accelerator

摘要： Graph convolutional networks(GCNs) have been applied successfully in social networks and recommendation systems to analyze graph data. Unlike conventional neural networks, GCNs introduce an aggregation phase, which is both computation-and memory-intensive. This phase aggregates features from the neighboring vertices in the graph, which incurs significant amounts of irregular data and memory access.The emerging computation-in-memory(CIM) architecture presents a promising solution to alleviate the problem of irregular accesses and provide fast near-data processing for GCN applications by integrating both three-dimensional stacked CIM and general-purpose processing units in the system. This paper presents Graph-CIM, which exploits the hybrid CIM architecture to determine the allocation of GCN applications.Graph-CIM models the GCN application process as a directed acyclic graph(DAG) and allocates tasks on the hybrid CIM architecture. It achieves fine-grained graph partitioning to capture the irregular characteristics of the aggregation phase of GCN applications. We use a set of representative GCN models and standard graph datasets to evaluate the effectiveness of Graph-CIM. The experimental results show that Graph-CIM can significantly reduce the processing latency and data-movement overhead compared with the representative schemes.

摘要： The assessment of activated concrete is particularly difficult during the decommissioning of an accelerator facility.Destructive analysis by core boring is the only method of investigating the activity of concrete material.To address this problem,an in-situ and nondestructive analysis method was developed to determineγ-ray-emitting nuclides and their specific activities in the concrete walls and floor by using a portable germanium semiconductor detector.In this work,we examined a substitute for Ge detector to establish a simpler and more convenient method.As candidates,we focused on some scintillation type spectrometers,and the possibility of a substitute for a Ge detector was examined by both simulation and experiment.The detection limits were roughly estimated through Monte Carlo simulation for various scintillation crystals,and it was found that 1.5-inch LaBr3,CeBr3,and SrI2 could distinguish the clearance level.It was confirmed that the 1.5-inch LaBr3 could reproduce the calibration curve of the Ge detector in the experiment.The required thickness and length of the radiation shield for suppressing the background radiation during the measurement was also determined for the convenience of an actual decommissioning work.

摘要： This study established a method for easily and quickly estimating the specific activity produced in the concrete walls and floors of accelerator rooms during long-term operation of accelerator,for advanced zoning of activated/nonactivated areas in planning the decommissioning of an accelerator.We propose a new,highly sensitive method for nondestructively estimating the specific activity in concrete that can be applied to activation zoning.In this method,instead of direct determination of the specific activities of important long-half-life radionuclides for decommissioning,such as 152 Eu and 60 Co,we determine the specific activities of short-half-life radionuclides,24 Na and 56 Mn,in situ to obtain neutron flux.The obtained neutron flux and accelerator operation history yield the specific activities of 152 Eu and 60 Co for the advance zoning of activated/non-activated concrete.This method is a powerful long-term prediction tool for concrete activation.

摘要： This is the second paper by the author describing versatile accelerator complexes that could be built at a Future Circular Collider (FCC) in order to produce e+e-, γγ and ep collisions. The facility described here features an ILC-based e+e- collider placed tangentially to the FCC tunnel. If the collider is positioned asymmetrically with respect to the FCC tunnel, electron (or positron) bunches could be accelerated by both linacs before they are brought into collision with the 50-TeV beams from the FCC proton storage ring (FCC-pp). The two linacs may also form a part of the injector chain for FCC-pp. The facility could be converted into a γγ collider or a source of multi-MW beams for fixed-target experiments.

摘要： Particle accelerators play an important role in a wide range of scientific discoveries and industrial applications. The self-consistent multi-particle simulation based on the particle-in-cell (PIC) method has been used to study charged particle beam dynamics inside those accelerators. However, the PIC simulation is time-consuming and needs to use modern parallel computers for high-resolution applications. In this paper, we implemented a parallel beam dynamics PIC code on multi-node hybrid architecture computers with multiple Graphics Processing Units (GPUs). We used two methods to parallelize the PIC code on multiple GPUs and observed that the replication method is a better choice for moderate problem size and current computer hardware while the domain decomposition method might be a better choice for large problem size and more advanced computer hardware that allows direct communications among multiple GPUs. Using the multi-node hybrid architectures at Oak Ridge Leadership Computing Facility (OLCF), the optimized GPU PIC code achieves a reasonable parallel performance and scales up to 64 GPUs with 16 million particles.

摘要： Innovative nuclear reactor concepts such as the Accelerator Driven Systems (ADSs) have imposed extra requirements of simulation capabilities on the existing stochastic neutronics codes. The combination of an accelerator and a nuclear reactor in the ADS requires the simulation of both subsystems for an integrated system analysis. Therefore, a need arises for more advanced simulation tools, able to cover the broad neutron energy spectrum involved in these systems. ANET (Advanced Neutronics with Evolution and Thermal hydraulic feedback) is an under development stochastic code for simulating conventional and hybrid nuclear reactors. Successive testing applications performed throughout the ANET development have been utilized to verify and validate the new code capabilities. In this context, the ANET reliability in simulating the spallation reaction and the corresponding neutron yield as well as computing the multiplication factor of an operating ADS are here examined. More specifically, three cores of the Kyoto University Critical Assembly (KUCA) facility in Japan were analyzed focusing on the spallation neutron yield and the neutron multiplication factor. The ANET-produced results are compared with independent results obtained using the stochastic codes MCNP6.1 and MCNPX. Satisfactory agreement is found between the codes, confirming thus ANET’s capability to successfully estimate both the neutron yield of the spallation reaction and the keff of a realistic ADS.

摘要： 意法半导体发布全新微控制器STM32H7。该新产品是业界性能最高的Arm CortexM通用MCU,集强劲的双核处理器和节能型功能以及强化的网络安保功能于一身。新产品采用Arm CortexM系列中性能最高的480 MHz CortexM7内核,并增加一颗240 MHz CortexM4内核。借助意法半导体的智能架构、高效的L1缓存和ART Accelerator自适应实时加速技术,当执行嵌入式闪存中的代码时,新MCU创下了1327 DMIPS和3224 CoreMark性能新记录。