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

[ML-31] Print time duration for each PCA step #32

Merged
merged 1 commit into from
Mar 5, 2021
Merged

[ML-31] Print time duration for each PCA step #32

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
21 changes: 17 additions & 4 deletions mllib-dal/src/main/native/PCADALImpl.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,6 +41,8 @@ JNIEXPORT jlong JNICALL Java_org_apache_spark_ml_feature_PCADALImpl_cPCATrainDAL
int nThreadsNew = services::Environment::getInstance()->getNumberOfThreads();
cout << "oneDAL (native): Number of threads used: " << nThreadsNew << endl;

auto t1 = std::chrono::high_resolution_clock::now();

pca::Distributed<step1Local, algorithmFPType, pca::svdDense> localAlgorithm;

/* Set the input data set to the algorithm */
Expand All @@ -49,6 +51,12 @@ JNIEXPORT jlong JNICALL Java_org_apache_spark_ml_feature_PCADALImpl_cPCATrainDAL
/* Compute PCA decomposition */
localAlgorithm.compute();

auto t2 = std::chrono::high_resolution_clock::now();
auto duration = std::chrono::duration_cast<std::chrono::seconds>( t2 - t1 ).count();
std::cout << "PCA (native): local step took " << duration << " secs" << std::endl;

t1 = std::chrono::high_resolution_clock::now();

/* Serialize partial results required by step 2 */
services::SharedPtr<byte> serializedData;
InputDataArchive dataArch;
Expand All @@ -60,26 +68,31 @@ JNIEXPORT jlong JNICALL Java_org_apache_spark_ml_feature_PCADALImpl_cPCATrainDAL
byte* nodeResults = new byte[perNodeArchLength];
dataArch.copyArchiveToArray(nodeResults, perNodeArchLength);

t2 = std::chrono::high_resolution_clock::now();

duration = std::chrono::duration_cast<std::chrono::seconds>( t2 - t1 ).count();
std::cout << "PCA (native): serializing partial results took " << duration << " secs" << std::endl;

vector<size_t> recv_counts(comm_size * perNodeArchLength);
for (int i = 0; i < comm_size; i++) recv_counts[i] = perNodeArchLength;

cout << "PCA (native): ccl_allgatherv receiving " << perNodeArchLength * nBlocks << " bytes" << endl;

auto t1 = std::chrono::high_resolution_clock::now();
t1 = std::chrono::high_resolution_clock::now();

/* Transfer partial results to step 2 on the root node */
// MPI_Gather(nodeResults, perNodeArchLength, MPI_CHAR, serializedData.get(),
// perNodeArchLength, MPI_CHAR, ccl_root, MPI_COMM_WORLD);
ccl::allgatherv(nodeResults, perNodeArchLength, serializedData.get(), recv_counts,
ccl::datatype::uint8, comm).wait();

auto t2 = std::chrono::high_resolution_clock::now();
t2 = std::chrono::high_resolution_clock::now();

auto duration = std::chrono::duration_cast<std::chrono::seconds>( t2 - t1 ).count();
duration = std::chrono::duration_cast<std::chrono::seconds>( t2 - t1 ).count();
std::cout << "PCA (native): ccl_allgatherv took " << duration << " secs" << std::endl;

if (rankId == ccl_root) {
auto t1 = std::chrono::high_resolution_clock::now();
auto t1 = std::chrono::high_resolution_clock::now();

/* Create an algorithm for principal component analysis using the svdDense method
* on the master node */
Expand Down