Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userHigh Performance and Portable Convolution Operators for ARM-based Multicore Processors
74
0
0.0
(
0
)
Added by
Pablo San Juan Sebasti\\'an
Publication date
2020
fields
Informatics Engineering
and research's language is
English
Ask ChatGPT about the research
No Arabic abstract
The considerable impact of Convolutional Neural Networks on many Artificial Intelligence tasks has led to the development of various high performance algorithms for the convolution operator present in this type of networks. One of these approaches leverages the imcol transform followed by a general matrix multiplication (GEMM) in order to take advantage of the highly optimized realizations of the GEMM kernel in many linear algebra libraries. The main problems of this approach are 1) the large memory workspace required to host the intermediate matrices generated by the IM2COL transform; and 2) the time to perform the IM2COL transform, which is not negligible for complex neural networks. This paper presents a portable high performance convolution algorithm based on the BLIS realization of the GEMM kernel that avoids the use of the intermediate memory by taking advantage of the BLIS structure. In addition, the proposed algorithm eliminates the cost of the explicit IM2COL transform, while maintaining the portability and performance of the underlying realization of GEMM in BLIS.
rate research
Read More
We evaluate the performance of Devito, a domain specific language (DSL) for finite differences on Arm ThunderX2 processors. Experiments with two common seismic computational kernels demonstrate that Arm processors can deliver competitive performance compared to other Intel Xeon processors.
This paper presents, implements, and evaluates a power-regulation technique for multicore processors, based on an integral controller with adjustable gain. The gain is designed for wide stability margins, and computed in real time as part of the control law. The tracking performance of the control system is robust with respect to modeling uncertainties and computational errors in the loop. The main challenge of designing such a controller is that the power dissipation of program-workloads varies widely and often cannot be measured accurately; hence extant controllers are either ad hoc or based on a-priori modeling characterizations of the processor and workloads. Our approach is different. Leveraging the aforementioned robustness it uses a simple textbook modeling framework, and adjusts its parameters in real time by a system-identification module. In this it trades modeling precision for fast computations in the loop making it suitable for on-line implementation in commodity data-center processors. Consequently, the proposed controller is agnostic in the sense that it does not require any a-priori system characterizations. We present an implementation of the controller on Intels fourth-generation microarchitecture, Haswell, and test it on a number of industry benchmark programs which are used in scientific computing and datacenter applications. Results of these experiments are presented in detail exposing the practical challenges of implementing provably-convergent power regulation solutions in commodity multicore processors.
Hardware platforms in high performance computing are constantly getting more complex to handle even when considering multicore CPUs alone. Numerous features and configuration options in the hardware and the software environment that are relevant for performance are not even known to most application users or developers. Microbenchmarks, i.e., simple codes that fathom a particular aspect of the hardware, can help to shed light on such issues, but only if they are well understood and if the results can be reconciled with known facts or performance models. The insight gained from microbenchmarks may then be applied to real applications for performance analysis or optimization. In this paper we investigate two modern Intel x86 server CPU architectures in depth: Broadwell EP and Cascade Lake SP. We highlight relevant hardware configuration settings that can have a decisive impact on code performance and show how to properly measure on-chip and off-chip data transfer bandwidths. The new victim L3 cache of Cascade Lake and its advanced replacement policy receive due attention. Finally we use DGEMM, sparse matrix-vector multiplication, and the HPCG benchmark to make a connection to relevant application scenarios.
In this paper, we use multithreaded fast Fourier transforms provided in three highly optimized packages, FFTW-2.1.5, FFTW-3.3.7, and Intel MKL FFT, to present a novel model-based parallel computing technique as a very effective and portable method for optimization of scientific multithreaded routines for performance, especially in the current multicore era where the processors have abundant number of cores. We propose two optimization methods, PFFT-FPM and PFFT-FPM-PAD, based on this technique. They compute 2D-DFT of a complex signal matrix of size NxN using p abstract processors. Both algorithms take as inputs, discrete 3D functions of performance against problem size of the processors and output the transformed signal matrix. Based on our experiments on a modern Intel Haswell multicore server consisting of 36 physical cores, the average and maximum speedups observed for PFFT-FPM using FFTW-3.3.7 are 1.9x and 6.8x respectively and the average and maximum speedups observed using Intel MKL FFT are 1.3x and 2x respectively. The average and maximum speedups observed for PFFT-FPM-PAD using FFTW-3.3.7 are 2x and 9.4x respectively and the average and maximum speedups observed using Intel MKL FFT are 1.4x and 5.9x respectively.
The increasing attention on deep learning has tremendously spurred the design of intelligence processing hardware. The variety of emerging intelligence processors requires standard benchmarks for fair comparison and system optimization (in both software and hardware). However, existing benchmarks are unsuitable for benchmarking intelligence processors due to their non-diversity and nonrepresentativeness. Also, the lack of a standard benchmarking methodology further exacerbates this problem. In this paper, we propose BENCHIP, a benchmark suite and benchmarking methodology for intelligence processors. The benchmark suite in BENCHIP consists of two sets of benchmarks: microbenchmarks and macrobenchmarks. The microbenchmarks consist of single-layer networks. They are mainly designed for bottleneck analysis and system optimization. The macrobenchmarks contain state-of-the-art industrial networks, so as to offer a realistic comparison of different platforms. We also propose a standard benchmarking methodology built upon an industrial software stack and evaluation metrics that comprehensively reflect the various characteristics of the evaluated intelligence processors. BENCHIP is utilized for evaluating various hardware platforms, including CPUs, GPUs, and accelerators. BENCHIP will be open-sourced soon.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
