AUGEM: Automatically generate high performance dense linear algebra kernels on x86 CPUs

ISCAS OpenIR

	AUGEM: Automatically generate high performance dense linear algebra kernels on x86 CPUs
	Wang, Qian (1); Zhang, Xianyi (1); Zhang, Yunquan (2); Yi, Qing (3)
	2013
会议名称	2013 International Conference for High Performance Computing, Networking, Storage and Analysis, SC 2013
会议日期	November 17, 2013 - November 22, 2013
会议地点	Denver, CO, United states
收录类别	EI
出版地	IEEE Computer Society
ISSN	21674329
ISBN	9781450323789
部门归属	(1) Institute of Software, Chinese Academy of Sciences, University of Chinese, Beijing, China; (2) Institute of Software, Chinese Academy of Sciences, State Key Lab of Computer Architecture, Beijing, China; (3) University of Colorado at Colorado Springs, Colorado, United States
摘要	Basic Liner algebra subprograms (BLAS) is a fundamental library in scientific computing. In this paper, we present a template-based optimization framework, AUGEM, which can automatically generate fully optimized assembly code for several dense linear algebra (DLA) kernels, such as GEMM, GEMV, AXPY and DOT, on varying multi-core CPUs without requiring any manual interference from developers. In particular, based on domain-specific knowledge about algorithms of the DLA kernels, we use a collection of parameterized code templates to formulate a number of commonly occurring instruction sequences within the optimized low-level C code of these DLA kernels. Then, our framework uses a specialized low-level C optimizer to identify instruction sequences that match the pre-defined code templates and thereby translates them into extremely efficient SSE/AVX instructions. The DLA kernels generated by our templatebased approach surpass the implementations of Intel MKL and AMD ACML BLAS libraries, on both Intel Sandy Bridge and AMD Piledriver processors. Copyright 2013 ACM.; Basic Liner algebra subprograms (BLAS) is a fundamental library in scientific computing. In this paper, we present a template-based optimization framework, AUGEM, which can automatically generate fully optimized assembly code for several dense linear algebra (DLA) kernels, such as GEMM, GEMV, AXPY and DOT, on varying multi-core CPUs without requiring any manual interference from developers. In particular, based on domain-specific knowledge about algorithms of the DLA kernels, we use a collection of parameterized code templates to formulate a number of commonly occurring instruction sequences within the optimized low-level C code of these DLA kernels. Then, our framework uses a specialized low-level C optimizer to identify instruction sequences that match the pre-defined code templates and thereby translates them into extremely efficient SSE/AVX instructions. The DLA kernels generated by our templatebased approach surpass the implementations of Intel MKL and AMD ACML BLAS libraries, on both Intel Sandy Bridge and AMD Piledriver processors. Copyright 2013 ACM.
语种	英语
内容类型	会议论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16662
专题	中国科学院软件研究所
推荐引用方式 GB/T 7714	Wang, Qian ,Zhang, Xianyi ,Zhang, Yunquan ,et al. AUGEM: Automatically generate high performance dense linear algebra kernels on x86 CPUs[C]. IEEE Computer Society,2013.