#!/bin/bash


#集合通信参数,不需要修改
export RANK_SIZE=1

# 数据集路径,保持为空,不需要修改
data_path=""

#网络名称,同目录名称,需要模型审视修改
Network="CRNN_ID0103_for_PyTorch"

#训练batch_size,,需要模型审视修改
batch_size=2560

#训练epoch，不需要修改
epochs=2
# 指定训练所使用的npu device卡id
device_id=0
# 参数校验，data_path为必传参数，其他参数的增删由模型自身决定；此处新增参数需在上面有定义并赋值
for para in $*
do
    if [[ $para == --data_path* ]];then
        data_path=`echo ${para#*=}`
    elif [[ $para == --epochs* ]];then
        epochs=`echo ${para#*=}`
    fi
done

#校验是否传入data_path,不需要修改
if [[ $data_path == "" ]];then
    echo "[Error] para \"data_path\" must be confing"
    exit 1
fi
# 校验是否指定了device_id,分动态分配device_id与手动指定device_id,此处不需要修改
if [ $ASCEND_DEVICE_ID ];then
    echo "device id is ${ASCEND_DEVICE_ID}"
elif [ ${device_id} ];then
    export ASCEND_DEVICE_ID=${device_id}
    echo "device id is ${ASCEND_DEVICE_ID}"
else
    "[Error] device id must be config"
    exit 1
fi
###############指定训练脚本执行路径###############
# cd到与test文件夹同层级目录下执行脚本，提高兼容性；test_path_dir为包含test文件夹的路径
cur_path=`pwd`
cur_path_last_dirname=${cur_path##*/}
if [ x"${cur_path_last_dirname}" == x"test" ];then
    test_path_dir=${cur_path}
    cd ..
    cur_path=`pwd`
else
    test_path_dir=${cur_path}/test
fi

#创建DeviceID输出目录，不需要修改
if [ -d ${test_path_dir}/output/${ASCEND_DEVICE_ID} ];then
    rm -rf ${test_path_dir}/output/$ASCEND_DEVICE_ID
    mkdir -p ${test_path_dir}/output/$ASCEND_DEVICE_ID
else
    mkdir -p ${test_path_dir}/output/$ASCEND_DEVICE_ID
fi

#################启动训练脚本#################
#训练开始时间，不需要修改
start_time=$(date +%s)
# 非平台场景时source 环境变量
check_etp_flag=`env | grep etp_running_flag`
etp_flag=`echo ${check_etp_flag#*=}`
if [ x"${etp_flag}" != x"true" ];then
    source ${test_path_dir}/env_npu.sh
fi

# 必要参数替换配置文件
cur_path=`pwd`
sed -i "0,/BATCH_SIZE_PER_GPU.*$/s//BATCH_SIZE_PER_GPU\: ${batch_size}/g" ${cur_path}/LMDB_config.yaml
sed -i "s/END_EPOCH.*$/END_EPOCH\: ${epochs}/g" ${cur_path}/LMDB_config.yaml
sed -i "s|TRAIN_ROOT.*$|TRAIN_ROOT\: ${data_path}/MJ_LMDB|g" ${cur_path}/LMDB_config.yaml
sed -i "s|TEST_ROOT.*$|TEST_ROOT\: ${data_path}/IIIT5K_lmdb|g" ${cur_path}/LMDB_config.yaml
#sed -i "s/DEVICE_ID.*$/DEVICE_ID\: ${device_id}/g" ${cur_path}/LMDB_config.yaml
#执行训练脚本，以下传参不需要修改，其他需要模型审视修改
export ASCEND_SLOG_PRINT_TO_STDOUT=0
export ASCEND_GLOBAL_LOG_LEVEL=3
export TASK_QUEUE_ENABLE=1
export PTCOPY_ENABLE=1
export COMBINED_ENABLE=1
export SWITCH_MM_OUTPUT_ENABLE=1

/usr/local/Ascend/driver/tools/msnpureport -g error -d 0
/usr/local/Ascend/driver/tools/msnpureport -g error -d 4

python3.7 main.py \
    --cfg LMDB_config.yaml > ${test_path_dir}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log 2>&1 &
    
wait

##################获取训练数据################
#训练结束时间，不需要修改
end_time=$(date +%s)
e2e_time=$(( $end_time - $start_time ))

#结果打印，不需要修改
echo "------------------ Final result ------------------"
#输出性能FPS，需要模型审视修改
FPS=`grep -a 'FPS@all'  ${test_path_dir}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log|awk -F " " '{print $NF}'|awk 'NR==1{max=$1;next}{max=max>$1?max:$1}END{print max}'`
#打印，不需要修改
echo "Final Performance images/sec : $FPS"

#打印，不需要修改
echo "E2E Training Duration sec : $e2e_time"
#输出编译时间
CompileTime=`grep "Epoch:"  ${test_path_dir}/output/${ASCEND_DEVICE_ID}/train_${ASCEND_DEVICE_ID}.log| head -2 |awk -F "Time " '{print $2}' | awk '{sum+=$1} END {print"",sum/1000}' |sed s/[[:space:]]//g`

#性能看护结果汇总
#训练用例信息，不需要修改
BatchSize=${batch_size}
DeviceType=`uname -m`
CaseName=${Network}_bs${BatchSize}_${RANK_SIZE}'p'_'perf'

#获取性能数据，不需要修改
#吞吐量
ActualFPS=${FPS}
#单迭代训练时长
TrainingTime=`awk 'BEGIN{printf "%.2f\n", '${batch_size}'*1000/'${FPS}'}'`

#从train_$ASCEND_DEVICE_ID.log提取Loss到train_${CaseName}_loss.txt中，需要模型审视修改
grep -a 'Loss' ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_$ASCEND_DEVICE_ID.log|awk -F "Loss " '{print $NF}' | awk -F " " '{print $1}' >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt
#最后一个迭代loss值，不需要修改
ActualLoss=`awk 'END {print}' ${test_path_dir}/output/$ASCEND_DEVICE_ID/train_${CaseName}_loss.txt`

#关键信息打印到${CaseName}.log中，不需要修改
echo "Network = ${Network}" > ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "RankSize = ${RANK_SIZE}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "BatchSize = ${BatchSize}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "DeviceType = ${DeviceType}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "CaseName = ${CaseName}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "ActualFPS = ${ActualFPS}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "TrainingTime = ${TrainingTime}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "ActualLoss = ${ActualLoss}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "E2ETrainingTime = ${e2e_time}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log
echo "CompileTime = ${CompileTime}" >> ${test_path_dir}/output/$ASCEND_DEVICE_ID/${CaseName}.log