Add utility to format ninja-log build times

build.sh

@@ -172,6 +172,9 @@ if buildAll || hasArg libcudf; then
         echo "Building for *ALL* supported GPU architectures..."
     fi
 
+    # get the current count before the compile starts
+    FILES_IN_CCACHE=$(ccache -s | grep "files in cache")
+
     cmake -S $REPODIR/cpp -B ${LIB_BUILD_DIR} \
           -DCMAKE_INSTALL_PREFIX=${INSTALL_PREFIX} \
           ${CUDF_CMAKE_CUDA_ARCHITECTURES} \
@@ -185,7 +188,19 @@ if buildAll || hasArg libcudf; then
 
     cd ${LIB_BUILD_DIR}
 
+    compile_start=$(date +%s)
     cmake --build . -j${PARALLEL_LEVEL} ${VERBOSE_FLAG}
+    compile_end=$(date +%s)
+    compile_total=$(( compile_end - compile_start ))
+
+    # Record build times
+    if [[ -f "${LIB_BUILD_DIR}/.ninja_log" ]]; then
+        echo "Formatting build times"
+        python ${REPODIR}/cpp/scripts/sort_ninja_log.py ${LIB_BUILD_DIR}/.ninja_log --fmt xml > ${LIB_BUILD_DIR}/ninja_log.xml
+        message="$FILES_IN_CCACHE <p>CI build time = $compile_total"
+        echo $message
+        python ${REPODIR}/cpp/scripts/sort_ninja_log.py ${LIB_BUILD_DIR}/.ninja_log --fmt html --msg "$message" > ${LIB_BUILD_DIR}/ninja_log.html
+    fi
 
     if [[ ${INSTALL_TARGET} != "" ]]; then
         cmake --build . -j${PARALLEL_LEVEL} --target install ${VERBOSE_FLAG}

ci/gpu/build.sh

@@ -166,16 +166,26 @@ 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
 
+    # Copy libcudf build time results
+    echo "Checking for build time log $LIB_BUILD_DIR/ninja_log.html"
+    if [[ -f "$LIB_BUILD_DIR/ninja_log.html" ]]; then
+        gpuci_logger "Copying build time results"
+        cp "$LIB_BUILD_DIR/ninja_log.xml" "$WORKSPACE/test-results/buildtimes-junit.xml"
+        mkdir -p "$WORKSPACE/build-metrics"
+        cp "$LIB_BUILD_DIR/ninja_log.html" "$WORKSPACE/build-metrics/BuildMetrics.html"
+    fi
+
     ################################################################################
     # MEMCHECK - Run compute-sanitizer on GoogleTest (only in nightly builds)
     ################################################################################

cpp/cmake/thirdparty/get_dlpack.cmake

@@ -21,7 +21,8 @@ function(find_and_configure_dlpack VERSION)
     dlpack ${VERSION}
     GIT_REPOSITORY https://github.com/dmlc/dlpack.git
     GIT_TAG v${VERSION}
-    GIT_SHALLOW TRUE DOWNLOAD_ONLY TRUE
+    GIT_SHALLOW TRUE
+    DOWNLOAD_ONLY TRUE
     OPTIONS "BUILD_MOCK OFF"
   )
 

cpp/cmake/thirdparty/get_jitify.cmake

@@ -20,7 +20,8 @@ function(find_and_configure_jitify)
     jitify 2.0.0
     GIT_REPOSITORY https://github.com/rapidsai/jitify.git
     GIT_TAG cudf_0.19
-    GIT_SHALLOW TRUE DOWNLOAD_ONLY TRUE
+    GIT_SHALLOW TRUE
+    DOWNLOAD_ONLY TRUE
   )
   set(JITIFY_INCLUDE_DIR
       "${jitify_SOURCE_DIR}"

cpp/cmake/thirdparty/get_libcudacxx.cmake

@@ -17,8 +17,9 @@ function(find_and_configure_libcudacxx)
   include(${rapids-cmake-dir}/cpm/libcudacxx.cmake)
 
   rapids_cpm_libcudacxx(
-    BUILD_EXPORT_SET cudf-exports INSTALL_EXPORT_SET cudf-exports PATCH_COMMAND patch
-    --reject-file=- -p1 -N < ${CUDF_SOURCE_DIR}/cmake/libcudacxx.patch || true
+    BUILD_EXPORT_SET cudf-exports
+    INSTALL_EXPORT_SET cudf-exports PATCH_COMMAND patch --reject-file=- -p1 -N <
+                                    ${CUDF_SOURCE_DIR}/cmake/libcudacxx.patch || true
   )
 
   set(LIBCUDACXX_INCLUDE_DIR

cpp/scripts/sort_ninja_log.py

@@ -0,0 +1,117 @@
+#
+# Copyright (c) 2021, NVIDIA CORPORATION.
+#
+import argparse
+import os
+import sys
+import xml.etree.ElementTree as ET
+from xml.dom import minidom
+
+parser = argparse.ArgumentParser()
+parser.add_argument(
+    "log_file", type=str, default=".ninja_log", help=".ninja_log file"
+)
+parser.add_argument(
+    "--fmt", type=str, default="csv", help="output format (csv, xml, html)"
+)
+parser.add_argument(
+    "--msg",
+    type=str,
+    default=None,
+    help="optional message to include in html output",
+)
+args = parser.parse_args()
+
+log_file = args.log_file
+log_path = os.path.dirname(os.path.abspath(log_file))
+
+output_fmt = args.fmt
+
+# 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])
+            obj_file = entry[3]
+            file_size = (
+                os.path.getsize(os.path.join(log_path, obj_file))
+                if os.path.exists(obj_file)
+                else 0
+            )
+            entries[entry[3]] = (elapsed, file_size)
+
+# 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][0], reverse=True)
+
+if output_fmt == "xml":
+    # output results in XML format
+    root = ET.Element("testsuites")
+    testsuite = ET.Element(
+        "testsuite",
+        attrib={
+            "name": "build-time",
+            "tests": str(len(keys)),
+            "failures": str(0),
+            "errors": str(0),
+        },
+    )
+    root.append(testsuite)
+    for key in sl:
+        entry = entries[key]
+        elapsed = float(entry[0]) / 1000
+        item = ET.Element(
+            "testcase",
+            attrib={
+                "classname": "BuildTime",
+                "name": key,
+                "time": str(elapsed),
+            },
+        )
+        testsuite.append(item)
+
+    tree = ET.ElementTree(root)
+    xmlstr = minidom.parseString(ET.tostring(root)).toprettyxml(indent="   ")
+    print(xmlstr)
+
+elif output_fmt == "html":
+    # output results in HTML format
+    print("<html><head><title>Sorted Ninja Build Times</title>")
+    print("<style>", "table, th, td { border:1px solid black; }", "</style>")
+    print("</head><body>")
+    if args.msg is not None:
+        print("<p>", args.msg, "</p>")
+    print("<table>")
+    print(
+        "<tr><th>File</th>",
+        "<th align='right'>Compile time (ms)</th>",
+        "<th align='right'>Size (bytes)</th><tr>",
+        sep="",
+    )
+    for key in sl:
+        result = entries[key]
+        print(
+            "<tr><td>",
+            key,
+            "</td><td align='right'>",
+            result[0],
+            "</td><td align='right'>",
+            result[1],
+            "</td></tr>",
+            sep="",
+        )
+    print("</table></body></html>")
+
+else:
+    # output results in CSV format
+    print("time,size,file")
+    for key in sl:
+        result = entries[key]
+        print(result[0], result[1], key, sep=",")