Addressing the Last Roadblock for Message Logging in HPC: Alleviating the Memory Requirement Using Dedicated Resources

The sub-pixel motion estimation (SME), together with the interpolation of reference frames, is a computationally extensive part of the H.264 encoder that increases the memory requirement 16-times for each reference frame. Due to the huge computational complexity and memory requirement of the H.264 SME, its hardware architecture design is an important issue especially in high resolution or low power applications. To solve the above difficulties, we propose several optimization techniques in both algorithm and architecture levels. In the algorithm level, we propose a parabolic based algorithm for SME with quarter-pixel accuracy which reduces the computational budget by 94.35% and the memory access requirement by 98.5% in comparison to the standard interpolate and search method. In addition, a fast version of the proposed algorithm is presented that reduces the computational budget 46.28% further while maintaining the video quality. In the architecture level, we propose a novel bit-serial architecture for our algorithm. Due to advantages of the bit-serial architecture, it has a low gate count, high speed operation frequency, low density interconnection, and a reduced number of I/O pins. Also, several optimization techniques including the sum of absolute differences truncation, source sharing exploiting and power saving techniques are applied to the proposed architecture which reduce power consumption and area. Our design can save between 57.71–90.01% of area cost and improves the macroblock (MB) processing speed between 1.7–8.44 times when compared to previous designs. Implementation results show that our design can support real time HD1080 format with 20.3 k gate counts at the operation frequency of 144.9 MHz.

Download Full-text

Transient elastodynamic analysis with a combination of convolution quadrature method and pseudo-initial condition method

Engineering Computations ◽

10.1108/ec-06-2018-0263 ◽

2018 ◽

Vol 36 (1) ◽

pp. 334-355

Author(s):

Yuan Li ◽

J. Zhang ◽

Yudong Zhong ◽

Xiaomin Shu ◽

Yunqiao Dong

Keyword(s):

Initial Conditions ◽

Initial Condition ◽

Quadrature Method ◽

Memory Requirement ◽

Content Type ◽

Convolution Quadrature ◽

Force Method ◽

Cpu Time ◽

Condition Method ◽

Convolution Quadrature Method

Purpose The Convolution Quadrature Method (CQM) has been widely applied to solve transient elastodynamic problems because of its stability and generality. However, the CQM suffers from the problems of huge memory requirement in case of direct implementation in time domain or CPU time in case of its reformulation in Laplace domain. The purpose of this paper is to combine the CQM with the pseudo-initial condition method (PICM) to achieve a good balance between memory requirement and CPU time. Design/methodology/approach The combined methods first subdivide the whole analysis into a few sub-analyses, which is dealt with the PICM, namely, the results obtained by previous sub-analysis are used as the initial conditions for the next sub-analysis. In each sub-analysis, the time interval is further discretized into a number of sub-steps and dealt with the CQM. For non-zero initial conditions, the pseudo-force method is used to transform them into equivalent body forces. The boundary face method is employed in the numerical implementation. Three examples are analyzed. Results are compared with analytical solutions or FEM results and the results of reformulated CQM. Findings Results demonstrate that the computation time and the storage requirement can be reduced significantly as compared to the CQM, by using the combined approach. Originality/value The combined methods can be successfully applied to the problems of long-time dynamic response, which requires a large amount of computer memory when CQM is applied, while preserving the CQM stability. If the number of time steps is high, then the accuracy of the proposed approach can be deteriorated because of the pseudo-force method.

Download Full-text