index range can be reversed now

.gitlab-ci-correaa.yml

@@ -814,31 +814,31 @@ cuda-12.6.3:
     - mkdir build && cd build
     - g++ --version
     - /usr/local/cuda/bin/nvcc --version
-    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=75 -DCMAKE_CUDA_HOST_COMPILER=g++
+    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=native -DCMAKE_CUDA_HOST_COMPILER=g++
     - cmake --build . --parallel 2 || cmake --build . --verbose
     - ctest || ctest --rerun-failed --output-on-failure
   needs: ["cuda"]
 
-cuda-12.9.0:
-  stage: build
-  allow_failure: false
-  image: nvcr.io/nvidia/cuda:12.9.0-devel-ubuntu24.04  # https://catalog.ngc.nvidia.com/orgs/nvidia/containers/cuda/tags
-  tags:
-    - non-shared
-    - nvidia-gpu
-  interruptible: true
-  script:
-    - nvidia-smi
-    - apt-get -qq update
-    - apt-get install --no-install-recommends -y cmake g++ wget pkg-config make libboost-serialization-dev libboost-timer-dev libblas-dev libfftw3-dev
-    - cmake --version
-    - mkdir build && cd build
-    - g++ --version
-    - /usr/local/cuda/bin/nvcc --version
-    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=75 -DCMAKE_CUDA_HOST_COMPILER=g++
-    - cmake --build . --verbose --parallel 2 || cmake --build . --verbose
-    - ctest || ctest --rerun-failed --output-on-failure
-  needs: ["g++", "clang++", "cuda"]
+# cuda-12.9.0:
+#   stage: build
+#   allow_failure: false
+#   image: nvcr.io/nvidia/cuda:12.9.0-devel-ubuntu24.04  # https://catalog.ngc.nvidia.com/orgs/nvidia/containers/cuda/tags
+#   tags:
+#     - non-shared
+#     - nvidia-gpu
+#   interruptible: true
+#   script:
+#     - nvidia-smi
+#     - apt-get -qq update
+#     - apt-get install --no-install-recommends -y cmake g++ wget pkg-config make libboost-serialization-dev libboost-timer-dev libblas-dev libfftw3-dev
+#     - cmake --version
+#     - mkdir build && cd build
+#     - g++ --version
+#     - /usr/local/cuda/bin/nvcc --version
+#     - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=native -DCMAKE_CUDA_HOST_COMPILER=g++
+#     - cmake --build . --verbose --parallel 2 || cmake --build . --verbose
+#     - ctest || ctest --rerun-failed --output-on-failure
+#   needs: ["g++", "clang++", "cuda"]
 
 cuda-13.0.1:
   stage: build
@@ -847,6 +847,7 @@ cuda-13.0.1:
   tags:
     - non-shared
     - nvidia-gpu
+    - arch75
   interruptible: true
   script:
     - nvidia-smi
@@ -856,7 +857,7 @@ cuda-13.0.1:
     - mkdir build && cd build
     - g++ --version
     - /usr/local/cuda/bin/nvcc --version
-    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_FLAGS="-ftemplate-backtrace-limit=0" -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=75 -DCMAKE_CUDA_HOST_COMPILER=g++
+    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_FLAGS="-ftemplate-backtrace-limit=0" -DENABLE_CUDA=1 -DCMAKE_CXX_STANDARD=20 -DCMAKE_CUDA_STANDARD=20 -DCMAKE_CUDA_ARCHITECTURES=native -DCMAKE_CUDA_HOST_COMPILER=g++
     - cmake --build . --verbose --parallel 2 || cmake --build . --verbose
     - ctest || ctest --rerun-failed --output-on-failure
   needs: ["g++", "clang++", "cuda"]
@@ -882,7 +883,7 @@ culang++-19 cuda-12.6.3 tidy:  # https://catalog.ngc.nvidia.com/orgs/nvidia/cont
     - mkdir build && cd build
     - clang++-19 --version
     - clang-tidy-19 --version
-    - cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_FLAGS="-Wno-unknown-cuda-version" -DCMAKE_CXX_CLANG_TIDY=clang-tidy-19 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=clang++-19 -DCMAKE_CXX_COMPILER=clang++-19 -DCMAKE_CUDA_ARCHITECTURES=75
+    - cmake .. -DCMAKE_BUILD_TYPE=Release -DCMAKE_CUDA_FLAGS="-Wno-unknown-cuda-version" -DCMAKE_CXX_CLANG_TIDY=clang-tidy-19 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=clang++-19 -DCMAKE_CXX_COMPILER=clang++-19 -DCMAKE_CUDA_ARCHITECTURES=native
     - cmake --build . --parallel 2 || cmake --build . --verbose
     - ctest -j 2 --output-on-failure
   needs: ["cuda", "clang++"]
@@ -1144,12 +1145,14 @@ qmcpack:
 
 qmcpack cuda-12.3.1:
   stage: test
-  image: nvcr.io/nvidia/cuda:12.3.1-devel-ubuntu22.04
+  image:
+    name: nvcr.io/nvidia/cuda:12.9.0-devel-ubuntu24.04  # nvcr.io/nvidia/cuda:12.3.1-devel-ubuntu24.04
   tags:
     - non-shared
     - nvidia-gpu
     - high-bandwidth
   interruptible: true
+  allow_failure: true
   before_script:
     - apt-get -qq update && apt-get -qq install --no-install-recommends -y ca-certificates cmake git libopenmpi-dev cmake g++ git gfortran libblas-dev libboost-serialization-dev libfftw3-dev libhdf5-dev liblapack-dev libopenmpi-dev make numdiff pkg-config python3 python3-h5py python3-numpy python3-mpi4py python3-scipy libxml2-dev
   script:
@@ -1163,7 +1166,7 @@ qmcpack cuda-12.3.1:
     - cd build
     - nvcc --version
     - __nvcc_device_query
-    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_C_COMPILER=mpicc -DCMAKE_CXX_COMPILER=mpicxx -DBUILD_AFQMC=1 -DQMC_CXX_STANDARD=17 -DQMC_GPU=cuda -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_FLAGS="-Wno-deprecated -Wno-deprecated-declarations" -DCMAKE_CUDA_ARCHITECTURES=75  # =80
+    - CUDACXX=/usr/local/cuda/bin/nvcc cmake .. -DCMAKE_C_COMPILER=mpicc -DCMAKE_CXX_COMPILER=mpicxx -DBUILD_AFQMC=1 -DQMC_CXX_STANDARD=17 -DQMC_GPU=cuda -DCMAKE_CUDA_COMPILER=/usr/local/cuda/bin/nvcc -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_FLAGS="-Wno-deprecated -Wno-deprecated-declarations" -DCMAKE_CUDA_ARCHITECTURES=native  # =80
     - make -j 4 ppconvert afqmc test_afqmc_matrix test_afqmc_numerics test_afqmc_slaterdeterminantoperations test_afqmc_walkers test_afqmc_hamiltonians test_afqmc_hamiltonian_operations test_afqmc_phmsd test_afqmc_wfn_factory test_afqmc_prop_factory test_afqmc_estimators qmc-afqmc-performance
     - OMPI_ALLOW_RUN_AS_ROOT=1 OMPI_ALLOW_RUN_AS_ROOT_CONFIRM=1 ctest -R ppconvert --output-on-failure
     - OMPI_ALLOW_RUN_AS_ROOT=1 OMPI_ALLOW_RUN_AS_ROOT_CONFIRM=1 ctest -R afqmc     --output-on-failure

include/boost/multi/detail/index_range.hpp

@@ -22,6 +22,10 @@
 #include <type_traits>  // for declval, true_type, decay_t, enable_if_t
 #include <utility>      // for forward
 
+#if defined(__cplusplus) && (__cplusplus >= 202002L) && __has_include(<ranges>)
+#include <ranges>  // IWYU pragma: keep  // NOLINT(misc-include-cleaner)
+#endif
+
 #ifdef __NVCC__
 #define BOOST_MULTI_HD __host__ __device__
 #else
@@ -129,11 +133,14 @@ class range {
 	using pointer         = value_type;
 
 	range() = default;
-
 	// range(range const&) = default;
+	// range(range&&)      = default;
+	// auto operator=(range const&) -> range& = default;
+	// auto operator=(range&&) -> range& = default;
+	// ~range() = default;
 
 	template<class Range,
-	         std::enable_if_t<!std::is_base_of_v<range, std::decay_t<Range>>, int> = 0,  // NOLINT(modernize-type-traits) for C++20
+	         std::enable_if_t<!std::is_same_v<range, std::decay_t<Range>>, int> = 0,  // NOLINT(modernize-type-traits) for C++20
 	         decltype(detail::implicit_cast<IndexType>(std::declval<Range&&>().first()),
 	                  detail::implicit_cast<IndexTypeLast>(std::declval<Range&&>().last())
 	         )*                                                                    = nullptr>
@@ -143,7 +150,7 @@ class range {
 
 	template<
 		class Range,
-		std::enable_if_t<!std::is_base_of_v<range, std::decay_t<Range>>, unsigned> = 0,
+		std::enable_if_t<!std::is_same_v<range, std::decay_t<Range>>, unsigned> = 0,  // NOLINT(modernize-type-traits) for C++20
 		decltype(detail::explicit_cast<IndexType>(std::declval<Range&&>().first()),
 		         detail::explicit_cast<IndexTypeLast>(std::declval<Range&&>().last())
 		)*                                                                    = nullptr>
@@ -178,7 +185,7 @@ class range {
 			++curr_;
 			return *this;
 		}
-		constexpr auto operator--() -> const_iterator& {
+		constexpr auto operator--() noexcept(noexcept(--curr_)) -> const_iterator& {
 			--curr_;
 			return *this;
 		}
@@ -213,8 +220,8 @@ class range {
 
 	constexpr auto operator[](difference_type n) const -> const_reference { return first() + n; }
 
-	[[nodiscard]] BOOST_MULTI_HD constexpr auto front() const -> const_reference { return first(); }
-	[[nodiscard]] BOOST_MULTI_HD constexpr auto back() const -> const_reference { return last() - 1; }
+	[[nodiscard]] BOOST_MULTI_HD constexpr auto front() const -> value_type { return first(); }
+	[[nodiscard]] BOOST_MULTI_HD constexpr auto back() const -> value_type { return last() - 1; }
 
 	[[nodiscard]] constexpr auto cbegin() const { return const_iterator{first_}; }
 	[[nodiscard]] constexpr auto cend() const { return const_iterator{last_}; }
@@ -225,11 +232,11 @@ class range {
 	[[nodiscard]] constexpr auto begin() const -> const_iterator { return cbegin(); }
 	[[nodiscard]] constexpr auto end() const -> const_iterator { return cend(); }
 
-	BOOST_MULTI_HD constexpr auto        is_empty() const& noexcept { return first_ == last_; }
-	friend BOOST_MULTI_HD constexpr auto is_empty(range const& self) noexcept { return self.is_empty(); }
+	BOOST_MULTI_HD constexpr auto        is_empty() const noexcept { return first_ == last_; }
+	// friend BOOST_MULTI_HD constexpr auto is_empty(range const& self) noexcept { return self.is_empty(); }
 
 	[[nodiscard]] BOOST_MULTI_HD constexpr auto empty() const& noexcept { return is_empty(); }
-	friend BOOST_MULTI_HD constexpr auto        empty(range const& self) noexcept { return self.empty(); }
+	// friend BOOST_MULTI_HD constexpr auto        empty(range const& self) noexcept { return self.empty(); }
 
 	#ifdef __NVCC__
 	#pragma nv_diagnostic push
@@ -242,10 +249,10 @@ class range {
 	#pragma nv_diagnostic pop
 	#endif
 
-	friend BOOST_MULTI_HD constexpr auto size(range const& self) noexcept -> size_type { return self.size(); }
+	// friend BOOST_MULTI_HD constexpr auto size(range const& self) noexcept -> size_type { return self.size(); }
 
-	friend constexpr auto begin(range const& self) { return self.begin(); }
-	friend constexpr auto end(range const& self) { return self.end(); }
+	// friend constexpr auto begin(range const& self) { return self.begin(); }
+	// friend constexpr auto end(range const& self) { return self.end(); }
 
 	friend BOOST_MULTI_HD constexpr auto operator==(range const& self, range const& other) {
 		return (self.empty() && other.empty()) || (self.first_ == other.first_ && self.last_ == other.last_);
@@ -344,8 +351,16 @@ struct extension_t : public range<IndexType, IndexTypeLast> {
 	BOOST_MULTI_HD constexpr extension_t(IndexTypeLast last) noexcept  // NOLINT(google-explicit-constructor,hicpp-explicit-conversions) // NOSONAR(cpp:S1709) allow terse syntax
 	: range<IndexType, IndexTypeLast>(IndexType{}, IndexType{} + last) {}
 
+	// extension_t() = default;
+	// extension_t(extension_t const&) = default;
+	// extension_t(extension_t&&)      = default;
+	// auto operator=(extension_t const&) -> extension_t& = default;
+	// auto operator=(extension_t&&) -> extension_t& = default;
+	// ~extension_t() = default;
+
 	template<
 		class OtherExtension,
+		// std::enable_if_t<!std::is_same_v<extension_t, std::decay_t<OtherExtension>>, int> = 0,  // NOLINT(modernize-type-traits) bug in clang-tidy
 		decltype(
 			detail::implicit_cast<IndexType>(std::declval<OtherExtension>().first()),
 			detail::implicit_cast<IndexTypeLast>(std::declval<OtherExtension>().last())
@@ -365,13 +380,16 @@ struct extension_t : public range<IndexType, IndexTypeLast> {
 	// BOOST_MULTI_HD constexpr explicit extension_t(OtherExtension const& other) noexcept
 	// : extension_t{other.first(), other.last()} {}
 
-	template<class OtherExtension>
+	template<class OtherExtension//,
+		//std::enable_if_t<!std::is_base_of_v<extension_t, OtherExtension>, int> =0  // NOLINT(modernize-use-constraints,modernize-type-traits)
+	>
 	BOOST_MULTI_HD constexpr auto operator=(OtherExtension const& other) -> extension_t& {
 		(*this) = extension_t{other};
 		return *this;
 	}
 
-	BOOST_MULTI_HD constexpr extension_t() noexcept : range<IndexType, IndexTypeLast>() {}
+	extension_t() = default;
+	// BOOST_MULTI_HD constexpr extension_t() noexcept : range<IndexType, IndexTypeLast>() {}
 
 	// friend constexpr auto size(extension_t const& self) -> typename extension_t::size_type { return self.size(); }
 

include/boost/multi/detail/layout.hpp

@@ -502,7 +502,7 @@ struct extensions_t : boost::multi::detail::tuple_prepend_t<index_extension, typ
 				if(n > 0) {  // mull-ignore: cxx_gt_to_ge
 					curr_ += (rest_it_ - rest_begin_ + n) / (rest_end_ - rest_begin_);
 					rest_it_ = rest_begin_ + ((rest_it_ - rest_begin_ + n) % (rest_end_ - rest_begin_));
-				} else if(n < 0) {  // mull-ignore: cxx_lt_to_ge, cxx_lt_to_le
+				} else if(n < 0) {  // mull-ignore
 					curr_ -= (rest_end_ - rest_it_ - n) / (rest_end_ - rest_begin_);
 					rest_it_ = rest_end_ - ((rest_end_ - rest_it_ - n) % (rest_end_ - rest_begin_));
 					if(rest_it_ == rest_end_) {
@@ -521,7 +521,7 @@ struct extensions_t : boost::multi::detail::tuple_prepend_t<index_extension, typ
 						rest_it_ = rest_begin_;
 						++curr_;
 					}
-				} else if(n < 0) {  // mull-ignore: cxx_lt_to_ge, cxx_lt_to_le
+				} else if(n < 0) {  // mull-ignore
 					curr_ += (rest_it_ - rest_begin_ - n) / (rest_end_ - rest_begin_);
 					rest_it_ = rest_begin_ + ((rest_it_ - rest_begin_ - n) % (rest_end_ - rest_begin_));
 				}

pre-push

@@ -8,7 +8,7 @@
 #     curl https://developer.download.nvidia.com/hpc-sdk/ubuntu/DEB-GPG-KEY-NVIDIA-HPC-SDK | sudo gpg --dearmor -o /usr/share/keyrings/nvidia-hpcsdk-archive-keyring.gpg
 #     echo 'deb [signed-by=/usr/share/keyrings/nvidia-hpcsdk-archive-keyring.gpg] https://developer.download.nvidia.com/hpc-sdk/ubuntu/amd64 /' | sudo tee /etc/apt/sources.list.d/nvhpc.list
 #     sudo apt-get update -y
-#     sudo apt-get install -y nvhpc-24-7
+#     sudo apt-get install -y nvhpc
 #   on Fedora:
 #   sudo yum-config-manager --add-repo https://developer.download.nvidia.com/hpc-sdk/rhel/nvhpc.repo && sudo yum install -y nvhpc-cuda-multi-24.5
 # install oneAPI
@@ -47,11 +47,11 @@ export UBSAN_OPTIONS=print_stacktrack=1
 if [[ $(uname -m) != 'aarch64' ]]; then
  clang-format-19 --dry-run -Werror include/**/*.hpp test/*.cpp
 
- (CXX="$HOME/bin/circle" cmake -S . -B .build.circle     -DCMAKE_BUILD_TYPE=Debug   -DCMAKE_CXX_STANDARD=20 -DDISABLE_MPI=1 -DCMAKE_CXX_EXTENSIONS=OFF -DENABLE_CIRCLE=1 -DCMAKE_THREAD_LIBS_INIT="-lpthread"                                                                                          && cmake --build .build.circle     $CMT && (ctest --test-dir .build.circle --parallel 8 $CTR || ctest --test-dir .build.circle --rerun-failed --output-on-failure $CTR)) || exit 666
- (                       cmake -S . -B .build.nvcc.relax -DCMAKE_BUILD_TYPE=Debug   -DCMAKE_CXX_STANDARD=20 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=nvcc -DCMAKE_CUDA_FLAGS="--expt-relaxed-constexpr --threads 2" -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_COMPILER=g++ -DCMAKE_CUDA_ARCHITECTURES=75 && cmake --build .build.nvcc.relax $CMT && (ctest --test-dir .build.nvcc.relax   `#--parallel 8` $CTR || ctest --test-dir .build.nvcc.relax   --rerun-failed --output-on-failure $CTR)) || exit 666
- (                       cmake -S . -B .build.nvcc       -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=20 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=nvcc -DCMAKE_CUDA_FLAGS="--threads 2" -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_COMPILER=g++ -DCMAKE_CUDA_ARCHITECTURES=75                          && cmake --build .build.nvcc       $CMT && (ctest --test-dir .build.nvcc   --parallel 8 $CTR || ctest --test-dir .build.nvcc   --rerun-failed --output-on-failure $CTR)) || exit 666
+#(CXX="$HOME/bin/circle" cmake -S . -B .build.circle     -DCMAKE_BUILD_TYPE=Debug   -DCMAKE_CXX_STANDARD=20 -DDISABLE_MPI=1 -DCMAKE_CXX_EXTENSIONS=OFF -DENABLE_CIRCLE=1 -DCMAKE_THREAD_LIBS_INIT="-lpthread"                                                                                          && cmake --build .build.circle     $CMT && (ctest --test-dir .build.circle --parallel 8 $CTR || ctest --test-dir .build.circle --rerun-failed --output-on-failure $CTR)) || exit 666
+ (                       cmake -S . -B .build.nvcc.relax -DCMAKE_BUILD_TYPE=Debug   -DCMAKE_CXX_STANDARD=20 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=nvcc -DCMAKE_CUDA_FLAGS="--expt-relaxed-constexpr --threads 2" -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_COMPILER=g++ -DCMAKE_CUDA_ARCHITECTURE=native && cmake --build .build.nvcc.relax $CMT && (ctest --test-dir .build.nvcc.relax   `#--parallel 8` $CTR || ctest --test-dir .build.nvcc.relax   --rerun-failed --output-on-failure $CTR)) || exit 666
+ (                       cmake -S . -B .build.nvcc       -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=20 -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=nvcc -DCMAKE_CUDA_FLAGS="--threads 2" -DCMAKE_CUDA_HOST_COMPILER=g++ -DCMAKE_CXX_COMPILER=g++ -DCMAKE_CUDA_ARCHITECTURES=native                          && cmake --build .build.nvcc       $CMT && (ctest --test-dir .build.nvcc   --parallel 8 $CTR || ctest --test-dir .build.nvcc   --rerun-failed --output-on-failure $CTR)) || exit 666
 #(source /opt/intel/oneapi/setvars.sh && CXX=/opt/intel/oneapi/compiler/latest/bin/icpx cmake -S . -B .build.icpx  -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=20                                                                                                                                && cmake --build .build.icpx       $CMT && (ctest --test-dir .build.icpx   --parallel 8 $CTR || ctest --test-dir .build.icpx   --rerun-failed --output-on-failure $CTR)) || exit 666
- (CXX=/opt/nvidia/hpc_sdk/Linux_x86_64/2024/compilers/bin/nvc++ cmake -S . -B .build.nvc++ -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=17 -DBLA_PREFER_PKGCONFIG=ON -DCMAKE_CXX_COMPILER=/opt/nvidia/hpc_sdk/Linux_x86_64/2024/compilers/bin/nvc++                                               && cmake --build .build.nvc++  $CMT && (ctest --test-dir .build.nvc++  --parallel 8 $CTR || ctest --test-dir .build.nvc++  --rerun-failed --output-on-failure $CTR)) || exit 666
+ (CXX=/opt/nvidia/hpc_sdk/Linux_x86_64/2025/compilers/bin/nvc++ cmake -S . -B .build.nvc++ -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_STANDARD=17 -DBLA_PREFER_PKGCONFIG=ON -DCMAKE_CXX_COMPILER=/opt/nvidia/hpc_sdk/Linux_x86_64/2025/compilers/bin/nvc++                                               && cmake --build .build.nvc++  $CMT && (ctest --test-dir .build.nvc++  --parallel 8 $CTR || ctest --test-dir .build.nvc++  --rerun-failed --output-on-failure $CTR)) || exit 666
  (                       cmake -S . -B .build.culang       -DCMAKE_BUILD_TYPE=Release -DENABLE_CUDA=1 -DCMAKE_CUDA_COMPILER=clang++-19 -DCMAKE_CUDA_HOST_COMPILER=clang++-19 -DCMAKE_CXX_COMPILER=clang++-19                                                                    && cmake --build .build.culang $CMT && (ctest --test-dir .build.culang --parallel 8 $CTR || ctest --test-dir .build.culang --rerun-failed --output-on-failure $CTR)) || exit 666
 fi