libcudacxx icon indicating copy to clipboard operation
libcudacxx copied to clipboard

[ARCHIVED] The C++ Standard Library for your entire system. See https://github.com/NVIDIA/cccl

Results 117 libcudacxx issues
Sort by recently updated
recently updated
newest added

The __PGIC__ macro still works for detecting NVC++, but it's deprecated and confusing because it doesn't match the current branding. In ``, we should use `__NVCOMPILER` instead.

enhancement
P1: should have
compiler: nvc++

There's a defect in `*_semaphore::try_acquire()`, where it can block indefinitely when it can't acquire because of a coding fail by yours truly. We should engage with the solution on Phabricator,...

According to the thread scope documentation: > A thread scope specifies the kind of threads that can synchronize with each other using a primitive such as an atomic or a...

Example (https://cuda.godbolt.org/z/W541eW): ```c++ #include #include #include #include #include using T = float4; __global__ void test_pipe_intr(T* input, T volatile* out) { __shared__ T smem[32]; __pipeline_memcpy_async(smem + threadIdx.x, input + threadIdx.x, sizeof(T));...

It can be beneficial to test the state of a barrier prior to needing the barrier to be resolved. This allows one to hide the latency of testing the barrier...

assert within device code is relatively expensive. While users can opt-in to disabling asserts by defining NDEBUG optimzied builds should not have to define this to get performant library code....

As a minimal example, save the following as `main.cpp` and compile with MSVC / `cl.exe`, e.g.: `cl main.cpp -I "C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.2\include"` ``` #include ``` The same code...

In our application related to 3D geometry, we use `near` and `far` names. Recently, we started using libcu++ and more specifically ``. However, this causes issues as there is an...

Hi! :wave: Is there any roadmap describing planned support for newer, smaller headers? I want to use [std::numbers](https://en.cppreference.com/w/cpp/header/numbers) in my CUDA application, but NVCC doesn't seem to support it ([godbolt](https://godbolt.org/z/37r7nd))....

Reference for how to set it up: https://docs.github.com/en/github/using-git/configuring-git-to-handle-line-endings This should help us avoid having to manually review changes for line ending differences.

enhancement