TEMPI: An Interposed MPI Library with a Canonical Representation of CUDA-aware Datatypes
TEMPI: An Interposed MPI Library with a Canonical Representation of CUDA-aware Datatypes
复制标题
TEMPI：具有 CUDA 感知数据类型规范表示的插入式 MPI 库
DOI：
10.1145/3431379.3460645
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Wen
中科院分区：
文献类型：
--
作者：
Carl Pearson;Kun Wu;I. Chung;Jinjun Xiong;Wen
MPI派生数据类型是一种抽象，它简化了MPI应用程序中非连续数据的处理。这些数据类型是在运行时从MPI标准中定义的基本命名类型递归地构造的。最近，支持cuda的MPI实现的开发和部署鼓励了分布式高性能MPI代码向使用gpu的过渡。这样的实现允许MPI函数直接在GPU缓冲区上操作，简化了GPU计算与MPI代码的集成。这项工作首先为嵌套跨行数据类型提出了一种新的数据类型处理策略，它在先前工作中的专门化或泛型处理之间找到了一个中间地带。这项工作还表明，非连续数据处理的性能特征可以用经验系统测量来建模，并用于透明地改善MPI_Send/Recv延迟。最后，尽管大量关注非连续GPU数据和cuda感知MPI实现，但良好的性能不能被视为理所当然。这项工作通过MPI中介程序库TEMPI展示了它的贡献。TEMPI可以与现有的MPI部署一起使用，而无需更改系统或应用程序。最后，与部署在领导级超级计算机上的MPI实现相比，本工作的插入库模型演示了高达242000x的MPI_Pack加速和高达59000x的MPI_Send加速。这在3072个进程的3D光晕交换中产生了超过917倍的加速。
MPI derived datatypes are an abstraction that simplifies handling of non-contiguous data in MPI applications. These datatypes are recursively constructed at runtime from primitive Named Types defined in the MPI standard. More recently, the development and deployment of CUDA-aware MPI implementations has encouraged the transition of distributed high-performance MPI codes to use GPUs. Such implementations allow MPI functions to directly operate on GPU buffers, easing integration of GPU compute into MPI codes. This work first presents a novel datatype handling strategy for nested strided datatypes, which finds a middle ground between the specialized or generic handling in prior work. This work also shows that the performance characteristics of non-contiguous data handling can be modeled with empirical system measurements, and used to transparently improve MPI_Send/Recv latency. Finally, despite substantial attention to non-contiguous GPU data and CUDA-aware MPI implementations, good performance cannot be taken for granted. This work demonstrates its contributions through an MPI interposer library, TEMPI. TEMPI can be used with existing MPI deployments without system or application changes. Ultimately, the interposed-library model of this work demonstrates MPI_Pack speedup of up to 242000x and MPI_Send speedup of up to 59000x compared to the MPI implementation deployed on a leadership-class supercomputer. This yields speedup of more than 917x in a 3D halo exchange with 3072 processes.
MVAPICH 项目：将研究转化为 HPC 社区的高性能 MPI 库
影响因子：
3.3
作者：
Panda, Dhabaleswar Kumar;Subramoni, Hari;Chu, Ching-Hsiang;Bayatpour, Mohammadreza
通讯作者：
Bayatpour, Mohammadreza
FALCON-X：现代 CPU 和 GPU 架构上的零拷贝 MPI 派生数据类型处理
影响因子：
3.8
作者：
Hashmi, Jahanzeb Maqbool;Chu, Ching-Hsiang;Chakraborty, Sourav;Bayatpour, Mohammadreza;Subramoni, Hari;Panda, Dhabaleswar K.
通讯作者：
Panda, Dhabaleswar K.
用于 GPU 集群上批量非连续数据传输的动态内核融合
DOI：
--
发表时间：
2020
期刊：
22nd IEEE International Conference on Cluster Computing (IEEE Cluster 2020
影响因子：
--
作者：
C.-H. Chu, K. S.
通讯作者：
C.-H. Chu, K. S.