The Experts below are selected from a list of 195174 Experts worldwide ranked by ideXlab platform
Hiromi Nakai - One of the best experts on this subject based on the ideXlab platform.
-
new recurrence relations for the rapid evaluation of electron repulsion integrals based on the accompanying coordinate expansion formula
Journal of Chemical Physics, 2004Co-Authors: Masato Kobayashi, Hiromi NakaiAbstract:We present an algorithm for the rapid computation of electron repulsion integrals (ERIs) over Gaussian basis functions based on the accompanying coordinate expansion (ACE) formula. The present algorithm uses equations termed angular momentum reduced expressions and introduces two types of recurrence relations to ACE formulas. Numerical efficiencies are assessed for (p p|p p) and (sp sp|sp sp) ERIs by using the Floating-Point Operation count. The algorithm is suitable for calculating ERIs for the same exponents but different angular momentum functions, such as L shells and derivatives of ERIs. The present algorithm is also capable of calculating ERIs with highly contracted Gaussian basis functions.
Todd J Martinez - One of the best experts on this subject based on the ideXlab platform.
-
automated code engine for graphical processing units application to the effective core potential integrals and gradients
Journal of Chemical Theory and Computation, 2016Co-Authors: Chenchen Song, Leeping Wang, Todd J MartinezAbstract:We present an automated code engine (ACE) that automatically generates optimized kernels for computing integrals in electronic structure theory on a given graphical processing unit (GPU) computing platform. The code generator in ACE creates multiple code variants with different memory and floating point Operation trade-offs. A graph representation is created as the foundation of the code generation, which allows the code generator to be extended to various types of integrals. The code optimizer in ACE determines the optimal code variant and GPU configurations for a given GPU computing platform by scanning over all possible code candidates and then choosing the best-performing code candidate for each kernel. We apply ACE to the optimization of effective core potential integrals and gradients. It is observed that the best code candidate varies with differing angular momentum, floating point precision, and type of GPU being used, which shows that the ACE may be a powerful tool in adapting to fast evolving GP...
-
generating efficient quantum chemistry codes for novel architectures
Journal of Chemical Theory and Computation, 2013Co-Authors: Alexey V Titov, Ivan S Ufimtsev, Nathan Luehr, Todd J MartinezAbstract:We describe an extension of our graphics processing unit (GPU) electronic structure program TeraChem to include atom-centered Gaussian basis sets with d angular momentum functions. This was made possible by a "meta-programming" strategy that leverages computer algebra systems for the derivation of equations and their transformation to correct code. We generate a multitude of code fragments that are formally mathematically equivalent, but differ in their memory and Floating-Point Operation footprints. We then select between different code fragments using empirical testing to find the highest performing code variant. This leads to an optimal balance of Floating-Point Operations and memory bandwidth for a given target architecture without laborious manual tuning. We show that this approach is capable of similar performance compared to our hand-tuned GPU kernels for basis sets with s and p angular momenta. We also demonstrate that mixed precision schemes (using both single and double precision) remain stable and accurate for molecules with d functions. We provide benchmarks of the execution time of entire self-consistent field (SCF) calculations using our GPU code and compare to mature CPU based codes, showing the benefits of the GPU architecture for electronic structure theory with appropriately redesigned algorithms. We suggest that the meta-programming and empirical performance optimization approach may be important in future computational chemistry applications, especially in the face of quickly evolving computer architectures.
Masato Kobayashi - One of the best experts on this subject based on the ideXlab platform.
-
new recurrence relations for the rapid evaluation of electron repulsion integrals based on the accompanying coordinate expansion formula
Journal of Chemical Physics, 2004Co-Authors: Masato Kobayashi, Hiromi NakaiAbstract:We present an algorithm for the rapid computation of electron repulsion integrals (ERIs) over Gaussian basis functions based on the accompanying coordinate expansion (ACE) formula. The present algorithm uses equations termed angular momentum reduced expressions and introduces two types of recurrence relations to ACE formulas. Numerical efficiencies are assessed for (p p|p p) and (sp sp|sp sp) ERIs by using the Floating-Point Operation count. The algorithm is suitable for calculating ERIs for the same exponents but different angular momentum functions, such as L shells and derivatives of ERIs. The present algorithm is also capable of calculating ERIs with highly contracted Gaussian basis functions.
Wang Jack - One of the best experts on this subject based on the ideXlab platform.
-
A HIGH-PERFORMANCE CHIP WITH DATAFLOW ARCHITECTURE FOR ATTITUDE AND HEADING REFERENCE SYSTEM ACCELERATING
IEEE International Conference on Cyber Technology in Automation Control and Intelligent Systems (CYBER), 2015Co-Authors: Shi Guangyi, Zhang Hanhui, Wang Zhenyu, Han Peng, Yan Wei, Jin Yufeng, Wang JackAbstract:Based on parallel processing methodology, a new design of high-performance chip with data flow architecture for accelerating AHRS is presented in this paper. Based gradient descent algorithm, we divided the serial algorithm into four primary pipeline function blocks on chip: Communication Block, Quaternion Initialization Block, Register file Block and Quaternion Update Block Every block was constructed with several parallel pipeline channels and embedded a number of acceleration units. Acceleration units were the technical-designed cells working for special mathematical Operation, high speed reading and writing process, such as floating point Operation, multi-dimension vectors computation and high speed cache on chip, which is proven as a great assistance for the data flow process. Compared to the traditional method estimating by serial MCU architecture, this data flow architecture got nearly a 10 times resolution of accuracy. Meanwhile, the speed of estimation could be raised up to over 40 times.CPCI-S(ISTP)jack@mems.cc1735-173
-
A high-performance chip with dataflow architecture for attitude and heading reference system accelerating
5th Annual IEEE International Conference on Cyber Technology in Automation Control and Intelligent Systems IEEE-CYBER 2015, 2015Co-Authors: Shi Guangyi, Zhang Hanhui, Wang Zhenyu, Han Peng, Yan Wei, Jin Yufeng, Wang JackAbstract:Based on parallel processing methodology, a new design of high-performance chip with data flow architecture for accelerating AHRS is presented in this paper. Based gradient descent algorithm, we divided the serial algorithm into four primary pipeline function blocks on chip: Communication Block, Quaternion Initialization Block, Register file Block and Quaternion Update Block. Every block was constructed with several parallel pipeline channels and embedded a number of acceleration units. Acceleration units were the technical-designed cells working for special mathematical Operation, high speed reading and writing process, such as floating point Operation, multi-dimension vectors computation and high speed cache on chip, which is proven as a great assistance for the data flow process. Compared to the traditional method estimating by serial MCU architecture, this data flow architecture got nearly a 10 times resolution of accuracy. Meanwhile, the speed of estimation could be raised up to over 40 times. ? 2015 IEEE.EI1735-173
Michael Bussmann - One of the best experts on this subject based on the ideXlab platform.
-
tuning and optimization for a variety of many core architectures without changing a single line of implementation code using the alpaka library
IEEE International Conference on High Performance Computing Data and Analytics, 2017Co-Authors: Alexander Matthes, Erik Zenker, Benjamin Worpitz, Axel Huebl, R. Widera, Michael BussmannAbstract:We present an analysis on optimizing performance of a single C++11 source code using the Alpaka hardware abstraction library. For this we use the general matrix multiplication (GEMM) algorithm in order to show that compilers can optimize Alpaka code effectively when tuning key parameters of the algorithm. We do not intend to rival existing, highly optimized DGEMM versions, but merely choose this example to prove that Alpaka allows for platform-specific tuning with a single source code. In addition we analyze the optimization potential available with vendor-specific compilers when confronted with the heavily templated abstractions of Alpaka. We specifically test the code for bleeding edge architectures such as Nvidia’s Tesla P100, Intel’s Knights Landing (KNL) and Haswell architecture as well as IBM’s Power8 system. On some of these we are able to reach almost 50% of the peak floating point Operation performance using the aforementioned means. When adding compiler-specific Open image in new window we are able to reach 5 Open image in new window on a P100 and over 1 Open image in new window on a KNL system.
-
Tuning and Optimization for a Variety of Many-Core Architectures Without Changing a Single Line of Implementation Code Using the Alpaka Library
High Performance Computing, 2017Co-Authors: Alexander Matthes, Erik Zenker, Benjamin Worpitz, Axel Huebl, R. Widera, Michael BussmannAbstract:We present an analysis on optimizing performance of a single C++11 source code using the Alpaka hardware abstraction library. For this we use the general matrix multiplication (GEMM) algorithm in order to show that compilers can optimize Alpaka code effectively when tuning key parameters of the algorithm. We do not intend to rival existing, highly optimized DGEMM versions, but merely choose this example to prove that Alpaka allows for platform-specific tuning with a single source code. In addition we analyze the optimization potential available with vendor-specific compilers when confronted with the heavily templated abstractions of Alpaka. We specifically test the code for bleeding edge architectures such as Nvidia’s Tesla P100, Intel’s Knights Landing (KNL) and Haswell architecture as well as IBM’s Power8 system. On some of these we are able to reach almost 50% of the peak floating point Operation performance using the aforementioned means. When adding compiler-specific we are able to reach 5 on a P100 and over 1 on a KNL system.