From c1b4366834f4c46ddc0f44c2d0bf4e8e70252fe9 Mon Sep 17 00:00:00 2001 From: huangyuxin Date: Sat, 9 Oct 2021 02:52:41 +0000 Subject: [PATCH] update the installing script for openfst --- benchmark/README.md | 14 ---- benchmark/run_benchmark.sh | 65 ------------------- deepspeech/decoders/swig/setup.sh | 2 +- tests/benchmark/.gitignore | 2 - tests/benchmark/README.md | 21 ++++-- {benchmark => tests/benchmark}/analysis.py | 0 {benchmark => tests/benchmark}/prepare.sh | 4 +- {benchmark => tests/benchmark}/run.sh | 0 tests/benchmark/run_all.sh | 49 -------------- .../benchmark}/run_analysis_mp.sh | 0 .../benchmark}/run_analysis_sp.sh | 0 tests/benchmark/run_benchmark.sh | 46 +++++++------ 12 files changed, 46 insertions(+), 157 deletions(-) delete mode 100644 benchmark/README.md delete mode 100644 benchmark/run_benchmark.sh delete mode 100644 tests/benchmark/.gitignore rename {benchmark => tests/benchmark}/analysis.py (100%) rename {benchmark => tests/benchmark}/prepare.sh (71%) rename {benchmark => tests/benchmark}/run.sh (100%) delete mode 100755 tests/benchmark/run_all.sh rename {benchmark => tests/benchmark}/run_analysis_mp.sh (100%) rename {benchmark => tests/benchmark}/run_analysis_sp.sh (100%) mode change 100755 => 100644 tests/benchmark/run_benchmark.sh diff --git a/benchmark/README.md b/benchmark/README.md deleted file mode 100644 index 2d1760a34..000000000 --- a/benchmark/README.md +++ /dev/null @@ -1,14 +0,0 @@ -### Prepare the environment -Please follow the instructions shown in [here](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/docs/src/install.md) to install the Deepspeech first. - -### Prepare the benchmark environment -bash prepare.sh - -### Start benchmarking -bash run.sh - -### Analyse the sp -bash run_analysis_sp.sh - -### Analyse the mp -bash run_analysis_mp.sh diff --git a/benchmark/run_benchmark.sh b/benchmark/run_benchmark.sh deleted file mode 100644 index c745aaaf3..000000000 --- a/benchmark/run_benchmark.sh +++ /dev/null @@ -1,65 +0,0 @@ -#!/usr/bin/env bash -set -xe -# 运行示例：CUDA_VISIBLE_DEVICES=0 bash run_benchmark.sh ${run_mode} ${bs_item} ${fp_item} 500 ${model_mode} -# 参数说明 -function _set_params(){ - - run_mode=${1:-"sp"} # 单卡sp|多卡mp - config_path=${2:-"conf/conformer.yaml"} - output=${3:-"exp/conformer"} - seed=${4:-"0"} - ngpu=${5:-"1"} - profiler_options=${6:-"None"} - batch_size=${7:-"32"} - fp_item=${8:-"fp32"} - TRAIN_LOG_DIR=${9:-$(pwd)} - - benchmark_max_step=0 - - run_log_path=${TRAIN_LOG_DIR:-$(pwd)} # TRAIN_LOG_DIR 后续QA设置该参数 - -# 以下不用修改 - device=${CUDA_VISIBLE_DEVICES//,/ } - arr=(${device}) - num_gpu_devices=${#arr[*]} - log_file=${run_log_path}/recoder_${run_mode}_bs${batch_size}_${fp_item}_ngpu${ngpu}.txt -} - -function _train(){ - echo "Train on ${num_gpu_devices} GPUs" - echo "current CUDA_VISIBLE_DEVICES=$CUDA_VISIBLE_DEVICES, gpus=$num_gpu_devices, batch_size=$batch_size" - train_cmd="--config=${config_path} - --output=${output} - --seed=${seed} - --nproc=${ngpu} - --profiler-options "${profiler_options}" - --benchmark-batch-size ${batch_size} - --benchmark-max-step ${benchmark_max_step} " - - echo "run_mode "${run_mode} - - case ${run_mode} in - sp) train_cmd="python3 -u ${BIN_DIR}/train.py "${train_cmd} ;; - mp) train_cmd="python3 -u ${BIN_DIR}/train.py "${train_cmd} ;; - *) echo "choose run_mode(sp or mp)"; exit 1; - esac - echo ${train_cmd} -# 以下不用修改 - timeout 15m ${train_cmd} > ${log_file} 2>&1 - if [ $? -ne 0 ];then - echo -e "${model_name}, FAIL" - export job_fail_flag=1 - else - echo -e "${model_name}, SUCCESS" - export job_fail_flag=0 - fi - kill -9 `ps -ef|grep 'python'|awk '{print $2}'` - - if [ $run_mode = "mp" -a -d mylog ]; then - rm ${log_file} - cp mylog/workerlog.0 ${log_file} - fi -} - -_set_params $@ -_train diff --git a/deepspeech/decoders/swig/setup.sh b/deepspeech/decoders/swig/setup.sh index 73fa7aea7..302c55502 100755 --- a/deepspeech/decoders/swig/setup.sh +++ b/deepspeech/decoders/swig/setup.sh @@ -10,7 +10,7 @@ fi if [ ! -d openfst-1.6.3 ]; then echo "Download and extract openfst ..." - wget http://www.openfst.org/twiki/pub/FST/FstDownload/openfst-1.6.3.tar.gz + wget http://www.openfst.org/twiki/pub/FST/FstDownload/openfst-1.6.3.tar.gz --no-check-certificate tar -xzvf openfst-1.6.3.tar.gz echo -e "\n" fi diff --git a/tests/benchmark/.gitignore b/tests/benchmark/.gitignore deleted file mode 100644 index 7d166b066..000000000 --- a/tests/benchmark/.gitignore +++ /dev/null @@ -1,2 +0,0 @@ -old-pd_env.txt -pd_env.txt diff --git a/tests/benchmark/README.md b/tests/benchmark/README.md index d21999ab3..fdfc6328f 100644 --- a/tests/benchmark/README.md +++ b/tests/benchmark/README.md @@ -1,11 +1,22 @@ -# Benchmark Test +### Prepare the environment +Please follow the instructions shown in [here](https://github.com/PaddlePaddle/DeepSpeech/blob/develop/docs/src/install.md) to install the Deepspeech first. -## Data +### Prepare the benchmark environment +``` +bash prepare.sh +``` -* Aishell +### Start benchmarking +``` +bash run.sh +``` -## Docker +### Analyse the sp +``` +bash run_analysis_sp.sh +``` +### Analyse the mp ``` -registry.baidubce.com/paddlepaddle/paddle 2.1.1-gpu-cuda10.2-cudnn7 59d5ec1de486 +bash run_analysis_mp.sh ``` diff --git a/benchmark/analysis.py b/tests/benchmark/analysis.py similarity index 100% rename from benchmark/analysis.py rename to tests/benchmark/analysis.py diff --git a/benchmark/prepare.sh b/tests/benchmark/prepare.sh similarity index 71% rename from benchmark/prepare.sh rename to tests/benchmark/prepare.sh index 483caaab7..c934f487b 100644 --- a/benchmark/prepare.sh +++ b/tests/benchmark/prepare.sh @@ -1,7 +1,7 @@ source ../tools/venv/bin/activate -#ִĿ¼ +#Enter the example dir pushd ../examples/aishell/s1 -#׼ +#Prepare the data bash run.sh --stage 0 --stop_stage 0 diff --git a/benchmark/run.sh b/tests/benchmark/run.sh similarity index 100% rename from benchmark/run.sh rename to tests/benchmark/run.sh diff --git a/tests/benchmark/run_all.sh b/tests/benchmark/run_all.sh deleted file mode 100755 index 6f707cdcb..000000000 --- a/tests/benchmark/run_all.sh +++ /dev/null @@ -1,49 +0,0 @@ -#!/bin/bash - -CUR_DIR=${PWD} -ROOT_DIR=../../ - -# 提供可稳定复现性能的脚本，默认在标准docker环境内py37执行： -# collect env info -bash ${ROOT_DIR}/utils/pd_env_collect.sh -#cat pd_env.txt - - -# 1 安装该模型需要的依赖 (如需开启优化策略请注明) -#pushd ${ROOT_DIR}/tools; make; popd -#source ${ROOT_DIR}/tools/venv/bin/activate -#pushd ${ROOT_DIR}; bash setup.sh; popd - - -# 2 拷贝该模型需要数据、预训练模型 - -# 执行目录：需说明 -#pushd ${ROOT_DIR}/examples/aishell/s1 -pushd ${ROOT_DIR}/examples/tiny/s1 - -mkdir -p exp/log -. path.sh -#bash local/data.sh &> exp/log/data.log - -# 3 批量运行（如不方便批量，1，2需放到单个模型中） - -model_mode_list=(conformer transformer) -fp_item_list=(fp32) -bs_item_list=(32 64 96) -for model_mode in ${model_mode_list[@]}; do - for fp_item in ${fp_item_list[@]}; do - for bs_item in ${bs_item_list[@]} - do - echo "index is speed, 1gpus, begin, ${model_name}" - run_mode=sp - CUDA_VISIBLE_DEVICES=0 bash ${CUR_DIR}/run_benchmark.sh ${run_mode} ${bs_item} ${fp_item} 500 ${model_mode} # (5min) - sleep 60 - echo "index is speed, 8gpus, run_mode is multi_process, begin, ${model_name}" - run_mode=mp - CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 bash ${CUR_DIR}/run_benchmark.sh ${run_mode} ${bs_item} ${fp_item} 500 ${model_mode} - sleep 60 - done - done -done - -popd # aishell/s1 diff --git a/benchmark/run_analysis_mp.sh b/tests/benchmark/run_analysis_mp.sh similarity index 100% rename from benchmark/run_analysis_mp.sh rename to tests/benchmark/run_analysis_mp.sh diff --git a/benchmark/run_analysis_sp.sh b/tests/benchmark/run_analysis_sp.sh similarity index 100% rename from benchmark/run_analysis_sp.sh rename to tests/benchmark/run_analysis_sp.sh diff --git a/tests/benchmark/run_benchmark.sh b/tests/benchmark/run_benchmark.sh old mode 100755 new mode 100644 index bd4655d19..c745aaaf3 --- a/tests/benchmark/run_benchmark.sh +++ b/tests/benchmark/run_benchmark.sh @@ -1,40 +1,50 @@ -#!/bin/bash - +#!/usr/bin/env bash set -xe - # 运行示例：CUDA_VISIBLE_DEVICES=0 bash run_benchmark.sh ${run_mode} ${bs_item} ${fp_item} 500 ${model_mode} # 参数说明 function _set_params(){ + run_mode=${1:-"sp"} # 单卡sp|多卡mp - batch_size=${2:-"64"} - fp_item=${3:-"fp32"} # fp32|fp16 - max_iter=${4:-"500"} # 可选，如果需要修改代码提前中断 - model_name=${5:-"model_name"} + config_path=${2:-"conf/conformer.yaml"} + output=${3:-"exp/conformer"} + seed=${4:-"0"} + ngpu=${5:-"1"} + profiler_options=${6:-"None"} + batch_size=${7:-"32"} + fp_item=${8:-"fp32"} + TRAIN_LOG_DIR=${9:-$(pwd)} + + benchmark_max_step=0 + run_log_path=${TRAIN_LOG_DIR:-$(pwd)} # TRAIN_LOG_DIR 后续QA设置该参数 # 以下不用修改 device=${CUDA_VISIBLE_DEVICES//,/ } arr=(${device}) num_gpu_devices=${#arr[*]} - log_file=${run_log_path}/${model_name}_${run_mode}_bs${batch_size}_${fp_item}_${num_gpu_devices} + log_file=${run_log_path}/recoder_${run_mode}_bs${batch_size}_${fp_item}_ngpu${ngpu}.txt } function _train(){ echo "Train on ${num_gpu_devices} GPUs" echo "current CUDA_VISIBLE_DEVICES=$CUDA_VISIBLE_DEVICES, gpus=$num_gpu_devices, batch_size=$batch_size" + train_cmd="--config=${config_path} + --output=${output} + --seed=${seed} + --nproc=${ngpu} + --profiler-options "${profiler_options}" + --benchmark-batch-size ${batch_size} + --benchmark-max-step ${benchmark_max_step} " - train_cmd="--benchmark-batch-size ${batch_size} - --benchmark-max-step ${max_iter} - conf/${model_name}.yaml ${model_name}" + echo "run_mode "${run_mode} case ${run_mode} in - sp) train_cmd="bash local/train.sh "${train_cmd}"" ;; - mp) - train_cmd="bash local/train.sh "${train_cmd}"" ;; + sp) train_cmd="python3 -u ${BIN_DIR}/train.py "${train_cmd} ;; + mp) train_cmd="python3 -u ${BIN_DIR}/train.py "${train_cmd} ;; *) echo "choose run_mode(sp or mp)"; exit 1; esac - - # 以下不用修改 + echo ${train_cmd} +# 以下不用修改 timeout 15m ${train_cmd} > ${log_file} 2>&1 if [ $? -ne 0 ];then echo -e "${model_name}, FAIL" @@ -43,8 +53,7 @@ function _train(){ echo -e "${model_name}, SUCCESS" export job_fail_flag=0 fi - - trap 'for pid in $(jobs -pr); do kill -KILL $pid; done' INT QUIT TERM + kill -9 `ps -ef|grep 'python'|awk '{print $2}'` if [ $run_mode = "mp" -a -d mylog ]; then rm ${log_file} @@ -54,4 +63,3 @@ function _train(){ _set_params $@ _train -