You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

run_train.sh 1.8 kB

12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364
  1. #!/bin/bash
  2. # Copyright 2020 Huawei Technologies Co., Ltd
  3. #
  4. # Licensed under the Apache License, Version 2.0 (the "License");
  5. # you may not use this file except in compliance with the License.
  6. # You may obtain a copy of the License at
  7. #
  8. # http://www.apache.org/licenses/LICENSE-2.0
  9. #
  10. # Unless required by applicable law or agreed to in writing, software
  11. # distributed under the License is distributed on an "AS IS" BASIS,
  12. # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. # See the License for the specific language governing permissions and
  14. # limitations under the License.
  15. # ============================================================================
  16. if [ $# != 1 ] && [ $# != 2 ]
  17. then
  18. echo "Usage: sh run_train.sh [RANK_TABLE_FILE] [cifar10|imagenet]"
  19. exit 1
  20. fi
  21. if [ ! -f $1 ]
  22. then
  23. echo "error: RANK_TABLE_FILE=$1 is not a file"
  24. exit 1
  25. fi
  26. dataset_type='cifar10'
  27. if [ $# == 2 ]
  28. then
  29. if [ $2 != "cifar10" ] && [ $2 != "imagenet" ]
  30. then
  31. echo "error: the selected dataset is neither cifar10 nor imagenet"
  32. exit 1
  33. fi
  34. dataset_type=$2
  35. fi
  36. ulimit -u unlimited
  37. export DEVICE_NUM=8
  38. export RANK_SIZE=8
  39. RANK_TABLE_FILE=$(realpath $1)
  40. export RANK_TABLE_FILE
  41. echo "RANK_TABLE_FILE=${RANK_TABLE_FILE}"
  42. export SERVER_ID=0
  43. rank_start=$((DEVICE_NUM * SERVER_ID))
  44. for((i=0; i<${DEVICE_NUM}; i++))
  45. do
  46. export DEVICE_ID=$i
  47. export RANK_ID=$((rank_start + i))
  48. rm -rf ./train_parallel$i
  49. mkdir ./train_parallel$i
  50. cp -r ./src ./train_parallel$i
  51. cp ./train.py ./train_parallel$i
  52. echo "start training for rank $RANK_ID, device $DEVICE_ID, $dataset_type"
  53. cd ./train_parallel$i ||exit
  54. env > env.log
  55. python train.py --device_id=$i --dataset_name=$dataset_type> log 2>&1 &
  56. cd ..
  57. done