Skip to content

Latest commit

 

History

History
106 lines (82 loc) · 3.66 KB

README.md

File metadata and controls

106 lines (82 loc) · 3.66 KB
Raw

PyTorch DLRM inference

Description

This document has instructions for running DLRM inference using Intel-optimized PyTorch for bare metal.

Bare Metal

General Setup

Follow link to install Conda and build Pytorch, IPEX, and Jemalloc.

Model Specific Setup

  • Install dependencies

    cd <clone of the model zoo>/quickstart/recommendation/pytorch/dlrm
pip install requirements.txt

  • Set Jemalloc Preload for better performance

    The jemalloc should be built from the General setup section.

    export LD_PRELOAD="path/lib/libjemalloc.so":$LD_PRELOAD
export MALLOC_CONF="oversize_threshold:1,background_thread:true,metadata_thp:auto,dirty_decay_ms:9000000000,muzzy_decay_ms:9000000000"

  • Set IOMP preload for better performance

    IOMP should be installed in your conda env from the General setup section.

    export LD_PRELOAD=path/lib/libiomp5.so:$LD_PRELOAD

  • Set ENV to use AMX if you are using SPR

    export DNNL_MAX_CPU_ISA=AVX512_CORE_AMX

Datasets

Criteo Terabyte Dataset

The Criteo Terabyte Dataset is used to run DLRM. To download the dataset, you will need to visit the Criteo website and accept their terms of use: https://labs.criteo.com/2013/12/download-terabyte-click-logs/. Copy the download URL into the command below as the <download url> and replace the <dir/to/save/dlrm_data> to any path where you want to download and save the dataset.

export DATASET_DIR=<dir/to/save/dlrm_data>

mkdir ${DATASET_DIR} && cd ${DATASET_DIR}
curl -O <download url>/day_{$(seq -s , 0 23)}.gz
gunzip day_*.gz

The raw data will be automatically preprocessed and saved as day_*.npz to the DATASET_DIR when DLRM is run for the first time. On subsequent runs, the scripts will automatically use the preprocessed data.

Quick Start Scripts

Script name Description
inference_performance.sh Run inference to verify performance for the specified precision (fp32, avx-fp32, int8, avx-int8, or bf16).
accuracy.sh Measures the inference accuracy for the specified precision (fp32, avx-fp32, int8, avx-int8, or bf16).

Run the model

Download the DLRM PyTorch weights (tb00_40M.pt, 90GB) from the MLPerf repo and set the WEIGHT_PATH to point to the weights file.

export WEIGHT_PATH=<path to the tb00_40M.pt file>

Follow the instructions above to setup your bare metal environment, do the model-specific setup and download and prepropcess the datsaet. Once all the setup is done, the Model Zoo can be used to run a quickstart script. Ensure that you have enviornment variables set to point to the dataset directory, precision, weights file, and an output directory.

# Clone the model zoo repo and set the MODEL_DIR
git clone https://github.com/IntelAI/models.git
cd models
export MODEL_DIR=$(pwd)

# Env vars
export PRECISION=<specify the precision to run>
export WEIGHT_PATH=<path to the tb00_40M.pt file> # only needed for testing accuracy
export DATASET_DIR=<path to the dataset>
export OUTPUT_DIR=<directory where log files will be written>

# Run a quickstart script (for example, bare metal performance)
cd ${MODEL_DIR}/quickstart/recommendation/pytorch/dlrm/inference/cpu
bash inference_performance.sh

License

LICENSE