Search NASA⌕ Search

Engineering topics

Meng, Jintao

Publications and source records attributed to Meng, Jintao.

autoGEMM: Pushing the Limits of Irregular Matrix Multiplication on Arm Architectures

This paper presents an open-source library that pushes the limits of performance portability for irregular General Matrix Multiplication (GEMM) on the widely-used Arm architectures. Our library, autoGEMM, is designed to support a wide range of Arm processors: from edge devices to HPC-grade CPUs. autoGEMM generates optimized kernels for various hardware configurations by auto-combining fragments of autogenerated micro-kernels that employ hand-written optimizations to maximize computational efficiency. We optimize the kernel pipeline by tuning the register reuse and the data load/store overlapping. In addition, we use a dynamic tiling scheme to generate balanced tile shapes. Finally, we position autoGEMM on top of the TVM framework where our dynamic tiling scheme prunes the search space for TVM to identify the optimal combination of parameters for code optimization. Evaluations on five different classes of Arm chips demonstrate the advantages of autoGEMM. For small matrices, autoGEMM achieves 98% of peak and up to 2.0x speedup over state-of-the-art libraries such as LIBXSMM and LibShalom. For irregular matrices (i.e. tall skinny and long rectangles), autoGEMM is 1.3-2.0x faster than widely-used libraries such as OpenBLAS and Eigen. autoGEMM is publicly available at: https://github.com/wudu98/autoGEMM.

Wu, Du↗

PERKS: a Locality-Optimized Execution Model for Iterative Memory-bound GPU Applications

Iterative memory-bound solvers commonly occur in HPC codes. Typical GPU implementations have a loop on the host side that invokes the GPU kernel as much as time/algorithm steps there are. The termination of each kernel implicitly acts the barrier required after advancing the solution every time step. We propose an execution model for running memory-bound iterative GPU kernels: PERsistent KernelS (PERKS). In this model, the time loop is moved inside persistent kernel, and device-wide barriers are used for synchronization. We then reduce the traffic to device memory by caching subset of the output in each time step in the unused registers and shared memory. PERKS can be generalized to any iterative solver: they largely independent of the solver's implementation. We explain the design principle of PERKS and demonstrate effectiveness of PERKS for a wide range of iterative 2D/3D stencil benchmarks (geomean speedup of 2.12x for 2D stencils and 1.24x for 3D stencils over state-of-art libraries), and a Krylov subspace conjugate gradient solver (geomean speedup of 4.86x in smaller SpMV datasets from SuiteSparse and 1.43x in larger SpMV datasets over a state-of-art library). All PERKS-based implementations available at: https://github.com/neozhang307/PERKS.

Zhang, Lingqi↗

Revisiting Temporal Blocking Stencil Optimizations

Iterative stencils are used widely across the spectrum of High Performance Computing (HPC) applications. Many efforts have been put into optimizing stencil GPU kernels, given the prevalence of GPU-accelerated supercomputers. To improve the data locality, temporal blocking is an optimization that combines a batch of time steps to process them together. Under the observation that GPUs are evolving to resemble CPUs in some aspects, we revisit temporal blocking optimizations for GPUs. We explore how temporal blocking schemes can be adapted to the new features in the recent Nvidia GPUs, including large scratchpad memory, hardware prefetching, and device-wide synchronization. We propose a novel temporal blocking method, EBISU, which champions low device occupancy to drive aggressive deep temporal blocking on large tiles that are executed tile-by-tile. We compare EBISU with state-of-the-art temporal blocking libraries: STENCILGEN and AN5D. We also compare with state-of-the-art stencil auto-tuning tools that are equipped with temporal blocking optimizations: ARTEMIS and DRSTENCIL. Over a wide range of stencil benchmarks, EBISU achieves speedups up to 2.53x and a geometric mean speedup of 1.49x over the best state-of-the-art performance in each stencil benchmark.

Zhang, Lingqi↗

Automatic Generation of High-Performance Convolution Kernels on ARM CPUs for Deep Learning

In this work, we present FastConv, a template-based code auto-generation open source library that can automatically generate high-performance deep learning convolution kernels of arbitrary matrices/tensors shapes. FastConv is based on the Winograd algorithm, which is reportedly the highest performing algorithm for the time-consuming convolution layers of convolutional neural networks. ARM CPUs cover a wide range designs and specifications, from embedded devices to HPC-grade CPUs. The leads to the dilemma of how to consistently optimize Winograd-based convolution solvers for convolution layers of different shapes. FastConv addresses this problem by using templates to auto-generate multiple shapes of tuned kernels variants suitable for skinny tall matrices. As a performance portable library, FastConv transparently searches for the best combination of kernel shapes, cache tiles, scheduling of loop orders, packing strategies, access patterns, and online/offline computations. Auto-tuning is used to search the parameter configuration space for the best performance for a given target architecture and problem size. The experiments with layer-wise evaluation on the VGG--16 model confirms a 1.25x performance gains is got by tuning the Winograd library. Integrated comparison results shows 1.02x to 1.40x, 1.14x to 2.17x, and 1.22x and 2.48x speedup is achieved over NNPACK, Arm NN, and FeatherCNN on the Kunpeng 920 beside few cases. Furthermore, problem size performance portability experiments with various convolution shapes shows that FastConv achieves 1.2x to 1.7x speedup and 2x to 22x speedup over NNPACK and ARM NN inference engine using Winograd on Kunpeng 920 . CPU performance portability evaluation on the VGG--16 show an average speedup over NNPACK of 1.42x, 1.21x, 1.26x, 1.37x, 2.26x, and 11.02x is observed on Kunpeng 920, Snapdragon 835, 855, 888, Apple M1, and AWS Graviton2, respectively.

97 MATHEMATICS AND COMPUTING↗

Counting Kmers for Biological Sequences at Large Scale

Counting the abundance of all the distinct kmers in biological sequence data is a fundamental step in bioinformatics. These applications include de novo genome assembly, error correction, etc. With the development of sequencing technology, the sequence data in a single project can reach Petabyte-scale or Terabyte-scale nucleotides. Counting demand for the abundance of these sequencing data is beyond the memory and computing capacity of single computing node, and how to process it efficiently is a challenge on a high-performance computing cluster. As such, we propose SWAPCounter, a highly scalable distributed approach for kmer counting. This approach is embedded with an MPI streaming I/O module for loading huge data set at high speed, and a counting bloom filter module for both memory and communication efficiency. By overlapping all the counting steps, SWAPCounter achieves high scalability with high parallel efficiency. In this work, the experimental results indicate that SWAPCounter has competitive performance with two other tools on shared memory environment, KMC 2 , and MSPKmerCounter. Moreover, SWAPCounter also shows the highest scalability under strong scaling experiments. In our experiment on Cetus supercomputer, SWAPCounter scales to 32,768 cores with 79% parallel efficiency (using 2048 cores as baseline) when processing 4 TB sequence data of 1000 Genomes. The source code of SWAPCounter is publicly available at https://github.com/mengjintao/SWAPCounter.

59 BASIC BIOLOGICAL SCIENCES↗

SWAP-Assembler 2: Optimization of De Novo Genome Assembler at Large Scale

In this paper, we analyze and optimize the most time-consuming steps of the SWAP-Assembler, a parallel genome assembler, so that it can scale to a large number of cores for huge genomes with the size of sequencing data ranging from terabyes to petabytes. According to the performance analysis results, the most time-consuming steps are input parallelization, k-mer graph construction, and graph simplification (edge merging). For the input parallelization, the input data is divided into virtual fragments with nearly equal size, and the start position and end position of each fragment are automatically separated at the beginning of the reads. In k-mer graph construction, in order to improve the communication efficiency, the message size is kept constant between any two processes by proportionally increasing the number of nucleotides to the number of processes in the input parallelization step for each round. The memory usage is also decreased because only a small part of the input data is processed in each round. With graph simplification, the communication protocol reduces the number of communication loops from four to two loops and decreases the idle communication time. The optimized assembler is denoted as SWAP-Assembler 2 (SWAP2). In our experiments using a 1000 Genomes project dataset of 4 terabytes (the largest dataset ever used for assembling) on the supercomputer Mira, the results show that SWAP2 scales to 131,072 cores with an efficiency of 40%. We also compared our work with both the HipMER assembler and the SWAP-Assembler. On the Yanhuang dataset of 300 gigabytes, SWAP2 shows a 3X speedup and 4X better scalability compared with the HipMer assembler and is 45 times faster than the SWAP-Assembler. The SWAP2 software is available at https://sourceforge.net/projects/swapassembler.

Meng, Jintao↗

SWAP-Assembler: scalable and efficient genome assembly towards thousands of cores

Background: There is a widening gap between the throughput of massive parallel sequencing machines and the ability to analyze these sequencing data. Traditional assembly methods requiring long execution time and large amount of memory on a single workstation limit their use on these massive data. Results: This paper presents a highly scalable assembler named as SWAP-Assembler for processing massive sequencing data using thousands of cores, where SWAP is an acronym for Small World Asynchronous Parallel model. In the paper, a mathematical description of multi-step bi-directed graph (MSG) is provided to resolve the computational interdependence on merging edges, and a highly scalable computational framework for SWAP is developed to automatically preform the parallel computation of all operations. Graph cleaning and contig extension are also included for generating contigs with high quality. Experimental results show that SWAP-Assembler scales up to 2048 cores on Yanhuang dataset using only 26 minutes, which is better than several other parallel assemblers, such as ABySS, Ray, and PASHA. Results also show that SWAP-Assembler can generate high quality contigs with good N50 size and low error rate, especially it generated the longest N50 contig sizes for Fish and Yanhuang datasets. Conclusions: In this paper, we presented a highly scalable and efficient genome assembly software, SWAPAssembler. Compared with several other assemblers, it showed very good performance in terms of scalability and contig quality. This software is available at: https://sourceforge.net/projects/swapassembler

59 BASIC BIOLOGICAL SCIENCES↗