dgsparse / dgsparse-lib Goto Github PK

View Code? Open in Web Editor NEW

94.0 2.0 24.0 8.4 MB

PyTorch-Based Fast and Efficient Processing for Various Machine Learning Applications with Diverse Sparsity

Home Page: https://dgsparse.github.io/

License: MIT License

C 1.36% Cuda 68.41% C++ 13.83% Makefile 0.73% Python 13.78% Dockerfile 0.25% Shell 1.64%

gnn sparse-matrix spconv

dgsparse-lib's Introduction

dgSPARSE Library

Introdution

The dgSPARSE Library (Deep Geometric Sparse Library) is a high performance library for sparse kernel acceleration on GPUs based on CUDA. Now we aims to provide PyTorch-Based Fast and Efficient Processing kernel for users to have better experience in running applications like GNN, Rec sys and 3D pointcloud detection.

Installation

First, setup the the following environment variables:

export CUDA_HOME=/usr/local/cuda # your cuda path
export LD_LIBRARY_PATH=$CUDA_HOME/lib64 # your cuda lib path

Then, install with conda.

conda install -c dgsparse dgsparse

Or you can build from source

pip install -e .

A demo for SpMM inference time compared to other main-stream library. (Tested on RTX 3090 with feature=64).

Run Examples

Previously we provide C++ examples for SpMM and SDDMM kernels. To run these examples, please build dgsparse through make exp. Then, you could run our kernels in the example folder. Check more details in README under example directory.

Documentation

Our new docs for python API will be coming soon! Now you can refer to dgSPARSE Library Documentation for more details.

dgsparse-lib's People

Contributors

Stargazers

Watchers

dgsparse-lib's Issues

example/ge-spmm

When running example ./spmm.out ../data/p2p-Gnutella31.mtx X
if X is not a 2^n integer
it reports like Wrong result: i = 363, j = 0, result = 0.000000, reference = 0.070000.

Only when X is a 2^n integer we can view the result correctly

[ERROR] Compile error with cuda_version(noexecpt torch custom op)

Command Line

python setup.py install

Env

Ubuntu-18.04
gcc-9.4.0
cuda-11.8
python-3.8.10
pytorch-2.1.0

Error info

/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/utils/cpp_extension.py:399: UserWarning: There are no x86_64-linux-gnu-g++ version bounds defined for CUDA version 11.8
  warnings.warn(f'There are no {compiler_name} version bounds defined for CUDA version {cuda_str_version}')
building 'dgsparse._version_cuda' extension
x86_64-linux-gnu-gcc -pthread -Wno-unused-result -Wsign-compare -DNDEBUG -g -fwrapv -O2 -Wall -g -fstack-protector-strong -Wformat -Werror=format-security -g -fwrapv -O2 -g -fstack-protector-strong -Wformat -Werror=format-security -Wdate-time -D_FORTIFY_SOURCE=2 -fPIC -DWITH_PYTHON -DWITH_CUDA -Isrc -I/home/lzy/intel/oneapi/mkl -I/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include -I/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/csrc/api/include -I/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/TH -I/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/THC -I/home/lzy/BASE/cuda_118/include -I/home/lzy/dgsparse/env/include -I/usr/include/python3.8 -c src/version.cpp -o build/temp.linux-x86_64-3.8/src/version.o -O2 -DTORCH_API_INCLUDE_EXTENSION_H -DPYBIND11_COMPILER_TYPE="_gcc" -DPYBIND11_STDLIB="_libstdcpp" -DPYBIND11_BUILD_ABI="_cxxabi1011" -DTORCH_EXTENSION_NAME=_version_cuda -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++17
src/version.cpp:19:73: error: no matching function for call to ‘c10::RegisterOperators::op(const char [23], int64_t (*)() noexcept)’
   19 |    torch::RegisterOperators().op("dgsparse::cuda_version", &cuda_version);
      |                                                                         ^
In file included from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/library.h:67,
                 from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/csrc/api/include/torch/types.h:13,
                 from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/script.h:3,
                 from src/version.cpp:4:
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:455:23: note: candidate: ‘c10::RegisterOperators&& c10::RegisterOperators::op(c10::RegisterOperators::Options&&) &&’
  455 |   RegisterOperators&& op(Options&& options) && {
      |                       ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:455:23: note:   candidate expects 1 argument, 2 provided
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:461:22: note: candidate: ‘c10::RegisterOperators& c10::RegisterOperators::op(c10::RegisterOperators::Options&&) &’
  461 |   RegisterOperators& op(Options&& options) & {
      |                      ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:461:22: note:   candidate expects 1 argument, 2 provided
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:471:23: note: candidate: ‘c10::RegisterOperators&& c10::RegisterOperators::op(const string&, c10::RegisterOperators::Options&&) &&’
  471 |   RegisterOperators&& op(const std::string& schemaOrName, Options&& options = RegisterOperators::options()) && {
      |                       ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:471:69: note:   no known conversion for argument 2 from ‘int64_t (*)() noexcept’ {aka ‘long int (*)() noexcept’} to ‘c10::RegisterOperators::Options&&’
  471 |   RegisterOperators&& op(const std::string& schemaOrName, Options&& options = RegisterOperators::options()) && {
      |                                                           ~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:476:23: note: candidate: ‘c10::RegisterOperators&& c10::RegisterOperators::op(c10::FunctionSchema, c10::RegisterOperators::Options&&) &&’
  476 |   RegisterOperators&& op(FunctionSchema schema, Options&& options) && {
      |                       ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:476:41: note:   no known conversion for argument 1 from ‘const char [23]’ to ‘c10::FunctionSchema’
  476 |   RegisterOperators&& op(FunctionSchema schema, Options&& options) && {
      |                          ~~~~~~~~~~~~~~~^~~~~~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:522:4: note: candidate: ‘template<class FuncType> std::enable_if_t<(c10::guts::is_function_type<T>::value && (! std::is_same<FuncType, void(const c10::OperatorHandle&, std::vector<c10::IValue>*)>::value)), c10::RegisterOperators&&> c10::RegisterOperators::op(const string&, FuncType*, c10::RegisterOperators::Options&&) &&’
  522 |    op(const std::string& schemaOrName, FuncType* func, Options&& options = RegisterOperators::options()) && {
      |    ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:522:4: note:   template argument deduction/substitution failed:
In file included from /usr/include/c++/9/bits/move.h:55,
                 from /usr/include/c++/9/bits/stl_pair.h:59,
                 from /usr/include/c++/9/bits/stl_algobase.h:64,
                 from /usr/include/c++/9/bits/specfun.h:45,
                 from /usr/include/c++/9/cmath:1927,
                 from /usr/include/c++/9/math.h:36,
                 from /usr/include/python3.8/pyport.h:212,
                 from /usr/include/python3.8/Python.h:63,
                 from src/version.cpp:2:
/usr/include/c++/9/type_traits: In substitution of ‘template<bool _Cond, class _Tp> using enable_if_t = typename std::enable_if::type [with bool _Cond = (((bool)std::integral_constant<bool, false>::value) && (!(bool)std::integral_constant<bool, false>::value)); _Tp = c10::RegisterOperators&&]’:
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:522:4:   required by substitution of ‘template<class FuncType> std::enable_if_t<(c10::guts::is_function_type<T>::value && (! std::is_same<FuncType, void(const c10::OperatorHandle&, std::vector<c10::IValue>*)>::value)), c10::RegisterOperators&&> c10::RegisterOperators::op(const string&, FuncType*, c10::RegisterOperators::Options&&) && [with FuncType = long int() noexcept]’
src/version.cpp:19:73:   required from here
/usr/include/c++/9/type_traits:2378:11: error: no type named ‘type’ in ‘struct std::enable_if<false, c10::RegisterOperators&&>’
 2378 |     using enable_if_t = typename enable_if<_Cond, _Tp>::type;
      |           ^~~~~~~~~~~
In file included from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/library.h:67,
                 from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/csrc/api/include/torch/types.h:13,
                 from /home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/torch/script.h:3,
                 from src/version.cpp:4:
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:551:5: note: candidate: ‘template<class Lambda> std::enable_if_t<(c10::guts::is_functor<LambdaType>::value && c10::guts::detail::is_stateless_lambda_<typename std::decay<typename std::decay<_Tp>::type>::type, void>::value), c10::RegisterOperators&&> c10::RegisterOperators::op(const string&, Lambda&&, c10::RegisterOperators::Options&&) &&’
  551 |     op(const std::string& schemaOrName, Lambda&& lambda, Options&& options = RegisterOperators::options()) && {
      |     ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:551:5: note:   template argument deduction/substitution failed:
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:568:5: note: candidate: ‘template<class Lambda> std::enable_if_t<(c10::guts::is_functor<LambdaType>::value && (! c10::guts::detail::is_stateless_lambda_<typename std::decay<typename std::decay<_Tp>::type>::type, void>::value)), c10::RegisterOperators&&> c10::RegisterOperators::op(const string&, Lambda&&, c10::RegisterOperators::Options&&) &&’
  568 |     op(const std::string& schemaOrName, Lambda&& lambda, Options&& options = RegisterOperators::options()) && {
      |     ^~
/home/lzy/dgsparse/env/lib/python3.8/site-packages/torch/include/ATen/core/op_registration/op_registration.h:568:5: note:   template argument deduction/substitution failed:
src/version.cpp:18:13: warning: ‘registry’ defined but not used [-Wunused-variable]
   18 | static auto registry =
      |             ^~~~~~~~
error: command 'x86_64-linux-gnu-gcc' failed with exit status 1

src/ge-spmm make failure

Steps to reproduce

git clone https://github.com/dgSPARSE/dgSPARSE-Library.git
cd dgSPARSE-Library/src/ge-spmm
make

Errors:

# output of 'make'
...
cp gespmm_v2.cu.o csrspmm_parreduce.cu.o gespmm_csrcoo_v2.cu.o csrspmm_rowcaching.cu.o csrspmm_seqreduce.cu.o csrspmm_non_transpose.cu.o gespmm.cc gespmm_v2.cu csrspmm_parreduce.cu gespmm_csrcoo_v2.cu csrspmm_rowcaching.cu csrspmm_seqreduce.cu csrspmm_non_transpose.cu gespmm.cc.o ../../objs
cp: target '../../objs' is not a directory
make: *** [Makefile:29: libgespmm.a] Error 1

Recommend Projects

React

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

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

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

An Open Source Machine Learning Framework for Everyone
Django

The Web framework for perfectionists with deadlines.
Laravel

A PHP framework for web artisans
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.
Visualization

Some thing interesting about visualization, use data art
Game

Some thing interesting about game, make everyone happy.

Recommend Org

Facebook

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

Open source projects and samples from Microsoft.
Google

Google ❤️ Open Source for everyone.
Alibaba

Alibaba Open Source for everyone
D3

Data-Driven Documents codes.
Tencent

China tencent open source team.