Skip to content

4.4 Cooperative Groups ​

待翻译

本节已建立目录框架,正文尚未翻译。

4.4.1 Introduction ​

4.4.2 Cooperative Group Handle & Member Functions ​

4.4.3 Default Behavior / Groupless Execution ​

4.4.3.1 Create Implicit Group Handles As Early As Possible ​

4.4.3.2 Only Pass Group Handles by Reference ​

4.4.4 Creating Cooperative Groups ​

4.4.4.1 Avoiding Group Creation Hazards ​

4.4.5 Synchronization ​

4.4.5.1 Sync ​

4.4.5.2 Barriers ​

4.4.6 Collective Operations ​

4.4.6.1 Reduce ​

4.4.6.2 Scans ​

4.4.6.3 Invoke One ​

4.4.7 Asynchronous Data Movement ​

4.4.7.1 Memcpy Async Alignment Requirements ​

4.4.8 Large Scale Groups ​

4.4.8.1 When to use cudaLaunchCooperativeKernel ​