Skip to content

README.md

Latest commit

 

History

History
129 lines (94 loc) · 6.08 KB

README.md

File metadata and controls

129 lines (94 loc) · 6.08 KB

ReadMe

Setup

Requirements

  • cmake (>= 3.10)
  • GNU Make (4.2.1)
  • LLVM (10.X) - src, release
  • Python (3.10), C++17
  • gRPC v1.34 and protobuf v3.13 - for gRPC Model Runner
    • Building GRPC from Source: Please follow Build GRPC with cmake v1.34 (protobuf v3.13) to build GRPC from source.
    • In the above tutorial setting DCMAKE_INSTALL_PREFIX is necessary as it would give you an easy way to uninstall GRPC later.
    • The following dependencies will be required for Python: . # remove this
  • ONNXRuntime v1.16.3
  • TensorFlow - for TF Model Runner (AOT flow) # this should be in the yml only don't need to set it up separately
    • Tested with TensorFlow 2.13.0
  • Other python requirements are available in [mlbridge.yml] # needs to be updated with sangamesh's.yml
    • Conda/Anaconda based virtual environment is assumed

(Experiments are done on an Ubuntu 20.04 machine) (LMAO)

Build

set up env :

cp -r /Pramana/ML_LLVM_Tools/AE/envs/ ~/

# install the env using the following commands
conda env create -f ~/env/LOF_original_env.yml
conda env create -f ~/env/mlgo-new

wget https://github.com/microsoft/onnxruntime/releases/download/v1.16.3/onnxruntime-linux-x64-1.16.3.tgz
tar -xvf onnxruntime-linux-x64-1.16.3.tgz

# get GRPC working 

# switch to mlgo-new env as you will need it to build the setup
conda activate mlgo-new 

# rename files in your conda enviornment
mv ~/anaconda3/envs/mlgo-new/lib/python3.10/site-packages/tensorflow/include/google/ ~/anaconda3/envs/mlgo-new/lib/python3.10/site-packages/tensorflow/include/google_new/

mv ~/anaconda3/envs/mlgo-new/include/google/ ~/anaconda3/envs/mlgo-new/include/google_new/

git clone git@github.com:IITH-Compilers/ml-llvm-project.git
cd ml-llvm-project
git branch mlbridge-lib
git pull
git submodule update --init --recursive

# go to mono repo dir

mkdir build
cd build

# build command 
	cmake -G "Unix Makefiles" -S ../llvm -B . \                                         
	-DCMAKE_BUILD_TYPE="Release" \
	-DLLVM_ENABLE_PROJECTS="clang;IR2Vec;ml-llvm-tools;mlir;MLCompilerBridge" \
	-DLLVM_TARGETS_TO_BULID="X86" \
	-DLLVM_ENABLE_ASSERTIONS=on \
	-DCMAKE_EXPORT_COMPILE_COMMANDS=1 \
	-DLLVM_CCACHE_BUILD=ON \
	-DONNXRUNTIME_ROOTDIR=$HOME/onnxruntime-linux-x64-1.16.3 \ # change to your path where you wget the onnxruntime
	-DLLVM_TF_AOT_RUNTIME=/home/intern24005/miniconda3/envs/mlgo-new/lib/python3.10/site-packages/tensorflow \ # change to your path 
	-DTENSORFLOW_AOT_PATH=/home/intern24005/miniconda3/envs/mlgo-new/lib/python3.10/site-packages/tensorflow \ # change to your path
	-DLLVM_INLINER_MODEL_PATH=download \
	-DLLVM_INLINER_MODEL_CURRENT_URL=https://github.com/google/ml-compiler-opt/releases/download/inlining-Oz-v1.1/inlining-Oz-99f0063-v1.1.tar.gz \
	-DLLVM_RAEVICT_MODEL_PATH=download \
	-DLLVM_RAEVICT_MODEL_CURRENT_URL=https://github.com/google/ml-compiler-opt/releases/download/regalloc-evict-v1.0/regalloc-evict-e67430c-v1.0.tar.gz
	       

make clang opt -j50

#for inference 

# add the onnxruntime to your path
export LIBRARY_PATH=$LIBRARY_PATH:/home/intern24005/onnxruntime-linux-x64-1.16.3/lib   

# add the

Static training

conda activate LPD

Config changes

  • Make changed to run.sh if required
    • LLVM build path

How to train?

  • bash run.sh trainInv <dataset> <POST distribution passes choice: {0, 1, 2}> <rewardtype R or S> [<reamrks>]

    • trainInv - Indicate to run the training
    • <dataset> - Path of the dataset (dataset should have graphs/train folder)
    • POST distribution passes choice - {0 -> No transformation, 1 -> Loop Vectorization pass, 2 -> Whole Post distribution passes sequence}
    • S - Specify use of static rewards
    • remarks - Optional -Add some remarks for training

  • Example - bash trainInv ../../../../data/New_mutated_data/generated_final/ 2 S Regular

    • Print the location of the logs
    • Print the location of the trained_model LLVM build directory selected for training : /home/venkat/IF-DV/Rohit/IR2Vec-LoopOptimizationFramework/build_LoopCost_REL_AsrtON Running the training using trainInv.py................... Location of the trained model: /home/venkat/IF-DV/Rohit/IR2Vec-LoopOptimizationFramework/model/ggnn_drl/static_v2/trained_model/New_mutated_data/generated_final/ELC_False_AMF_False_LoopCost_REL_AsrtON_PDP_2_Regular/trainInv Location of the generated llfiles and outfiles : /home/venkat/IF-DV/Rohit/IR2Vec-LoopOptimizationFramework/model/ggnn_drl/static_v2/trained_model/New_mutated_data/generated_final/ELC_False_AMF_False_LoopCost_REL_AsrtON_PDP_2_Regular/trainInv
      Logs files: /home/venkat/IF-DV/Rohit/IR2Vec-LoopOptimizationFramework/model/ggnn_drl/static_v2/trained_model/New_mutated_data/generated_final/ELC_False_AMF_False_LoopCost_REL_AsrtON_PDP_2_Regular/trainInv/log

How to test the model?

  • bash run.sh test <path dataset> <path of model> <disable runtime calc:Y> <POST distribution passes choice: {0, 1, 2}> <rewardtype R or S> <remarks>

    • test - Indicate to run the testing
    • <dataset> - Path of the dataset to test (dataset should have graphs/test folder)
    • path of model - Path of the model
    • disable runtime calc - Pass Y to disable runtime calucations
    • POST distribution passes choice - {0 -> No transformation, 1 -> Loop Vectorization pass, 2 -> Whole Post distribution passes sequence}
    • S - Specify use of static rewards
    • remarks - Optional -Add some remarks for training

  • Example - bash run.sh test ../../../../data/SPEC_NEW_UNLINK_Ind_iv_REL_AsrtON/generated_final/ /home/venkat/IF-DV/Rohit/IR2Vec-LoopOptimizationFramework/model/ggnn_drl/static_v2/trained_model/New_mutated_data/generated_final/ELC_False_AMF_False_LoopCost_REL_AsrtON_PDP_2_Regular/trainInv/checkpoint-graphs-11.pth Y 2 S parest

Note

  • POST distribution passes choice - value should be same for training and testing
  • On every run for training, directories are created if not exists and if exists, delete the old data and use it.