Add utility to format ninja-log build times

build.sh

@@ -187,6 +187,14 @@ if buildAll || hasArg libcudf; then
 
     cmake --build . -j${PARALLEL_LEVEL} ${VERBOSE_FLAG}
 
+    # Record build times
+    if [[ -f "${LIB_BUILD_DIR}/.ninja_log" ]]; then
+        echo "Formatting build times"
+        python ${REPODIR}/ci/utils/sort_ninja_log.py ${LIB_BUILD_DIR}/.ninja_log > ${LIB_BUILD_DIR}/ninja_log.csv
+        echo "===Top 50 build time offenders==="
+        cat ${LIB_BUILD_DIR}/ninja_log.csv | head -n 50
+    fi
+
     if [[ ${INSTALL_TARGET} != "" ]]; then
         cmake --build . -j${PARALLEL_LEVEL} --target install ${VERBOSE_FLAG}
     fi

ci/gpu/build.sh

@@ -127,6 +127,14 @@ if [[ -z "$PROJECT_FLASH" || "$PROJECT_FLASH" == "0" ]]; then
         "$WORKSPACE/build.sh" clean libcudf cudf dask_cudf libcudf_kafka cudf_kafka benchmarks tests -l --ptds
     fi
 
+    ###
+    echo "Checking for build time log $WORKSPACE/cpp/build/ninja_log.csv"
+    gpuci_logger "Checking for build time logger $WORKSPACE/cpp/build/ninja_log.csv"
+    if [[ -f "$WORKSPACE/cpp/build/ninja_log.csv" ]]; then
+        echo "Copying build time results"
+        cp "$WORKSPACE/cpp/build/ninja_log.csv" "$WORKSPACE/test-results/."
+    fi
+
     ################################################################################
     # TEST - Run GoogleTest
     ################################################################################
@@ -166,16 +174,24 @@ else
     gpuci_logger "Check GPU usage"
     nvidia-smi
 
-    gpuci_logger "GoogleTests"
     set -x
     cd $LIB_BUILD_DIR
 
+    gpuci_logger "GoogleTests"
+
     for gt in gtests/* ; do
         test_name=$(basename ${gt})
         echo "Running GoogleTest $test_name"
         ${gt} --gtest_output=xml:"$WORKSPACE/test-results/"
     done
 
+    ###
+    echo "Checking for build time log $LIB_BUILD_DIR/ninja_log.csv"
+    if [[ -f "$LIB_BUILD_DIR/ninja_log.csv" ]]; then
+        gpuci_logger "Copying build time results"
+        cp "$LIB_BUILD_DIR/ninja_log.csv" "$WORKSPACE/test-results/."
+    fi
+
     CUDF_CONDA_FILE=`find ${CONDA_ARTIFACT_PATH} -name "libcudf-*.tar.bz2"`
     CUDF_CONDA_FILE=`basename "$CUDF_CONDA_FILE" .tar.bz2` #get filename without extension
     CUDF_CONDA_FILE=${CUDF_CONDA_FILE//-/=} #convert to conda install

ci/utils/sort_ninja_log.py

@@ -0,0 +1,33 @@
+#
+# Copyright (c) 2021, NVIDIA CORPORATION.
+#
+import os
+import sys
+
+log_file = ".ninja_log"
+if len(sys.argv) > 1:
+    log_file = sys.argv[1]
+
+# build a map of the log entries
+entries = {}
+with open(log_file, "r") as log:
+    for line in log:
+        entry = line.split()
+        if len(entry) > 4:
+            elapsed = int(entry[1]) - int(entry[0])
+            entries[entry[3]] = elapsed
+
+# check file could be loaded
+if len(entries) == 0:
+    print("Could not parse", log_file)
+    exit()
+
+# sort the keys by build time (descending order)
+keys = list(entries.keys())
+sl = sorted(keys, key=lambda k: entries[k], reverse=True)
+
+# output results in CSV format
+print("time,file")
+for key in sl:
+    print(entries[key], key, sep=",")
+