Code Monkey home page Code Monkey logo

leaftl's Introduction

LeaFTL

LeaFTL is a learning-based flash translation layer (FTL), which learns the address mapping and tolerates dynamic data access patterns via linear regression at runtime to reduce the memory footprint of the address mapping.

1. System Setup

The following packages and benchmarks are necessary to install before running the experiments. It is recommended to allocate a dedicated server for these experiments as they are both cpu and memory intensive.

git clone https://github.com/my-HenryS/LeaFTL.git
cd LeaFTL
pushd .
# Download traces
cd wiscsee/leaftl_scripts
pip3 install gdown
gdown 13gS89MNWESywUA3ZGycSwYTFjZ29qdGu
unzip leaftl_traces.zip
popd

# Set Env Variables
export PYTHONPATH=$PYTHONPATH:$(pwd)/wiscsee

# Install Pypy to speedup the experiment running
# You can run WiscSim 2-3× faster with Pypy. You can install Pypy2 with this guide: https://doc.pypy.org/en/latest/install.html.
wget https://downloads.python.org/pypy/pypy2.7-v7.3.9-linux64.tar.bz2
tar xf pypy2.7-v7.3.9-linux64.tar.bz2
export PATH=$PATH:$(pwd)/pypy2.7-v7.3.9-linux64/bin
pypy -m ensurepip

# Install dependencies
sudo apt-get install libblas-dev liblapack-dev libatlas-base-dev gfortran

# Install Python packages
cd wiscsee/leaftl_scripts
./setup.sh

2. Simulation-based Experiments

2.0 Directory Structures

leaftl_scripts/
├── batch  # script to run a batch of experiments
├── leaftl_traces  # leaftl traces
├── plot_all   # script to plot figures
├── plots     # directory for figures
├── raw_results     # raw experiment output
├── reference_plots   # reference figures
├── run_ftl      # simulator entry function
├── setup.sh    # env setup script
└── warmup    # warmup workloads

2.1 Memory Reduction Comparison with Baseline FTLs

cd LeaFTL/wiscsee/leaftl_scripts
# Run batch of experiments (replace $PARALLELISM with an integer, e.g. 10; higher parallelism will consume memory but run the experiments faster)
./batch memory_batch $PARALLELISM
# Run plot scripts
./plot_all memory_batch

2.1.1 Expected Results

memory2.2 Performance Improvement Comparison with Baseline FTLs

cd LeaFTL/wiscsee/leaftl_scripts
# Run warmup workloads (replace $PARALLELISM with an integer in [1, 5], higher parallelism will consume memory but run the experiments faster)
pypy warmup $PARALLELISM
# Run batch of experiments (replace $PARALLELISM with an integer, e.g. 10)
./batch main_batch $PARALLELISM
# Run plot scripts
./plot_all main_batch

2.2.1 Expected Results

latency

lookup_cdf

2.3 Sensitivity Analsysis with Different Gamma

cd LeaFTL/wiscsee/leaftl_scripts
# Run batch of experiments (replace $PARALLELISM with an integer, e.g. 10)
./batch sensitivity_batch $PARALLELISM
# Run plot scripts
./plot_all sensitivity_batch

2.3.1 Expected Results

latency_sensitivity

memory_sensitivity

misprediction

leaftl's People

Contributors

my-henrys avatar

Stargazers

Andy avatar Haoyang Zhang avatar

Watchers

 avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.