gunrock / graphblast Goto Github PK
View Code? Open in Web Editor NEWHigh-Performance Linear Algebra-based Graph Primitives on GPUs
License: Apache License 2.0
High-Performance Linear Algebra-based Graph Primitives on GPUs
License: Apache License 2.0
Hi. Just noticed that there were some commits made yesterday rendering the software not buildable (via Makefile method)
==> 29525: graphblast: Executing phase: 'edit' [82/11013]
==> 29525: graphblast: Executing phase: 'build'
==> Error: ProcessError: Command exited with status 2:
'make' '-j12'
18 errors found in build log:
17 nvcc -g -gencode arch=compute_35,code=compute_35 -gencode arch=compute_60,code=compute_60 -O3 -use_fast_math -w -std=c++11 -o bin/ggc example/ggc.cu -Iext/moderngpu/include/ -Iex
t/cub/cub/ -I/data/ctcyang/boost_1_58_0/ -I./ ext/moderngpu/src/mgpucontext.cu ext/moderngpu/src/mgpuutil.cpp -L/data/ctcyang/boost_1_58_0/stage/lib/ -lboost_program_options -lcu
blas -lcusparse -lcurand
18 mkdir -p bin
19 mkdir -p bin
20 nvcc -g -gencode arch=compute_35,code=compute_35 -gencode arch=compute_60,code=compute_60 -O3 -use_fast_math -w -std=c++11 -o bin/ggc_cusparse example/ggc_cusparse.cu -Iext/moder
ngpu/include/ -Iext/cub/cub/ -I/data/ctcyang/boost_1_58_0/ -I./ ext/moderngpu/src/mgpucontext.cu ext/moderngpu/src/mgpuutil.cpp -L/data/ctcyang/boost_1_58_0/stage/lib/ -lboost_pr
ogram_options -lcublas -lcusparse -lcurand
21 nvcc -g -gencode arch=compute_35,code=compute_35 -gencode arch=compute_60,code=compute_60 -O3 -use_fast_math -w -std=c++11 -o bin/gpr example/gpr.cu -Iext/moderngpu/include/ -Iex
t/cub/cub/ -I/data/ctcyang/boost_1_58_0/ -I./ ext/moderngpu/src/mgpucontext.cu ext/moderngpu/src/mgpuutil.cpp -L/data/ctcyang/boost_1_58_0/stage/lib/ -lboost_program_options -lcu
blas -lcusparse -lcurand
22 nvcc -g -gencode arch=compute_35,code=compute_35 -gencode arch=compute_60,code=compute_60 -O3 -use_fast_math -w -std=c++11 -o bin/gtc example/gtc.cu -Iext/moderngpu/include/ -Iex
t/cub/cub/ -I/data/ctcyang/boost_1_58_0/ -I./ ext/moderngpu/src/mgpucontext.cu ext/moderngpu/src/mgpuutil.cpp -L/data/ctcyang/boost_1_58_0/stage/lib/ -lboost_program_options -lcu
blas -lcusparse -lcurand
>> 23 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
24
>> 25 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
26
>> 27 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
28
>> 29 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
30
>> 31 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
32
>> 33 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
34
>> 35 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
36
>> 37 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
38
>> 39 ./graphblas/backend/cuda/reduce.hpp(160): error: function template "graphblas::backend::reduceInner(T *, BinaryOpT, MonoidT, const graphblas::backend::SparseMatrix<a> *, graphbla
s::backend::Descriptor *)" has already been defined
40
41 1 error detected in the compilation of "/cache//tmpxft_000074a3_00000000-14_gpr.compute_60.cpp1.ii".
>> 42 make: *** [gpr] Error 1
43 make: *** Waiting for unfinished jobs....
44 1 error detected in the compilation of "/cache//tmpxft_000074a6_00000000-14_gtc.compute_60.cpp1.ii".
>> 45 make: *** [gtc] Error 1
46 1 error detected in the compilation of "/cache//tmpxft_0000749f_00000000-14_ggc.compute_60.cpp1.ii".
>> 47 make: *** [ggc] Error 1
48 1 error detected in the compilation of "/cache//tmpxft_00007493_00000000-14_gbfs.compute_60.cpp1.ii".
>> 49 make: *** [gbfs] Error 1
50 1 error detected in the compilation of "/cache//tmpxft_00007495_00000000-14_gsssp.compute_60.cpp1.ii".
>> 51 make: *** [gsssp] Error 1
52 1 error detected in the compilation of "/cache//tmpxft_00007497_00000000-14_glgc.compute_60.cpp1.ii".
53 1 error detected in the compilation of "/cache//tmpxft_00007494_00000000-14_gdiameter.compute_60.cpp1.ii".
>> 54 make: *** [glgc] Error 1
>> 55 make: *** [gdiameter] Error 1
56 1 error detected in the compilation of "/cache//tmpxft_0000749e_00000000-14_gmis.compute_60.cpp1.ii".
57 1 error detected in the compilation of "/cache//tmpxft_000074a2_00000000-14_ggc_cusparse.compute_60.cpp1.ii".
>> 58 make: *** [gmis] Error 1
>> 59 make: *** [ggc_cusparse] Error 1
I am attempting to build it using Spack HPC package manager, to which the graphblast package will be added tomorrow or so (details can be found here spack/spack#17289 )
Which version of boost, gcc, cuda do you guys use to build the latest version of the package? Thanks.
If you have a DAG of binary operations, you can traverse it in some topological order and generate proper bitcodes for your GPU kernel. OmniSci does it on parsed SQL queries, and more specifically different filter combinations, etc. TensorFlow does it based on the equation that needs to be minimized for gradient descent. Can we do it for GraphBLAS?
If you want to keep your kernel code as simple as possible, with minimal branches, etc., then there's no way doing it at compile-time. Unless you know what you're going to solve at compile time for SQL queries and tensor flow optimizations, you don't know about the exact details of the queries/equations at compile-time.
Both TensorFlow and OmniSci do real-time code generation using LLVM.
Hey,
I am trying to reproduce your Page Rank number on the socLiveJournal-1. I installed CUDA-9.1 and gcc-5.4.0. I downloaded the data from here. I remove the first comment lines and added the required ones to convert the data to an .mtx file:
%%MatrixMarket matrix coordinate pattern general
4847571 4847571 68993773
I compiled the Page Rank example (using the Makefile with make gpr
). Here is my output after running it:
-> % ./run_pr.sh
bin/gpr --timing 1 --mxvmode 0 --niter 100 --max_niter 1000 data/topc-datasets/soc-LiveJournal1.mtx
Undirected due to mtx: 0
Undirected due to cmd: 0
Undirected: 0
Remove self-loop: 1
Reading data/topc-datasets/.soc-LiveJournal1.mtx.d.nosl.bin
Allocate 4847572
Allocate 82170360
Allocate 82170360
Do not allocate 4847571 0x7f92c5367010
Do not allocate 68475300 0x7f92b19f2010
Do not allocate 68475300 0x7f929e07d010
Do not allocate 4847571 0x7f92c5367010
Do not allocate 68475300 0x7f92b19f2010
Do not allocate 68475300 0x7f929e07d010
output:
[0]:4.39931e-06 [1]:2.28478e-06 [2]:1.69434e-06 [3]:1.92988e-06 [4]:1.22651e-06 [5]:2.72542e-06 [6]:1.1593e-06 [7]:6.15954e-07 [8]:1.37225e-06 [9]:7.86397e-07 [10]:1.81577e-06 [11]:1.33162e-06 [12]:9.23166e-06 [13]:1.69406e-06 [14]:3.96623e-07 [15]:1.44772e-06 [16]:1.2477e-06 [17]:3.52301e-07 [18]:9.64398e-06 [19]:1.0859e-06 [20]:1.2196e-06 [21]:3.94725e-07 [22]:5.35532e-07 [23]:5.91095e-07 [24]:2.39766e-07 [25]:1.27003e-06 [26]:4.75216e-07 [27]:1.35969e-06 [28]:4.24832e-07 [29]:2.08527e-07 [30]:1.8405e-06 [31]:1.94205e-06 [32]:5.78724e-07 [33]:7.93741e-07 [34]:7.44357e-08 [35]:6.85211e-07 [36]:2.066e-07 [37]:8.49346e-07 [38]:1.00569e-06 [39]:6.59399e-07
CPU PR finished in 935.825134 msec. Search depth is: 3. Resultant: 0.000000
CORRECT
cpu, 957.859,
warmup, 294.753, 0
tight, 264.629
vxm, 290.707
CORRECT
If I interpret the numbers correctly one GPU iteration does 290ms whereas in the paper you mention that an iteration does 21ms. Also my GPU is a Tesla P100, which I believe is better than the one that you have used in your experiments. What am I doing wrong? I would appreciate any help!
Thank you for your time in advance.
Some TODOs:
spgemm_test.cu
program_options
and test
with commandline flag library such as gflags and googletestIs it okay that all examples fail on graphs small/test_spgemm.mtx and small/test_sgm.mtx?
Hi,
Thanks for the authors' contributions to GraphBLAS.
I am a beginner of GraphBLAS. I try to compile the GraphBLAS on a server with Ubuntu 20.04 and NVIDIA RTX 3080Ti GPU with CUDA Version=11.5. When I execute the make -j16
, there is some error information listed below.
(graphblast) bizhao.shi@dasys21-lc:~/research/compiler/graphblast/build$ make
Scanning dependencies of target graphblas
[ 1%] Linking CXX static library libgraphblas.a
[ 1%] Built target graphblas
[ 3%] Building NVCC (Device) object CMakeFiles/gspgemm.dir/ext/moderngpu/src/gspgemm_generated_mgpucontext.cu.o
nvcc warning : The 'compute_35', 'compute_37', 'sm_35', and 'sm_37' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-ta
rgets to suppress warning).
nvcc warning : The 'compute_35', 'compute_37', 'sm_35', and 'sm_37' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-ta
rgets to suppress warning).
ptxas info : 0 bytes gmem
ptxas info : Compiling entry function '_ZN4mgpu17KernelVersionShimEv' for 'sm_35'
ptxas info : Function properties for _ZN4mgpu17KernelVersionShimEv
0 bytes stack frame, 0 bytes spill stores, 0 bytes spill loads
ptxas info : Used 2 registers, 320 bytes cmem[0]
/rshome/bizhao.shi/research/compiler/graphblast/ext/moderngpu/src/mgpucontext.cu:126:6: warning: ‘template<class> class std::auto_ptr’ is deprecated [-Wdeprecated-d
eclarations]
126 | std::auto_ptr<DeviceGroup> deviceGroup;
| ^~~~~~~~
/usr/include/c++/9/bits/unique_ptr.h:53:25: note: declared here
53 | template<typename> class auto_ptr;
| ^~~~~~~~
/rshome/bizhao.shi/research/compiler/graphblast/ext/moderngpu/src/mgpucontext.cu:216:6: warning: ‘template<class> class std::auto_ptr’ is deprecated [-Wdeprecated-d
eclarations]
216 | std::auto_ptr<ContextGroup> contextGroup;
| ^~~~~~~~
/usr/include/c++/9/bits/unique_ptr.h:53:25: note: declared here
53 | template<typename> class auto_ptr;
| ^~~~~~~~
[ 4%] Building NVCC (Device) object CMakeFiles/gspgemm.dir/test/gspgemm_generated_gspgemm.cu.o
nvcc warning : The 'compute_35', 'compute_37', 'sm_35', and 'sm_37' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
In file included from /usr/local/cuda/include/thrust/detail/config/config.h:27,
from /usr/local/cuda/include/thrust/detail/config.h:23,
from /usr/local/cuda/include/thrust/iterator/iterator_facade.h:35,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/device/../iterator/arg_index_input_iterator.cuh:48,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/device/device_reduce.cuh:41,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/cub.cuh:53,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/backend/cuda/spmspv_inner.hpp:8,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/backend/cuda/cuda.hpp:13,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/graphblas.hpp:16,
from /rshome/bizhao.shi/research/compiler/graphblast/test/gspgemm.cu:12:
/usr/local/cuda/include/thrust/detail/config/cpp_dialect.h:131:13: warning: Thrust requires at least C++14. C++11 is deprecated but still supported. C++11 support will be removed in a future release. Define THRUST_IGNORE_DEPRECATED_CPP_DIALECT to suppress this message.
131 | THRUST_COMPILER_DEPRECATION_SOFT(C++14, C++11);
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
In file included from /usr/local/cuda/include/thrust/system/cuda/config.h:33,
from /usr/local/cuda/include/thrust/system/cuda/detail/execution_policy.h:35,
from /usr/local/cuda/include/thrust/iterator/detail/device_system_tag.h:23,
from /usr/local/cuda/include/thrust/iterator/detail/iterator_facade_category.h:22,
from /usr/local/cuda/include/thrust/iterator/iterator_facade.h:37,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/device/../iterator/arg_index_input_iterator.cuh:48,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/device/device_reduce.cuh:41,
from /rshome/bizhao.shi/research/compiler/graphblast/ext/cub/cub/cub.cuh:53,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/backend/cuda/spmspv_inner.hpp:8,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/backend/cuda/cuda.hpp:13,
from /rshome/bizhao.shi/research/compiler/graphblast/./graphblas/graphblas.hpp:16,
from /rshome/bizhao.shi/research/compiler/graphblast/test/gspgemm.cu:12:
/usr/local/cuda/include/cub/util_namespace.cuh:46:2: error: #error CUB requires a definition of CUB_NS_QUALIFIER when CUB_NS_PREFIX/POSTFIX are defined.
46 | #error CUB requires a definition of CUB_NS_QUALIFIER when CUB_NS_PREFIX/POSTFIX are defined.
| ^~~~~
CMake Error at gspgemm_generated_gspgemm.cu.o.cmake:220 (message):
Error generating
/rshome/bizhao.shi/research/compiler/graphblast/build/CMakeFiles/gspgemm.dir/test/./gspgemm_generated_gspgemm.cu.o
make[2]: *** [CMakeFiles/gspgemm.dir/build.make:65: CMakeFiles/gspgemm.dir/test/gspgemm_generated_gspgemm.cu.o] Error 1
make[1]: *** [CMakeFiles/Makefile2:101: CMakeFiles/gspgemm.dir/all] Error 2
make: *** [Makefile:84: all] Error 2
I would like to ask for the solution of CUB error, thx!
Currently, desc.loadArgs
is required to load many default values. This is undesirable from an user interface perspective. For example, "Error: Simple kernel unmasked not implemented yet! will be thrown if desc.loadArgs
is not called by test/gmxv.cu
.
I found it hard to decipher algorithm names such as lgc vs gc. One would think the former is a special case of the latter but in fact they are completely different, because "c" refers to "clustering" in lgc but to "coloring" in the second. Perhaps longer names with more comments would help?
Hi,
I tried compiling this against CUDA 11.0 on Ubuntu 18.04 with the nvidia HPC SDK version 2021_212, and running
../bin/gbfs ../data/small/chesapeake.mtx
prints a plethora of error messages:
Runtime error: reduceInner(val, accum, op, &u->sparse_, desc) returned 12 at /home/mgara/software/graphblast/./graphblas/backend/cuda/operations.hpp:1012
none of which are very informative. At this point I assume this project is highly experimental and unstable?
I've tested mis algorithm on graph test_mis.mtx and got this error:
Cuda error in file './graphblas/backend/cuda/reduce.hpp' in line 40 : invalid configuration argument.
System configuration:
GPU: NVidia GeForce GT 1030
NVIDIA-SMI 440.33.01
OS: Ubuntu 16.04 (I run in docker container)
CUDA Version: 9.2
g++ version 4.9
Hello,
Thank you for hosting Graphblast on a public repo to help the research community.
I was wondering whether there is any plan to get GraphBlast working for the latest sms. I am finding the mgpu version leveraged by GraphBlast a little bit challenging to get it to work on latest sms for Graphblast. I tried to put in some patches in the mgpu version currently being used by Grtaphblast, in particular, for the synchronization primitives (mostly shuffles and ballots suggested by @neoblizz in the mgpu repo). and I am encountering hangs for algorithms such as bfs with matrices of medium size.
I would really appreciate any insight. Thanks in advance!
Hi There,
I tried to install the code on my ubuntu-16.04 node, with gcc-4.8.5, cuda 9.0.
But I got the following errors. Hope you can let me know how to solve this problem.
mkdir -p bin nvcc -g -gencode arch=compute_60,code=compute_60 -O3 -use_fast_math -w -std=c++11 -o bin/gbfs example/gbfs.cu -Iext/moderngpu/include/ -Iext/cub/cub/ -Iext/boost_1_58_/ -I./ ext/moderngpu/src/mgpucontext.cu ext/moderngpu/src/mgpuutil.cpp -Lext/boost_1_58_0/stage/lib/ -lboost_program_options -lcublas -lcusparse -lcurand /tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function
boost::program_options::variables_map::operator[](std::string const&) const':
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /usr/include/boost/program_options/variables_map.hpp:155: undefined reference to
boost::program_options::abstract_variables_map::operator[](std::string const&) const'
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /usr/include/boost/program_options/variables_map.hpp:155: undefined reference to
boost::program_options::abstract_variables_map::operator[](std::string const&) const'
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:/usr/include/boost/program_options/variables_map.hpp:155: more undefined references to
boost::program_options::abstract_variables_map::operator[](std::string const&) const' follow
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function parseArgs(int, char**, boost::program_options::variables_map*)': /vpublic01/frog/zhengzhigao/addtest/graphblast/./graphblas/util.hpp:41: undefined reference to
boost::program_options::options_description::options_description(std::string const&, unsigned int, unsigned int)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<std::string, char>::xparse(boost::any&, std::vector<std::string, std::allocator<std::string> > const&) const': /usr/include/boost/program_options/detail/value_semantic.hpp:167: undefined reference to
boost::program_options::validate(boost::any&, std::vector<std::string, std::allocatorstd::string > const&, std::string*, int)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<bool, char>::xparse(boost::any&, std::vector<std::string, std::allocator<std::string> > const&) const': /usr/include/boost/program_options/detail/value_semantic.hpp:167: undefined reference to
boost::program_options::validate(boost::any&, std::vector<std::string, std::allocatorstd::string > const&, bool*, int)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::validation_error::validation_error(boost::program_options::validation_error::kind_t, std::string const&, std::string const&, int)': /usr/include/boost/program_options/errors.hpp:372: undefined reference to
boost::program_options::validation_error::get_template(boost::program_options::validation_error::kind_t)'
/usr/include/boost/program_options/errors.hpp:372: undefined reference to boost::program_options::error_with_option_name::error_with_option_name(std::string const&, std::string const&, std::string const&, int)' /tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function
boost::program_options::variables_map::operator[](std::string const&) const':
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /usr/include/boost/program_options/variables_map.hpp:155: undefined reference to
boost::program_options::abstract_variables_map::operator[](std::string const&) const'
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /usr/include/boost/program_options/variables_map.hpp:155: undefined reference to
boost::program_options::abstract_variables_map::operator[](std::string const&) const'
/usr/include/boost/program_options/variables_map.hpp:155: undefined reference to boost::program_options::abstract_variables_map::operator[](std::string const&) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:/usr/include/boost/program_options/variables_map.hpp:155: more undefined references to
boost::program_options::abstract_variables_map::operator[](std::string const&) const' follow
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<float, char>::name() const': /usr/include/boost/program_options/detail/value_semantic.hpp:19: undefined reference to
boost::program_options::arg'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<bool, char>::name() const': /usr/include/boost/program_options/detail/value_semantic.hpp:19: undefined reference to
boost::program_options::arg'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<std::string, char>::name() const': /usr/include/boost/program_options/detail/value_semantic.hpp:19: undefined reference to
boost::program_options::arg'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::typed_value<int, char>::name() const': /usr/include/boost/program_options/detail/value_semantic.hpp:19: undefined reference to
boost::program_options::arg'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function to_internal<std::basic_string<char> >': /usr/include/boost/program_options/detail/convert.hpp:79: undefined reference to
boost::program_options::to_internal(std::string const&)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function basic_command_line_parser': /usr/include/boost/program_options/detail/parsers.hpp:39: undefined reference to
boost::program_options::detail::cmdline::cmdline(std::vector<std::string, std::allocatorstd::string > const&)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function boost::program_options::basic_command_line_parser<char>::extra_parser(boost::function1<std::pair<std::string, std::string>, std::string const&>)': /usr/include/boost/program_options/detail/parsers.hpp:77: undefined reference to
boost::program_options::detail::cmdline::set_additional_parser(boost::function1<std::pair<std::string, std::string>, std::string const&>)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function to_internal<std::basic_string<char> >': /usr/include/boost/program_options/detail/convert.hpp:79: undefined reference to
boost::program_options::to_internal(std::string const&)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function void boost::program_options::validate<int, char>(boost::any&, std::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&, int*, long)': /usr/include/boost/program_options/detail/value_semantic.hpp:92: undefined reference to
boost::program_options::invalid_option_value::invalid_option_value(std::string const&)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o: In function void boost::program_options::validate<float, char>(boost::any&, std::vector<std::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::basic_string<char, std::char_traits<char>, std::allocator<char> > > > const&, float*, long)': /usr/include/boost/program_options/detail/value_semantic.hpp:92: undefined reference to
boost::program_options::invalid_option_value::invalid_option_value(std::string const&)'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost16exception_detail19error_info_injectorINS_15program_options20invalid_option_valueEEE[_ZTVN5boost16exception_detail19error_info_injectorINS_15program_options20invalid_option_valueEEE]+0x30): undefined reference to boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_15program_options20invalid_option_valueEEEEE[_ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_15program_options20invalid_option_valueEEEEE]+0x38): undefined reference to
boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost16exception_detail19error_info_injectorINS_15program_options16validation_errorEEE[_ZTVN5boost16exception_detail19error_info_injectorINS_15program_options16validation_errorEEE]+0x30): undefined reference to boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_15program_options16validation_errorEEEEE[_ZTVN5boost16exception_detail10clone_implINS0_19error_info_injectorINS_15program_options16validation_errorEEEEE]+0x38): undefined reference to
boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options16validation_errorE[_ZTVN5boost15program_options16validation_errorE]+0x30): undefined reference to boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options20invalid_option_valueE[_ZTVN5boost15program_options20invalid_option_valueE]+0x30): more undefined references to
boost::program_options::error_with_option_name::substitute_placeholders(std::string const&) const' follow
/tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options11typed_valueIicEE[_ZTVN5boost15program_options11typed_valueIicEE]+0x38): undefined reference to boost::program_options::value_semantic_codecvt_helper<char>::parse(boost::any&, std::vector<std::string, std::allocator<std::string> > const&, bool) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options11typed_valueISscEE[_ZTVN5boost15program_options11typed_valueISscEE]+0x38): undefined reference to
boost::program_options::value_semantic_codecvt_helper::parse(boost::any&, std::vector<std::string, std::allocatorstd::string > const&, bool) const'
/tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options11typed_valueIbcEE[_ZTVN5boost15program_options11typed_valueIbcEE]+0x38): undefined reference to boost::program_options::value_semantic_codecvt_helper<char>::parse(boost::any&, std::vector<std::string, std::allocator<std::string> > const&, bool) const' /tmp/tmpxft_0000012c_00000000-13_gbfs.o:(.rodata._ZTVN5boost15program_options11typed_valueIfcEE[_ZTVN5boost15program_options11typed_valueIfcEE]+0x38): undefined reference to
boost::program_options::value_semantic_codecvt_helper::parse(boost::any&, std::vector<std::string, std::allocatorstd::string > const&, bool) const'
collect2: error: ld returned 1 exit status
Makefile:17: recipe for target 'gbfs' failed
make: *** [gbfs] Error 1`
I test all algorithms from /example on graphs from /data/small and some of them (ggc, gmis, gpr) give the error message: Error: eWiseAdd sparse-sparse not implemented yet!
. Could you tell me if you're planning to implement the operation, please? If yes, then when?
Also, I've got this error while running gmis:
Error: Feature not implemented yet!
To reproduce:
make -j8
bin/gpr --niter 0 --mxvmode 0 --directed 2 data/small/test_pr.mtx
System configuration:
GPU: NVidia GeForce GT 1030
NVIDIA-SMI 440.33.01
OS: Ubuntu 16.04 (I run in docker container)
CUDA Version: 9.2
g++ version 4.9
Running gspgemm unit test results in:
Cuda error in file '/home/ctcyang/graphblast-ben/./graphblas/backend/cuda/sparse_matrix.hpp' in line 167 : invalid device pointer.
The way to fix this is to either:
I've built the library using CMake with the corresponding script and tried to run test/gspgemm.cu
example. I have modified it a bit for it to read the matrix of mine. I had one run with PlusMultipliesSemiring
and got the correct output. Then I ran the same example with MinimumPlusSemiring
and got the same result as the PlusMultipliesSemiring
one (which is incorrect in the case of minplus semiring).
The minimal reproducible example (Note: another issue is that uncommenting the lines at the bottom leads to segfault):
#define GRB_USE_CUDA
#define private public
#include <iostream>
#include <algorithm>
#include <string>
#include <cstdio>
#include <cstdlib>
#include <boost/program_options.hpp>
#include "graphblas/graphblas.hpp"
#include "test/test.hpp"
int main( int argc, char** argv )
{
bool DEBUG = true;
std::vector<graphblas::Index> a_row_indices, b_row_indices;
std::vector<graphblas::Index> a_col_indices, b_col_indices;
std::vector<float> a_values, b_values;
graphblas::Index a_num_rows, a_num_cols, a_num_edges;
graphblas::Index b_num_rows, b_num_cols, b_num_edges;
char* dat_name;
// Load A
std::cout << "loading A" << std::endl;
readMtx("path/to/example/matrix", &a_row_indices, &a_col_indices,
&a_values, &a_num_rows, &a_num_cols, &a_num_edges, 1, false);
graphblas::Matrix<float> a(a_num_rows, a_num_cols);
a.build(&a_row_indices, &a_col_indices, &a_values, a_num_edges, GrB_NULL,
GrB_NULL);
if(DEBUG) a.print();
// Load B, i.e. the matrix is squared
std::cout << "loading B" << std::endl;
readMtx("path/to/example/matrix", &b_row_indices, &b_col_indices,
&b_values, &b_num_rows, &b_num_cols, &b_num_edges, 1, false);
graphblas::Matrix<float> b(b_num_rows, b_num_cols);
b.build(&b_row_indices, &b_col_indices, &b_values, b_num_edges, GrB_NULL,
GrB_NULL );
if(DEBUG) b.print();
//
graphblas::Matrix<float> c(a_num_rows, b_num_cols);
graphblas::Descriptor desc;
desc.descriptor_.debug_ = true;
graphblas::mxm<float,float,float,float>(
&c,
GrB_NULL,
GrB_NULL,
graphblas::MinimumPlusSemiring<float>(),
&a,
&b,
&desc
);
if(DEBUG) c.print();
// // Multiply using GPU array initialization.
// graphblas::Matrix<float> A(a_num_rows, a_num_cols);
// graphblas::Matrix<float> B(b_num_rows, b_num_cols);
// graphblas::Matrix<float> C(a_num_rows, b_num_cols);
// A.build(a.matrix_.sparse_.d_csrRowPtr_, a.matrix_.sparse_.d_csrColInd_, a.matrix_.sparse_.d_csrVal_, a.matrix_.sparse_.nvals_);
// B.build(b.matrix_.sparse_.d_csrRowPtr_, b.matrix_.sparse_.d_csrColInd_, b.matrix_.sparse_.d_csrVal_, b.matrix_.sparse_.nvals_);
// desc.descriptor_.debug_ = true;
// graphblas::mxm<T, T, T, T>(&C, GrB_NULL, GrB_NULL, graphblas::PlusMultipliesSemiring<float>(),
// &A, &B, &desc);
// Multiply using CPU array initialization.
// TODO(ctcyang): Add EXPECT_FAIL, because require pointers to be GPU.
/*graphblas::Matrix<float> a_(a_num_rows, a_num_cols);
graphblas::Matrix<float> b_(b_num_rows, b_num_cols);
graphblas::Matrix<float> c_(a_num_rows, b_num_cols);
a_.build(a.matrix_.sparse_.h_csrRowPtr_, a.matrix_.sparse_.h_csrColInd_, a.matrix_.sparse_.h_csrVal_, a.matrix_.sparse_.nvals_);
b_.build(b.matrix_.sparse_.h_csrRowPtr_, b.matrix_.sparse_.h_csrColInd_, b.matrix_.sparse_.h_csrVal_, b.matrix_.sparse_.nvals_);
desc.descriptor_.debug_ = true;
graphblas::mxm<T, T, T, T>(&c_, GrB_NULL, GrB_NULL, graphblas::PlusMultipliesSemiring<float>(),
&a_, &b_, &desc);*/
}
example matrix:
%%MatrixMarket matrix coordinate real general
3 3 6
1 1 1
1 2 2
1 3 3
2 1 2
2 3 1
3 3 2
gspgemm with PlusMultipliesSemiring
yields:
%%MatrixMarket matrix coordinate real general
3 3 7
1 1 5
1 2 2
1 3 11
2 1 2
2 2 4
2 3 8
3 3 4
which is correct, while MinimumPlusSemiring
yields the same result as above when the correct one is:
%%MatrixMarket matrix coordinate real general
3 3 7
1 1 2
1 2 3
1 3 3
2 1 3
2 2 4
2 3 3
3 3 4
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.