Merge pull request #84 from bacpop/blas-fix

BLAS linking; CI fix; better message for deprecated function

CMakeLists.txt

@@ -57,9 +57,13 @@ include_directories(${CMAKE_CURRENT_SOURCE_DIR}/src)
 find_package(ZLIB)
 execute_process(COMMAND pybind11-config --cmakedir OUTPUT_STRIP_TRAILING_WHITESPACE OUTPUT_VARIABLE pybind11_DIR)
 find_package(pybind11 2.6 CONFIG REQUIRED)
+
+# Linear algebra packages
 find_package(Eigen3 3.3 REQUIRED NO_MODULE)
-find_package(Armadillo REQUIRED)
-include_directories(${ARMADILLO_INCLUDE_DIRS})
+find_package(BLAS)
+if(BLAS_FOUND)
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -DEIGEN_USE_BLAS")
+endif()
 
 # Define python library target
 add_library("${TARGET_NAME}" MODULE)
@@ -76,7 +80,7 @@ if(CMAKE_CUDA_COMPILER)
         # -Xptxas -dlcm=ca turns cache on, but not needed in recent nvcc versions
         # --cudart static: static linking of the CUDA libraries
         # -gencode arch=compute_35 etc compiles for each (minimum) device version listed (v3.5, v5.0, v7.5)
-        set(CUDA_OPTS "-Xcompiler -fPIC -Xptxas -dlcm=ca --relocatable-device-code=true --expt-relaxed-constexpr")
+        set(CUDA_OPTS "-Xcompiler -fPIC -Xptxas -dlcm=ca --relocatable-device-code=true --expt-relaxed-constexpr --diag-suppress 20236")
         # Turn on link time optimisation if available
         if(CMAKE_BUILD_TYPE MATCHES Release)
             string(APPEND CUDA_OPTS " -dlto -arch=sm_86")
@@ -123,6 +127,8 @@ target_sources("${TARGET_NAME}" PRIVATE src/sketchlib_bindings.cpp
                                     src/database/database.cpp
                                     src/api.cpp
                                     src/dist/linear_regression.cpp
+                                    src/random/kmeans/KMeansRexCore.cpp
+                                    src/random/kmeans/mersenneTwister2002.c
                                     src/random/rng.cpp
                                     src/random/random_match.cpp)
 set_target_properties("${TARGET_NAME}" PROPERTIES
@@ -141,7 +147,8 @@ if(CMAKE_CUDA_COMPILER)
     set_property(TARGET "${TARGET_NAME}" PROPERTY CUDA_RESOLVE_DEVICE_SYMBOLS ON)
     #set_property(TARGET "${TARGET_NAME}" PROPERTY CUDA_ARCHITECTURES OFF)
 endif()
-target_link_libraries("${TARGET_NAME}" PRIVATE pybind11::module Eigen3::Eigen ${HDF5_LIBRARIES} openblas lapack gfortran m dl)
+target_link_libraries("${TARGET_NAME}" PRIVATE pybind11::module Eigen3::Eigen
+    ${HDF5_LIBRARIES} ${BLAS_LIBRARIES} gfortran m dl)
 if(DEFINED ENV{CONDA_PREFIX} AND (NOT APPLE OR CMAKE_COMPILER_IS_GNUCC OR ENV{SKETCHLIB_INSTALL} EQUAL "conda"))
     target_link_libraries("${TARGET_NAME}" PRIVATE gomp z)
 else()

LICENSE_KMeansRex

@@ -0,0 +1,14 @@
+BSD 3-clause license for open-source software.
+
+Copyright (c) 2013-2015, Michael C. Hughes
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
+
+1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
+
+2. Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
+
+3. Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

azure-pipelines.yml

@@ -46,7 +46,7 @@ steps:
     export PATH=${CUDA_HOME}/bin${PATH:+:${PATH}}
     export LD_LIBRARY_PATH=${CUDA_HOME}/lib64${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}
     export SKETCHLIB_INSTALL=azure
-    python -m pip install --no-deps --ignore-installed .
+    python -m pip install --no-deps --ignore-installed . -vvv
   displayName: 'Build & install (setup.py)'
 
 - script: |

pp_sketch/__init__.py

@@ -3,4 +3,4 @@
 
 '''PopPUNK sketching functions'''
 
-__version__ = '2.0.1'
+__version__ = '2.1.0'

pp_sketch/__main__.py

@@ -166,7 +166,7 @@ def main():
         sequences = []
 
         if args['-l']:
-            with open(args['-l'], 'rU') as refFile:
+            with open(args['-l'], 'r') as refFile:
                 for refLine in refFile:
                     refFields = refLine.rstrip().split("\t")
                     names.append(refFields[0])

src/Makefile

@@ -11,7 +11,7 @@ endif
 
 UNAME_S := $(shell uname -s)
 LIBLOC = ${CONDA_PREFIX}
-LDLIBS = -lz -lhdf5_cpp -lhdf5 -lopenblas -llapack -lgomp
+LDLIBS = -lz -lhdf5_cpp -lhdf5 -lopenblas -lgomp
 ifeq ($(UNAME_S),Linux)
 	CXXFLAGS+= -m64
 	ifdef PROFILE
@@ -29,11 +29,11 @@ LDFLAGS+= -L$(LIBLOC)/lib
 CUDA_LDLIBS=-lcudadevrt -lcudart_static $(LDLIBS)
 
 CUDA_LDFLAGS =-L$(LIBLOC)/lib -L${CUDA_HOME}/targets/x86_64-linux/lib/stubs -L${CUDA_HOME}/targets/x86_64-linux/lib
-CUDAFLAGS +=-Xcompiler -fPIC -Xptxas -dlcm=ca --cudart static --relocatable-device-code=true -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75
+CUDAFLAGS +=-Xcompiler -fPIC -Xptxas -dlcm=ca --cudart static --relocatable-device-code=true --expt-relaxed-constexpr -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75
 ifdef GPU
 	CXXFLAGS += -DGPU_AVAILABLE
 	CUDAFLAGS += -gencode arch=compute_86,code=sm_86
-	CUDA_LDFLAGS += -L/usr/local/cuda-11.1/lib64
+	CUDA_LDFLAGS += -L/usr/local/cuda-11.2/lib64
 endif
 
 PYTHON_LIB = pp_sketchlib$(shell python3-config --extension-suffix)
@@ -43,7 +43,7 @@ python: CPPFLAGS += -DGPU_AVAILABLE -DPYTHON_EXT -DNDEBUG -Dpp_sketchlib_EXPORTS
 
 PROGRAMS=sketch_test matrix_test read_test gpu_dist_test
 
-SKETCH_OBJS=dist/dist.o dist/matrix_ops.o reference.o sketch/seqio.o sketch/sketch.o database/database.o sketch/countmin.o api.o dist/linear_regression.o random/rng.o random/random_match.o
+SKETCH_OBJS=dist/dist.o dist/matrix_ops.o reference.o sketch/seqio.o sketch/sketch.o database/database.o sketch/countmin.o api.o dist/linear_regression.o random/rng.o random/random_match.o random/kmeans/KMeansRexCore.o random/kmeans/mersenneTwister2002.o
 GPU_SKETCH_OBJS=gpu/gpu_api.o
 CUDA_OBJS=gpu/dist.cu.o gpu/sketch.cu.o gpu/device_reads.cu.o gpu/gpu_countmin.cu.o gpu/device_memory.cu.o
 

src/api.cpp

@@ -193,7 +193,7 @@ NumpyMatrix query_db(std::vector<Reference> &ref_sketches,
   if (ref_sketches == query_sketches) {
     // calculate dists
     distMat.resize(dist_rows, dist_cols);
-    arma::mat kmer_mat = kmer2mat<std::vector<size_t>>(kmer_lengths);
+    Eigen::MatrixXf kmer_mat = kmer2mat(kmer_lengths);
 
     // Iterate upper triangle
 #pragma omp parallel for schedule(dynamic, 5) num_threads(num_threads) shared(progress)
@@ -230,7 +230,7 @@ NumpyMatrix query_db(std::vector<Reference> &ref_sketches,
     distMat.resize(dist_rows, dist_cols);
 
     // Prepare objects used in distance calculations
-    arma::mat kmer_mat = kmer2mat<std::vector<size_t>>(kmer_lengths);
+    Eigen::MatrixXf kmer_mat = kmer2mat(kmer_lengths);
     std::vector<size_t> query_lengths(query_sketches.size());
     std::vector<uint16_t> query_random_idxs(query_sketches.size());
 
@@ -338,7 +338,7 @@ sparse_coo query_db_sparse(std::vector<Reference> &ref_sketches,
   ProgressMeter dist_progress(n_progress_ticks, true);
   volatile int progress = 0;
 
-  arma::mat kmer_mat = kmer2mat<std::vector<size_t>>(kmer_lengths);
+  Eigen::MatrixXf kmer_mat = kmer2mat(kmer_lengths);
 #pragma omp parallel for schedule(static) num_threads(num_threads) shared(progress)
   for (size_t i = 0; i < ref_sketches.size(); i++) {
     std::vector<float> row_dists(ref_sketches.size());