From f5d6937fe163f3ee774c25e95246371288ea2b1c Mon Sep 17 00:00:00 2001
From: wopeizl <peizhilin@baidu.com>
Date: Wed, 24 Apr 2019 13:08:33 +0800
Subject: [PATCH] =?UTF-8?q?specify=20the=20cuda=20arch=20name=20and=20bin?=
 =?UTF-8?q?=20to=20decrease=20the=20compile=20time=20for=20i=E2=80=A6=20(#?=
 =?UTF-8?q?17020)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

1. specify the cuda arch name and bin to decrease the compile time for inference test=develop
2. simplify the script and add comments
3. remove the fluid process from cicheck
---
 paddle/scripts/paddle_build.sh | 92 ++++++++--------------------------
 1 file changed, 21 insertions(+), 71 deletions(-)

diff --git a/paddle/scripts/paddle_build.sh b/paddle/scripts/paddle_build.sh
index 99320e27135..760a4437072 100755
--- a/paddle/scripts/paddle_build.sh
+++ b/paddle/scripts/paddle_build.sh
@@ -194,6 +194,7 @@ function cmake_gen() {
         -DWITH_AVX=${WITH_AVX:-OFF}
         -DWITH_GOLANG=${WITH_GOLANG:-OFF}
         -DCUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All}
+        -DCUDA_ARCH_BIN=${CUDA_ARCH_BIN} \
         -DWITH_PYTHON=${WITH_PYTHON:-ON}
         -DCUDNN_ROOT=/usr/
         -DWITH_TESTING=${WITH_TESTING:-ON}
@@ -228,6 +229,7 @@ EOF
         -DWITH_AVX=${WITH_AVX:-OFF} \
         -DWITH_GOLANG=${WITH_GOLANG:-OFF} \
         -DCUDA_ARCH_NAME=${CUDA_ARCH_NAME:-All} \
+        -DCUDA_ARCH_BIN=${CUDA_ARCH_BIN} \
         -DWITH_PYTHON=${WITH_PYTHON:-ON} \
         -DCUDNN_ROOT=/usr/ \
         -DWITH_TESTING=${WITH_TESTING:-ON} \
@@ -561,57 +563,6 @@ function bind_test() {
     wait
 }
 
-function parallel_test() {
-    mkdir -p ${PADDLE_ROOT}/build
-    cd ${PADDLE_ROOT}/build
-    if [ ${WITH_TESTING:-ON} == "ON" ] ; then
-    cat <<EOF
-    ========================================
-    Running unit tests ...
-    ========================================
-EOF
-
-        # calculate and set the memory usage for each process
-        # MEM_USAGE=$(printf "%.2f" `echo "scale=5; 1.0 / $NUM_PROC" | bc`)
-        # export FLAGS_fraction_of_gpu_memory_to_use=$MEM_USAGE
-
-        EXIT_CODE=0;
-        pids=()
-
-        # get the CUDA device count
-        CUDA_DEVICE_COUNT=$(nvidia-smi -L | wc -l)
-        # each test case would occupy two graph cards
-        NUM_PROC=$[CUDA_DEVICE_COUNT/2]
-        for (( i = 0; i < $NUM_PROC; i++ )); do
-            # CUDA_VISIBLE_DEVICES http://acceleware.com/blog/cudavisibledevices-masking-gpus
-            # ctest -I https://cmake.org/cmake/help/v3.0/manual/ctest.1.html?highlight=ctest
-            if [ ${TESTING_DEBUG_MODE:-OFF} == "ON" ] ; then
-                env CUDA_VISIBLE_DEVICES=$[i*2],$[i*2+1] ctest -I $i,,$NUM_PROC -V &
-                pids+=($!)
-            else
-                env CUDA_VISIBLE_DEVICES=$[i*2],$[i*2+1] ctest -I $i,,$NUM_PROC --output-on-failure &
-                pids+=($!)
-            fi
-        done
-
-        clen=`expr "${#pids[@]}" - 1` # get length of commands - 1
-        for i in `seq 0 "$clen"`; do
-            wait ${pids[$i]}
-            CODE=$?
-            if [[ "${CODE}" != "0" ]]; then
-                echo "At least one test failed with exit code => ${CODE}" ;
-                EXIT_CODE=1;
-            fi
-        done
-        wait; # wait for all subshells to finish
-
-        echo "EXIT_CODE => $EXIT_CODE"
-        if [[ "${EXIT_CODE}" != "0" ]]; then
-            exit "$EXIT_CODE"
-        fi
-    fi
-}
-
 EXIT_CODE=0;
 function caught_error() {
  for job in `jobs -p`; do
@@ -657,7 +608,6 @@ function card_test() {
                     cuda_list="$cuda_list,$[i*cardnumber+j]"
             fi
         done
-        # echo $cuda_list
         if [ ${TESTING_DEBUG_MODE:-OFF} == "ON" ] ; then
             if [[ $cardnumber == $CUDA_DEVICE_COUNT ]]; then
                 ctest -I $i,,$NUM_PROC -R "($testcases)" -V &
@@ -675,26 +625,27 @@ function card_test() {
     done
 
     wait; # wait for all subshells to finish
+    set +m
 }
 
-function aggresive_test() {
+function parallel_test() {
     mkdir -p ${PADDLE_ROOT}/build
     cd ${PADDLE_ROOT}/build
     if [ ${WITH_TESTING:-ON} == "ON" ] ; then
     cat <<EOF
     ========================================
-    Running unit tests ...
+    Running unit tests in parallel way ...
     ========================================
 EOF
 
 set +x
         EXIT_CODE=0;
-        test_cases=$(ctest -N -V)
-        exclusive_tests=''
-        single_card_tests=''
-        multiple_card_tests=''
-        is_exclusive=''
-        is_multicard=''
+        test_cases=$(ctest -N -V) # get all test cases
+        exclusive_tests=''        # cases list which would be run exclusively
+        single_card_tests=''      # cases list which would take one graph card
+        multiple_card_tests=''    # cases list which would take multiple GPUs, most cases would be two GPUs
+        is_exclusive=''           # indicate whether the case is exclusive type
+        is_multicard=''           # indicate whether the case is multiple GPUs type
         while read -r line; do
             if [[ "$line" == "" ]]; then
                 continue
@@ -703,7 +654,7 @@ set +x
                 if [[ "$matchstr" == "" ]]; then
                     # Any test case with LABELS property would be parse here
                     # RUN_TYPE=EXCLUSIVE mean the case would run exclusively
-                    # RUN_TYPE=DIST mean the case would take two graph cards during runtime
+                    # RUN_TYPE=DIST mean the case would take two graph GPUs during runtime
                     read is_exclusive <<< $(echo "$line"|grep -oEi "RUN_TYPE=EXCLUSIVE")
                     read is_multicard <<< $(echo "$line"|grep -oEi "RUN_TYPE=DIST")
                     continue
@@ -711,7 +662,7 @@ set +x
                 read testcase <<< $(echo "$line"|grep -oEi "\w+$")
 
                 if [[ "$is_multicard" == "" ]]; then
-                  # trick: treat all test case with prefix "test_dist" as dist case, and would run on 2 cards
+                  # trick: treat all test case with prefix "test_dist" as dist case, and would run on 2 GPUs
                   read is_multicard <<< $(echo "$testcase"|grep -oEi "test_dist")
                 fi
 
@@ -740,9 +691,9 @@ set +x
                 testcase=''
         done <<< "$test_cases";
 
-        card_test "$single_card_tests" 1
-        card_test "$multiple_card_tests" 2
-        card_test "$exclusive_tests"
+        card_test "$single_card_tests" 1    # run cases with single GPU
+        card_test "$multiple_card_tests" 2  # run cases with two GPUs
+        card_test "$exclusive_tests"        # run cases exclusively, in this cases would be run with 4/8 GPUs
         if [[ "$EXIT_CODE" != "0" ]]; then
             exit 1;
         fi
@@ -942,7 +893,8 @@ EOF
     if [[ "$1" != "" ]]; then
       parallel_number=$1
     fi
-    cmake .. -DWITH_DISTRIBUTE=OFF -DON_INFER=ON
+    cmake .. -DWITH_DISTRIBUTE=OFF -DON_INFER=ON -DCUDA_ARCH_NAME=${CUDA_ARCH_NAME:-Auto} -DCUDA_ARCH_BIN=${CUDA_ARCH_BIN}
+
     make -j ${parallel_number} fluid_lib_dist
     make -j ${parallel_number} inference_lib_dist
 }
@@ -994,7 +946,7 @@ function main() {
         gen_dockerfile ${PYTHON_ABI:-""}
         ;;
       test)
-        aggresive_test
+        parallel_test
         ;;
       single_test)
         single_test $2
@@ -1024,9 +976,7 @@ function main() {
         cmake_gen ${PYTHON_ABI:-""}
         build ${parallel_number}
         assert_api_not_changed ${PYTHON_ABI:-""}
-        aggresive_test
-        gen_fluid_lib ${parallel_number}
-        test_fluid_lib
+        parallel_test
         assert_api_spec_approvals
         ;;
       cicheck_brpc)
@@ -1057,7 +1007,7 @@ function main() {
       cicheck_py35)
         cmake_gen ${PYTHON_ABI:-""}
         build ${parallel_number}
-        aggresive_test
+        parallel_test
         assert_api_not_changed ${PYTHON_ABI:-""}
         ;;
       cmake_gen)
-- 
GitLab