[BUG]: A few parallel algorithms have started hanging at runtime after commit 35df3a

[zkhatami]

Is this a duplicate?

• I confirmed there appear to be no duplicate issues for this bug and that I agree to the Code of Conduct

Type of Bug

Silent Failure

Component

Thrust

Describe the bug

A few parallel algorithms have started hanging at runtime after commit 35df3a9.
For example:

#include <execution>
#include <algorithm>
#include <ranges>

constexpr int N = 1000;

int main() {
  auto data = std::views::iota(0, N);
  auto result = std::find_if_not(std::execution::par, data.begin(), data.end(),
      [](int x) { return x < N/2; });
}

when compiling this test with:

nvc++ --c++20 -stdpar -Ofast find_if_not.cpp; ./a.out

runtime hangs here:

at /proj/nv/Linux_x86_64/270541-dev/compilers/include-stdpar/thrust/iterator/detail/zip_iterator_base.h:155 in thrust::THRUST_300000_SM_70_NVHPC_NS::detail::tuple_impl_specific::tuple_for_each_helper<thrust::THRUST_300000_SM_70_NVHPC_NS::detail::advance_iterator<long>, thrust::THRUST_300000_SM_70_NVHPC_NS::transform_iterator<cuda::std::__4::__not_fn_t<main::{lambda(int)#1}>, std::ranges::iota_view<int, int>::_Iterator, bool, bool>, thrust::THRUST_300000_SM_70_NVHPC_NS::counting_iterator<long, thrust::THRUST_300000_SM_70_NVHPC_NS::use_default, thrust::THRUST_300000_SM_70_NVHPC_NS::use_default, thrust::THRUST_300000_SM_70_NVHPC_NS::use_default> > inlined from iterator_facade.h:427

I suspect the bug is in the iterators that replaced thrust::cuda_cub iterators in that commit.

How to Reproduce

nvc++ --c++20 -stdpar -Ofast find_if_not.cpp; ./a.out

Expected behavior

To verify that the runtime hang is resolved.

Reproduction link

No response

Operating System

No response

nvidia-smi output

No response

NVCC version

No response