China
Africa
Explore chapters and articles related to this topic
Vector and parallel processing of Monte Carlo particle transport
Published in Alireza Haghighat, Monte Carlo Methods for Particle Transport, 2020
Since the inception of computers, numerous architectures have been proposed and studied. These architectures can be classified into four groups based on the Flynn’s taxonomy [34], which considers the number of concurrent instruction and data streams: SISD (Single Instruction Single Data): A serial computer that does not have any parallelism in either instruction or data stream.SIMD (Single Instruction Multiple Data): A parallel computer that processes multiple data streams through a single instruction.MISD (Multiple Instruction Single Data): This taxonomy has not been considered!MIMD (Multiple Instruction Multiple Data): A computer environment that allows multiple instruction on multiple streams of data.
Distributed Systems
Published in Vivek Kale, Agile Network Businesses, 2017
They are briefly described as follows: Single instruction, single data stream (SISD): This is a sequential computer that exploits no parallelism, such as a PC (single core).Single instruction, multiple data stream (SIMD): This architecture supports multiple data streams to be processed simultaneously by replicating the computing hardware. Single Instruction means that all the data streams are processed using the same compute logic. Examples of parallel architectures that support this model are array processors or graphics processing unit (GPU).Multiple instruction, single data stream (MISD): This architecture operates on a single data stream but has multiple computing engines using the same data stream. This is not a very common architecture and is sometimes used to provide fault tolerance with heterogeneous systems operating on the same data to provide independent results that are compared with each other.Multiple instruction, multiple data stream (MIMD): This is the most generic parallel processing architecture where any type of distributed application can be programmed. Multiple autonomous processors executing in parallel work on independent streams of data. The application logic running on these processors can also be very different. All distributed systems are recognized to be MIMD architectures.
Towards Multicores: Technology and Software Complexity
Published in Marcello Coppola, Miltos D. Grammatikakis, Riccardo Locatelli, Giuseppe Maruccia, Lorenzo Pieralisi, Design of Cost-Efficient Interconnect Processing Units, 2020
Marcello Coppola, Miltos D. Grammatikakis, Riccardo Locatelli, Giuseppe Maruccia, Lorenzo Pieralisi
Single-instruction multiple-data (SIMD) enables a centralized data parallel approach, where all participating processing units execute the same instruction on multiple sets of data and synchronize before executing the next instruction. In addition, most SIMD (hybrid) architectures offer fast scalar instructions through deeply pipelined units. SIMD is frequently called single-program multiple-data (SPMD). In SPMD, some processors may be selected to abstain from a particular type of operation. SIMD architectures include versions of modern general-purpose processors, e.g. Intel’s Pentium and Motorola’s PowerPC, traditional parallel systems, such as the TMC CM-2, and DSP architectures.
FPGA-oriented lightweight multi-modal free-space detection network
Published in Connection Science, 2023
Feiyi Fang, Junzhu Mao, Wei Yu, Jianfeng Lu
The calculation of such a large model relies heavily on computational acceleration devices. Graphics Processing Units (GPUs) provide the infrastructure of multi-core parallel computing and has many cores, which can support parallel computing of large amounts of data. Its high floating-point computing capabilities are often used to accelerate CNN algorithms. For autonomous vehicles, both real-time processing speed and low power consumption are desirable. As GPU consumes much electricity, few GPU devices can be installed in an autonomous vehicle due to a limited power supply. But dozens of perception and planning tasks need to be processed simultaneously on the vehicle. Field Programmable Gate Arrays (FPGAs) are low-power devices with highly-parallel bit-oriented architecture, support both pipeline parallelism and data parallelism. The training of CNN models uses Single Instruction Multiple Data (SIMD) calculation, in which large batches of data can be processed in parallel by executing only one instruction. However, the calculation of model inference after the training is an application of Multiple Instruction Single Data (MISD), which is precisely the advantage of FPGA over GPU. We evaluate FPGAs as a platform to meet computational power requirements and constraints.
Design of a modern fast Fourier transform and cache effective bit-reversal algorithm
Published in International Journal of Parallel, Emergent and Distributed Systems, 2023
Single Instruction Multiple Data (SIMD) is a technology available on most nowadays computer architectures, which can process one instruction on multiple elements stored in the vector register unit in parallel. So-called butterfly is the innermost part of the FFT algorithm, gathering single elements with offset distance according to the actual stage and applying the convolution step.
Solving the Neutron Transport Equation for Microreactor Modeling Using Unstructured Meshes and Exascale Computing Architectures
Published in Nuclear Science and Engineering, 2023
William C. Dawn, Scott Palmtag
Graphics processing units are designed to efficiently perform single instruction, multiple data (SIMD) operations. An example of a SIMD operation would be adding two vectors together: the single instruction is the “add” operation and the multiple data are the two vectors. Many linear algebra operations can be reduced to a series of SIMD operations.