Skip to content

Latest commit

 

History

History
71 lines (49 loc) · 1.11 KB

README.SETUP.md

File metadata and controls

71 lines (49 loc) · 1.11 KB

Setup Instructions

Repo Setup

Clone repo:

Via SSH:

git clone \
    --branch lookup-profiling \
    --recursive [email protected]:UBC-ECE-Sasha/PIM-Embedding-Lookup.git

or https:

git clone \
    --branch lookup-profiling \
    --recursive https://github.com/UBC-ECE-Sasha/PIM-Embedding-Lookup.git

Python Setup

Enter PIM-Embedding-Lookup

cd PIM-Embedding-Lookup

Install minicona:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
chmod +x Miniconda3-latest-Linux-x86_64.sh
./Miniconda3-latest-Linux-x86_64.sh

Setup env:

conda env create -f environment.yml
conda activate pytorch

If graphviz fails to install change python-graphviz to graphviz in environment.yml, delete old env and try again.

Compiling PIM-Pytorch

Set required variables (subject to change, check upmem/run.sh)

export NR_TABLES=10
export NR_COLS=32
export MAX_NR_BATCHES=64
export NR_TASKLETS=14

Enter upmem, build and run 'random' workload.

./run.sh -br random

To only run, use -r

./run.sh -r random