Thread Groups

Thread Groups#

2026-10-01

3 min read time

Applies to Linux and Windows

This page documents the ThreadGroup concept, which defines the requirements for a type representing a group of cooperating threads in CUDA.

struct ThreadGroup {
  static constexpr cuda::thread_scope thread_scope;
  Integral size() const;
  Integral thread_rank() const;
  void sync() const;
};

The ThreadGroup concept defines the requirements of a type that represents a group of cooperating threads.

The CUDA Cooperative Groups Library provides a number of types that satisfy this concept.

Data Members#

thread_scope

The scope at which ThreadGroup::sync() synchronizes memory operations and thread execution.

Member Functions#

size

Returns the number of participating threads.

thread_rank

Returns a unique value for each participating thread (0 <= ThreadGroup::thread_rank() < ThreadGroup::size()).

sync

Synchronizes the participating threads.

Note

This concept is defined for documentation purposes but is not materialized in the library.

Example#

#include <cuda/atomic>
#include <cuda/std/cstddef>

struct single_thread_group {
  static constexpr cuda::thread_scope thread_scope = cuda::thread_scope::thread_scope_thread;
  cuda::std::size_t size() const { return 1; }
  cuda::std::size_t thread_rank() const { return 0; }
  void sync() const {}
};

See it on Godbolt

Contents