From e9e66d4c463b2ff9a487e42b489ed6d9b53e9f54 Mon Sep 17 00:00:00 2001 From: Daniel Lustig Date: Fri, 20 Jan 2023 14:25:13 -0500 Subject: [PATCH] Update thread scope for synchronization domains CUDA 12.0 introduced the "Memory Synchronization Domains" feature: https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#memory-synchronization-domains As described in that CUDA documentation, this feature changes the definition of `thread_scope_device`. Update libcu++ to match that definition update. --- docs/extended_api/memory_model.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/extended_api/memory_model.md b/docs/extended_api/memory_model.md index e0e08a54c1..7443c03fe8 100644 --- a/docs/extended_api/memory_model.md +++ b/docs/extended_api/memory_model.md @@ -36,7 +36,7 @@ enum thread_scope { Each program thread is related to each other program thread by one or more thread scope relations: - Each thread in the system is related to each other thread in the system by the *system* thread scope: `thread_scope_system`. -- Each GPU thread is related to each other GPU thread in the same CUDA device by the *device* thread scope: `thread_scope_device`. +- Each GPU thread is related to each other GPU thread in the same CUDA device and within the same [memory synchronization domain](https://docs.nvidia.com/cuda/cuda-c-programming-guide/index.html#memory-synchronization-domains) by the *device* thread scope: `thread_scope_device`. - Each GPU thread is related to each other GPU thread in the same CUDA thread block by the *block* thread scope: `thread_scope_block`. - Each thread is related to itself by the `thread` thread scope: `thread_scope_thread`.