|
| 1 | +# Slurm job template to probe EC2 informations |
| 2 | + |
| 3 | +Usage: review and customize [job-template.sbatch](job-template.sbatch) to your need. |
| 4 | + |
| 5 | +Below shows the sample output (trimmed) of customizing the template to run Megatron-LM, i.e., |
| 6 | +replace the `srun -l /usr/bin/hostname` with the relevant parts from the [Megatron-LM |
| 7 | +example](../../3.test_cases/1.megatron-lm/)). |
| 8 | + |
| 9 | +```text |
| 10 | +... |
| 11 | ++ validate_ec2_same_spine |
| 12 | +++ lstopo_ec2 |
| 13 | +++ INSTANCE_IDS=($(srun cat /sys/devices/virtual/dmi/id/board_asset_tag)) |
| 14 | ++++ srun cat /sys/devices/virtual/dmi/id/board_asset_tag |
| 15 | +++ local INSTANCE_IDS |
| 16 | +++ aws ec2 describe-instance-topology --instance-ids i-1111111111example i-0000000000example |
| 17 | ++ local 'TOPO_JSON={ |
| 18 | + "Instances": [ |
| 19 | + { |
| 20 | + "InstanceId": "i-0000000000example", |
| 21 | + "InstanceType": "p4de.24xlarge", |
| 22 | + "NetworkNodes": [ |
| 23 | + "nn-1111111111example", |
| 24 | + "nn-2222222222example", |
| 25 | + "nn-3333333333example" |
| 26 | + ], |
| 27 | + "AvailabilityZone": "us-west-2b", |
| 28 | + "ZoneId": "usw2-az2" |
| 29 | + }, |
| 30 | + { |
| 31 | + "InstanceId": "i-1111111111example", |
| 32 | + "InstanceType": "p4de.24xlarge", |
| 33 | + "NetworkNodes": [ |
| 34 | + "nn-1111111111example", |
| 35 | + "nn-2222222222example", |
| 36 | + "nn-3333333333example" |
| 37 | + ], |
| 38 | + "AvailabilityZone": "us-west-2b", |
| 39 | + "ZoneId": "usw2-az2" |
| 40 | + } |
| 41 | + ] |
| 42 | +}' |
| 43 | ++ echo '{ |
| 44 | + "Instances": [ |
| 45 | + { |
| 46 | + "InstanceId": "i-0000000000example", |
| 47 | + "InstanceType": "p4de.24xlarge", |
| 48 | + "NetworkNodes": [ |
| 49 | + "nn-1111111111example", |
| 50 | + "nn-2222222222example", |
| 51 | + "nn-3333333333example" |
| 52 | + ], |
| 53 | + "AvailabilityZone": "us-west-2b", |
| 54 | + "ZoneId": "usw2-az2" |
| 55 | + }, |
| 56 | + { |
| 57 | + "InstanceId": "i-1111111111example", |
| 58 | + "InstanceType": "p4de.24xlarge", |
| 59 | + "NetworkNodes": [ |
| 60 | + "nn-1111111111example", |
| 61 | + "nn-2222222222example", |
| 62 | + "nn-3333333333example" |
| 63 | + ], |
| 64 | + "AvailabilityZone": "us-west-2b", |
| 65 | + "ZoneId": "usw2-az2" |
| 66 | + } |
| 67 | + ] |
| 68 | +}' |
| 69 | +{ |
| 70 | + "Instances": [ |
| 71 | + { |
| 72 | + "InstanceId": "i-0000000000example", |
| 73 | + "InstanceType": "p4de.24xlarge", |
| 74 | + "NetworkNodes": [ |
| 75 | + "nn-1111111111example", |
| 76 | + "nn-2222222222example", |
| 77 | + "nn-3333333333example" |
| 78 | + ], |
| 79 | + "AvailabilityZone": "us-west-2b", |
| 80 | + "ZoneId": "usw2-az2" |
| 81 | + }, |
| 82 | + { |
| 83 | + "InstanceId": "i-1111111111example", |
| 84 | + "InstanceType": "p4de.24xlarge", |
| 85 | + "NetworkNodes": [ |
| 86 | + "nn-1111111111example", |
| 87 | + "nn-2222222222example", |
| 88 | + "nn-3333333333example" |
| 89 | + ], |
| 90 | + "AvailabilityZone": "us-west-2b", |
| 91 | + "ZoneId": "usw2-az2" |
| 92 | + } |
| 93 | + ] |
| 94 | +} |
| 95 | +++ echo '{ |
| 96 | + "Instances": [ |
| 97 | + { |
| 98 | + "InstanceId": "i-0000000000example", |
| 99 | + "InstanceType": "p4de.24xlarge", |
| 100 | + "NetworkNodes": [ |
| 101 | + "nn-1111111111example", |
| 102 | + "nn-2222222222example", |
| 103 | + "nn-3333333333example" |
| 104 | + ], |
| 105 | + "AvailabilityZone": "us-west-2b", |
| 106 | + "ZoneId": "usw2-az2" |
| 107 | + }, |
| 108 | + { |
| 109 | + "InstanceId": "i-1111111111example", |
| 110 | + "InstanceType": "p4de.24xlarge", |
| 111 | + "NetworkNodes": [ |
| 112 | + "nn-1111111111example", |
| 113 | + "nn-2222222222example", |
| 114 | + "nn-3333333333example" |
| 115 | + ], |
| 116 | + "AvailabilityZone": "us-west-2b", |
| 117 | + "ZoneId": "usw2-az2" |
| 118 | + } |
| 119 | + ] |
| 120 | +}' |
| 121 | +++ grep '^ *"nn\-.................\"' |
| 122 | +++ sort -n |
| 123 | +++ uniq -c |
| 124 | +++ wc -l |
| 125 | ++ local UNIQ_NN=3 |
| 126 | ++ echo Expected 3 nn ids, got 3 nn ids |
| 127 | +Expected 3 nn ids, got 3 nn ids |
| 128 | ++ [[ 3 -eq 3 ]] |
| 129 | +... |
| 130 | ++ srun -l bash -c 'echo "hostname <=> instance_id mapping: $(hostname) <=> $(cat /sys/devices/virtual/dmi/id/board_asset_tag)"' |
| 131 | +1: hostname <=> instance_id mapping: p4de-st-p4de-2 <=> i-0000000000example |
| 132 | +0: hostname <=> instance_id mapping: p4de-st-p4de-1 <=> i-1111111111example |
| 133 | ++ srun -l bash -c 'echo BEFORE: $(hostname) $(sudo lctl get_param llite.*.stats | grep write_bytes)' |
| 134 | +0: BEFORE: p4de-st-p4de-1 write_bytes 10361 samples [bytes] 1 2147479552 401369912220 |
| 135 | +1: BEFORE: p4de-st-p4de-2 |
| 136 | +... |
| 137 | +++ date |
| 138 | ++ BEGIN_TRAINING='Mon Apr 8 08:33:01 UTC 2024' |
| 139 | ++ SECONDS=0 |
| 140 | ++ srun -l --container-image ... --container-mounts ... python -m torch.distributed.run ... /workspace/Megatron-LM/pretrain_gpt.py ... |
| 141 | +... |
| 142 | +++ date |
| 143 | ++ END_TRAINING='Mon Apr 8 09:02:25 UTC 2024' |
| 144 | ++ echo 'BEGIN_TRAINING: Mon Apr 8 08:33:01 UTC 2024' |
| 145 | +BEGIN_TRAINING: Mon Apr 8 08:33:01 UTC 2024 |
| 146 | ++ echo 'END_TRAINING : Mon Apr 8 09:02:25 UTC 2024' |
| 147 | +END_TRAINING : Mon Apr 8 09:02:25 UTC 2024 |
| 148 | ++ echo 'Elapsed: 29min 24sec' |
| 149 | +Elapsed: 29min 24sec |
| 150 | ++ srun -l bash -c 'echo AFTER: $(hostname) $(sudo lctl get_param llite.*.stats | grep write_bytes)' |
| 151 | +srun: Step created for StepId=191.4 |
| 152 | +1: AFTER: p4de-st-p4de-2 |
| 153 | +0: AFTER: p4de-st-p4de-1 write_bytes 11553 samples [bytes] 1 2147479552 775980517197 |
| 154 | +``` |
0 commit comments