Similar articles

Abstract: As the hardware industry moves toward using specialized heterogeneous many-core processors to avoid the effects of the power wall, software developers are finding it hard to deal with the complexity of these systems. In this paper, we share our experience of developing a programming model and its supporting compiler and libraries for Matrix-3000, which is designed for next-generation exascale supercomputers but has a complex memory hierarchy and processor organization. To assist its software development, we have developed a software stack from scratch that includes a low-level programming interface and a high-level OpenCL compiler. Our low-level programming model offers native programming support for using the bare-metal accelerators of Matrix-3000, while the high-level model allows programmers to use the OpenCL programming standard. We detail our design choices and highlight the lessons learned from developing system software to enable the programming of bare-metal accelerators. Our programming models have been deployed in the production environment of an exascale prototype system.

以Matrix-3000为例研究面向裸金属加速器的异构多线程编程模型

[1]Alfieri RA, 1994. An efficient kernel-based implementation of POSIX threads. Proc USENIX Summer Technical Conf, p.59-72.

[2]Arevalo A, Matinata RM, Pandian M, et al., 2000. Programming the cell broadband engine examples and best practices. ACM Workshop. Available from https://www.autodesk.com/research/publications/programming-the-cell-broadband [Accessed on Aug. 25, 2022].

[3]Fang JB, Varbanescu AL, Sips H, 2011. A comprehensive performance comparison of CUDA and OpenCL. Int Conf on Parallel Processing, p.216-225.

[4]Fang JB, Huang C, Tang T, et al., 2020. Parallel programming models for heterogeneous many-cores: a comprehensive survey. CCF Trans High Perform Comput, 2(4):382-400.

[5]Jääskeläinen P, de la Lama CS, Schnetter E, et al., 2015. pocl: a performance-portable OpenCL implementation. Int J Parall Program, 43(5):752-785.

[6]Kudlur M, Mahlke S, 2008. Orchestrating the execution of stream programs on multicore platforms. Proc 29^th ACM SIGPLAN Conf on Programming Language Design and Implementation, p.114-124.

[7]Liao XK, Lu K, Yang CQ, et al., 2018. Moving from exascale to zettascale computing: challenges and techniques. Front Inform Technol Electron Eng, 19(10):1236-1244.

[8]Lu K, Wang YH, Guo Y, et al., 2022. MT-3000: a heterogeneous multi-zone processor for HPC. CCF Trans High Perform Comput, 4(2):150-164.

[9]Owens JD, Luebke D, Govindaraju N, et al., 2005. A survey of general-purpose computation on graphics hardware. Proc 26^th Annual Conf of the European Association for Computer Graphics, p.21-51.

[10]Owens JD, Houston M, Luebke D, et al., 2008. GPU computing. Proc IEEE, 96(5):879-899.

[11]Patterson D, 2018. 50 years of computer architecture: from the mainframe CPU to the domain-specific TPU and the open RISC-V instruction set. IEEE Int Solid-State Circuits Conf, p.27-31.

[12]Perez JM, Bellens P, Badia RM, et al., 2007. CellSs: making it easier to program the cell broadband engine processor. IBM J Res Dev, 51(5):593-604.

[13]Shen J, Fang JB, Sips H, et al., 2012. Performance gaps between OpenMP and OpenCL for multi-core CPUs. Proc 41^st Int Conf on Parallel Processing Workshops, p.116-125.

[14]Trott CR, Lebrun-Grandié D, Arndt D, et al., 2022. Kokkos 3: programming model extensions for the exascale era. IEEE Trans Parall Distrib Syst, 33(4):805-817.

[15]Zhai JD, Chen WG, 2018. A vision of post-exascale programming. Front Inform Technol Electron Eng, 19(10):1261-1266.

[16]Zhang P, Tang T, Fang J, et al., 2018. MOCL: an efficient OpenCL implementation for the Matrix-2000 architecture. Proc 15^th ACM Int Conf on Computing Frontiers, p.26-35.