improve python setup script

approxcdf/cpp_wrapper.pyx

@@ -1,3 +1,5 @@
+#cython: language_level=3
+
 ### BLAS and LAPACK
 from scipy.linalg.cython_blas cimport (
  ddot, daxpy, dgemv, dgemm, dtrmm, dtrsm

setup.py

@@ -5,6 +5,22 @@
 import sys, os, subprocess
 from os import environ
 
+found_omp = True
+def set_omp_false():
+ global found_omp
+ found_omp = False
+
+## Modify this to make the output of the compilation tests more verbose
+silent_tests = not (("verbose" in sys.argv)
+ or ("-verbose" in sys.argv)
+ or ("--verbose" in sys.argv))
+
+## Workaround for python<=3.9 on windows
+try:
+ EXIT_SUCCESS = os.EX_OK
+except AttributeError:
+ EXIT_SUCCESS = 0
+
 ## https://stackoverflow.com/questions/724664/python-distutils-how-to-get-a-compiler-that-is-going-to-be-used
 class build_ext_subclass( build_ext ):
  def build_extensions(self):
@@ -13,9 +29,9 @@ def build_extensions(self):
 
  if is_msvc:
  for e in self.extensions:
- e.extra_compile_args = ['/openmp', '/O2', '/std:c++14']
+ e.extra_compile_args = ['/openmp', '/O2', '/GL', '/std:c++14']
  else:
- if not self.check_for_variable_dont_set_march() and not self.is_arch_in_cflags():
+ if not self.check_for_variable_dont_set_march() and not self.check_cflags_contain_arch():
  self.add_march_native()
  self.set_cxxstd()
  self.add_openmp_linkage()
@@ -28,25 +44,37 @@ def build_extensions(self):
 
  build_ext.build_extensions(self)
 
+ def check_cflags_contain_arch(self):
+ if "CFLAGS" in os.environ:
+ arch_list = [
+ "-march", "-mcpu", "-mtune", "-msse", "-msse2", "-msse3",
+ "-mssse3", "-msse4", "-msse4a", "-msse4.1", "-msse4.2",
+ "-mavx", "-mavx2", "-mavx512"
+ ]
+ for flag in arch_list:
+ if flag in os.environ["CFLAGS"]:
+ return True
+ return False
+
  def check_for_variable_dont_set_march(self):
  return "DONT_SET_MARCH" in os.environ
 
  def add_march_native(self):
- arg_march_native = "-march=native"
- arg_mcpu_native = "-mcpu=native"
- if self.test_supports_compile_arg(arg_march_native):
- for e in self.extensions:
- e.extra_compile_args.append(arg_march_native)
- elif self.test_supports_compile_arg(arg_mcpu_native):
- for e in self.extensions:
- e.extra_compile_args.append(arg_mcpu_native)
+ args_march_native = ["-march=native", "-mcpu=native"]
+ for arg_march_native in args_march_native:
+ if self.test_supports_compile_arg(arg_march_native):
+ for e in self.extensions:
+ e.extra_compile_args.append(arg_march_native)
+ break
 
  def add_link_time_optimization(self):
- arg_lto = "-flto"
- if self.test_supports_compile_arg(arg_lto):
- for e in self.extensions:
- e.extra_compile_args.append(arg_lto)
- e.extra_link_args.append(arg_lto)
+ args_lto = ["-flto=auto", "-flto"]
+ for arg_lto in args_lto:
+ if self.test_supports_compile_arg(arg_lto):
+ for e in self.extensions:
+ e.extra_compile_args.append(arg_lto)
+ e.extra_link_args.append(arg_lto)
+ break
 
  def add_no_math_errno(self):
  arg_fnme = "-fno-math-errno"
@@ -60,6 +88,12 @@ def add_no_trapping_math(self):
  for e in self.extensions:
  e.extra_compile_args.append(arg_fntm)
 
+ def add_O3(self):
+ O3 = "-O3"
+ if self.test_supports_compile_arg(O3):
+ for e in self.extensions:
+ e.extra_compile_args.append(O3)
+
  def set_cxxstd(self):
  arg_std17 = "-std=c++17"
  arg_std14 = "-std=gnu++14"
@@ -75,45 +109,59 @@ def set_cxxstd(self):
 
  def add_openmp_linkage(self):
  arg_omp1 = "-fopenmp"
- arg_omp2 = "-qopenmp"
- arg_omp3 = "-xopenmp"
- arg_omp4 = "-fiopenmp"
+ arg_omp2 = "-fopenmp=libomp"
+ args_omp3 = ["-fopenmp=libomp", "-lomp"]
+ arg_omp4 = "-qopenmp"
+ arg_omp5 = "-xopenmp"
+ is_apple = sys.platform[:3].lower() == "dar"
  args_apple_omp = ["-Xclang", "-fopenmp", "-lomp"]
  args_apple_omp2 = ["-Xclang", "-fopenmp", "-L/usr/local/lib", "-lomp", "-I/usr/local/include"]
+ has_brew_omp = False
+ if is_apple:
+ res_brew_pref = subprocess.run(["brew", "--prefix", "libomp"], capture_output=silent_tests)
+ if res_brew_pref.returncode == EXIT_SUCCESS:
+ has_brew_omp = True
+ brew_omp_prefix = res_brew_pref.stdout.decode().strip()
+ args_apple_omp3 = ["-Xclang", "-fopenmp", f"-L{brew_omp_prefix}/lib", "-lomp", f"-I{brew_omp_prefix}/include"]
+
+
  if self.test_supports_compile_arg(arg_omp1, with_omp=True):
  for e in self.extensions:
  e.extra_compile_args.append(arg_omp1)
  e.extra_link_args.append(arg_omp1)
- elif (sys.platform[:3].lower() == "dar") and self.test_supports_compile_arg(args_apple_omp, with_omp=True):
+ elif is_apple and self.test_supports_compile_arg(args_apple_omp, with_omp=True):
  for e in self.extensions:
  e.extra_compile_args += ["-Xclang", "-fopenmp"]
  e.extra_link_args += ["-lomp"]
- elif (sys.platform[:3].lower() == "dar") and self.test_supports_compile_arg(args_apple_omp2, with_omp=True):
+ elif is_apple and self.test_supports_compile_arg(args_apple_omp2, with_omp=True):
  for e in self.extensions:
  e.extra_compile_args += ["-Xclang", "-fopenmp"]
  e.extra_link_args += ["-L/usr/local/lib", "-lomp"]
  e.include_dirs += ["/usr/local/include"]
+ elif is_apple and has_brew_omp and self.test_supports_compile_arg(args_apple_omp3, with_omp=True):
+ for e in self.extensions:
+ e.extra_compile_args += ["-Xclang", "-fopenmp"]
+ e.extra_link_args += [f"-L{brew_omp_prefix}/lib", "-lomp"]
+ e.include_dirs += [f"{brew_omp_prefix}/include"]
  elif self.test_supports_compile_arg(arg_omp2, with_omp=True):
  for e in self.extensions:
- e.extra_compile_args.append(arg_omp2)
- e.extra_link_args.append(arg_omp2)
- elif self.test_supports_compile_arg(arg_omp3, with_omp=True):
+ e.extra_compile_args += ["-fopenmp=libomp"]
+ e.extra_link_args += ["-fopenmp"]
+ elif self.test_supports_compile_arg(args_omp3, with_omp=True):
  for e in self.extensions:
- e.extra_compile_args.append(arg_omp3)
- e.extra_link_args.append(arg_omp3)
+ e.extra_compile_args += ["-fopenmp=libomp"]
+ e.extra_link_args += ["-fopenmp", "-lomp"]
  elif self.test_supports_compile_arg(arg_omp4, with_omp=True):
  for e in self.extensions:
  e.extra_compile_args.append(arg_omp4)
  e.extra_link_args.append(arg_omp4)
+ elif self.test_supports_compile_arg(arg_omp5, with_omp=True):
+ for e in self.extensions:
+ e.extra_compile_args.append(arg_omp5)
+ e.extra_link_args.append(arg_omp5)
  else:
  set_omp_false()
 
- def add_O3(self):
- O3 = "-O3"
- if self.test_supports_compile_arg(O3):
- for e in self.extensions:
- e.extra_compile_args.append(O3)
-
  def test_supports_compile_arg(self, comm, with_omp=False):
  is_supported = False
  try:
@@ -132,13 +180,12 @@ def test_supports_compile_arg(self, comm, with_omp=False):
  cmd = self.compiler.compiler_cxx
  except Exception:
  cmd = self.compiler.compiler_cxx
- val_good = subprocess.call(cmd + [fname])
  if with_omp:
  with open(fname, "w") as ftest:
  ftest.write(u"#include <omp.h>\nint main(int argc, char**argv) {return 0;}\n")
  try:
- val = subprocess.call(cmd + comm + [fname])
- is_supported = (val == val_good)
+ val = subprocess.run(cmd + comm + [fname], capture_output=silent_tests).returncode
+ is_supported = (val == EXIT_SUCCESS)
  except Exception:
  is_supported = False
  except Exception:
@@ -149,16 +196,6 @@ def test_supports_compile_arg(self, comm, with_omp=False):
  pass
  return is_supported
 
- def is_arch_in_cflags(self):
- arch_flags = '-march -mtune -msse -msse2 -msse3 -mssse3 -msse4 -msse4a -msse4.1 -msse4.2 -mavx -mavx2 -mcpu'.split()
- for env_var in ("CFLAGS", "CXXFLAGS"):
- if env_var in os.environ:
- for flag in arch_flags:
- if flag in os.environ[env_var]:
- return True
-
- return False
-
  def add_restrict_qualifier(self):
  supports_restrict = False
  try:
@@ -175,12 +212,11 @@ def add_restrict_qualifier(self):
  cmd = self.compiler.compiler_cxx
  except Exception:
  cmd = self.compiler.compiler_cxx
- val_good = subprocess.call(cmd + [fname])
  try:
  with open(fname, "w") as ftest:
  ftest.write(u"int main(int argc, char**argv) {double *__restrict x = 0; return 0;}\n")
- val = subprocess.call(cmd + [fname])
- supports_restrict = (val == val_good)
+ val = subprocess.run(cmd + comm + [fname], capture_output=silent_tests).returncode
+ is_supported = (val == EXIT_SUCCESS)
  except Exception:
  return None
  except Exception:
@@ -200,7 +236,6 @@ def add_restrict_qualifier(self):
  version = '0.0.1',
  description = 'Approximations for fast CDF calculation of MVN distributions',
  author = 'David Cortes',
- author_email = '[email protected]',
  url = 'https://github.com/david-cortes/approxcdf',
  keywords = ['cdf', 'tvbs', 'multivariate-normal', 'mvn'],
  cmdclass = {'build_ext': build_ext_subclass},
@@ -224,3 +259,14 @@ def add_restrict_qualifier(self):
  define_macros = [("FOR_PYTHON", None)]
  )]
 )
+
+if not found_omp:
+ omp_msg = "\n\n\nCould not detect OpenMP. Package will be built without multi-threading capabilities. "
+ omp_msg += " To enable multi-threading, first install OpenMP"
+ if (sys.platform[:3] == "dar"):
+ omp_msg += " - for macOS: 'brew install libomp'\n"
+ else:
+ omp_msg += " modules for your compiler. "
+
+ omp_msg += "Then reinstall this package from scratch: 'pip install --upgrade --no-deps --force-reinstall git+https://www.github.com/david-cortes/approxcdf.git'.\n"
+ warnings.warn(omp_msg)

src/plackett.cpp

@@ -63,8 +63,6 @@
  case it ends up giving reasonable results with an error of around 1e-8,
  but still very slow. */
 
-const int four = 4;
-
 /* https://stackoverflow.com/questions/2937702/i-want-to-find-determinant-of-4x4-matrix-in-c-sharp */
 double determinant4by4tri(const double x_tri[6])
 {