
|
|
|
|
Matrix expected_FtF = F.transpose() * F;
|
|
Matrix actual_FtF;
|
|
block_diagonal_ff->ToDenseMatrix(&actual_FtF);
|
|
|
|
// FtF might be not block-diagonal
|
|
auto bs = down_cast<BlockSparseMatrix*>(A_.get())->block_structure();
|
|
for (int i = 0; i < num_col_blocks_f; ++i) {
|
|
const auto col_block_f = bs->cols[num_col_blocks_e + i];
|
|
const int block_size = col_block_f.size;
|
|
const int block_pos = col_block_f.position - num_cols_e;
|
|
const auto cell_expected =
|
|
expected_FtF.block(block_pos, block_pos, block_size, block_size);
|
|
auto cell_actual =
|
|
actual_FtF.block(block_pos, block_pos, block_size, block_size);
|
|
cell_actual -= cell_expected;
|
|
EXPECT_NEAR(cell_actual.norm(), 0., kEpsilon);
|
|
}
|
|
// There should be nothing remaining outside block-diagonal
|
|
EXPECT_NEAR(actual_FtF.norm(), 0., kEpsilon);
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, BlockDiagonalEtE) {
|
|
std::unique_ptr<BlockSparseMatrix> block_diagonal_ee(
|
|
pmv_->CreateBlockDiagonalEtE());
|
|
const CompressedRowBlockStructure* bs = block_diagonal_ee->block_structure();
|
|
const int num_rows = pmv_->num_rows();
|
|
const int num_cols_e = pmv_->num_cols_e();
|
|
const int num_col_blocks_e = pmv_->num_col_blocks_e();
|
|
|
|
CHECK_EQ(block_diagonal_ee->num_rows(), num_cols_e);
|
|
CHECK_EQ(block_diagonal_ee->num_cols(), num_cols_e);
|
|
|
|
EXPECT_EQ(bs->cols.size(), num_col_blocks_e);
|
|
EXPECT_EQ(bs->rows.size(), num_col_blocks_e);
|
|
|
|
Matrix EF;
|
|
A_->ToDenseMatrix(&EF);
|
|
const auto E = EF.topLeftCorner(num_rows, num_cols_e);
|
|
|
|
Matrix expected_EtE = E.transpose() * E;
|
|
Matrix actual_EtE;
|
|
block_diagonal_ee->ToDenseMatrix(&actual_EtE);
|
|
|
|
EXPECT_NEAR((expected_EtE - actual_EtE).norm(), 0., kEpsilon);
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, UpdateBlockDiagonalEtE) {
|
|
std::unique_ptr<BlockSparseMatrix> block_diagonal_ete(
|
|
pmv_->CreateBlockDiagonalEtE());
|
|
const int num_cols = pmv_->num_cols_e();
|
|
|
|
Matrix multi_threaded(num_cols, num_cols);
|
|
pmv_->UpdateBlockDiagonalEtE(block_diagonal_ete.get());
|
|
block_diagonal_ete->ToDenseMatrix(&multi_threaded);
|
|
|
|
Matrix single_threaded(num_cols, num_cols);
|
|
pmv_single_threaded_->UpdateBlockDiagonalEtE(block_diagonal_ete.get());
|
|
block_diagonal_ete->ToDenseMatrix(&single_threaded);
|
|
|
|
EXPECT_NEAR((multi_threaded - single_threaded).norm(), 0., kEpsilon);
|
|
}
|
|
|
|
TEST_P(PartitionedMatrixViewTest, UpdateBlockDiagonalFtF) {
|
|
std::unique_ptr<BlockSparseMatrix> block_diagonal_ftf(
|
|
pmv_->CreateBlockDiagonalFtF());
|
|
const int num_cols = pmv_->num_cols_f();
|
|
|
|
Matrix multi_threaded(num_cols, num_cols);
|
|
pmv_->UpdateBlockDiagonalFtF(block_diagonal_ftf.get());
|
|
block_diagonal_ftf->ToDenseMatrix(&multi_threaded);
|
|
|
|
Matrix single_threaded(num_cols, num_cols);
|
|
pmv_single_threaded_->UpdateBlockDiagonalFtF(block_diagonal_ftf.get());
|
|
block_diagonal_ftf->ToDenseMatrix(&single_threaded);
|
|
|
|
EXPECT_NEAR((multi_threaded - single_threaded).norm(), 0., kEpsilon);
|
|
}
|
|
|
|
INSTANTIATE_TEST_SUITE_P(
|
|
ParallelProducts,
|
|
PartitionedMatrixViewTest,
|
|
::testing::Combine(::testing::Values(2, 4, 6),
|
|
::testing::Values(1, 2, 3, 4, 5, 6, 7, 8)),
|
|
ParamInfoToString);
|
|
|
|
} // namespace internal
|
|
} // namespace ceres
|