You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
PaddleSpeech/tests/benchmark/conformer/README.md

1.9 KiB

Prepare the environment

Please follow the instructions shown in here to install the Deepspeech first.

File list

└── benchmark # 模型名
├── README.md # 运行文档
├── analysis.py # log解析脚本,每个框架尽量统一,可参考paddle的analysis.py
├── recoder_mp_bs16_fp32_ngpu1.txt # 单卡数据 ├── recoder_mp_bs16_fp32_ngpu8.txt # 8卡数据
├── prepare.sh # 竞品PyTorch运行环境搭建
├── run_benchmark.sh # 运行脚本(包含性能、收敛性)
├── run_analysis_mp.sh # 分析8卡的脚本
├── run_analysis_sp.sh # 分析单卡的脚本
├── log │ ├── log_sp.out # 单卡的结果 │ └── log_mp.out # 8卡的结果 └── run.sh # 全量运行脚本

The physical environment

  • 单机单卡、8卡
    • 系统Ubuntu 16.04.6 LTS
    • GPUTesla V100-SXM2-16GB * 8
    • CPUIntel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz * 96
    • Driver Version: 440.64.00
    • 内存440 GB
    • CUDA、cudnn Version: cuda10.2-cudnn7
  • 多机32卡 TODO

Docker 镜像,如:

  • 镜像版本: registry.baidubce.com/paddlepaddle/paddle:2.1.0-gpu-cuda10.2-cudnn7
  • CUDA 版本: 10.2
  • cuDnn 版本: 7

Prepare the benchmark environment

bash prepare.sh

Start benchmarking

bash run.sh

The log

{"log_file": "recoder_sp_bs16_fp32_ngpu1.txt",
 "model_name": "Conformer",
 "mission_name": "one gpu",
 "direction_id": 1,
 "run_mode": "sp",
 "index": 1,
 "gpu_num": 1,
 "FINAL_RESULT": 23.228,
 "JOB_FAIL_FLAG": 0,
 "log_with_profiler": null,
 "profiler_path": null,
 "UNIT": "sent./sec"
}