公开(公告)号：US20160335143A1

公开(公告)日：2016-11-17

申请号：US14710879

申请日：2015-05-13

Applicant: Advanced Micro Devices, Inc.

Inventor： Rathijit Sen ,  Indrani Paul ,  Wei Huang

IPC: G06F9/54 ,  G06F9/445

CPC classification number: G06F9/545 ,  G06F9/44505 ,  Y02D10/43

Abstract: Disclosed is a method of determining concurrency factors for an application running on a parallel processor. Also disclosed is a system for implementing the method. In an embodiment, the method includes running at least a portion of the kernel as sequences of mini-kernels, each mini-kernel including a number of concurrently executing workgroups. The number of concurrently executing workgroups is defined as a concurrency factor of the mini-kernel. A performance measure is determined for each sequence of mini-kernels. From the sequences, a particular sequence is chosen that achieves a desired performance of the kernel, based on the performance measures. The kernel is executed with the particular sequence.

Abstract translation: 公开了一种确定并行处理器上运行的应用程序的并发因子的方法。 还公开了一种用于实现该方法的系统。 在一个实施例中，该方法包括将内核的至少一部分作为小型内核的序列运行，每个小型内核包括多个并发执行的工作组。 并发执行工作组的数量被定义为小型内核的并发因子。 针对每个小型内核序列确定性能指标。 从序列中，基于性能测量，选择实现内核所需性能的特定序列。 内核以特定顺序执行。

公开(公告)号：US09965343B2

公开(公告)日：2018-05-08

申请号：US14710879

申请日：2015-05-13

Applicant: Advanced Micro Devices, Inc.

Inventor： Rathijit Sen ,  Indrani Paul ,  Wei Huang

IPC: G06F9/54 ,  G06F9/445

CPC classification number: G06F9/545 ,  G06F9/44505 ,  Y02D10/43

Abstract: Disclosed is a method of determining concurrency factors for an application running on a parallel processor. Also disclosed is a system for implementing the method. In an embodiment, the method includes running at least a portion of the kernel as sequences of mini-kernels, each mini-kernel including a number of concurrently executing workgroups. The number of concurrently executing workgroups is defined as a concurrency factor of the mini-kernel. A performance measure is determined for each sequence of mini-kernels. From the sequences, a particular sequence is chosen that achieves a desired performance of the kernel, based on the performance measures. The kernel is executed with the particular sequence.