Related papers: Progress and status of APEmille
New concepts for next-generation wireless systems are being developed. It is expected that these 6G and beyond systems will incorporate more than only communication, but also sensing, positioning, (deep) edge computing, and other services.…
We adopt CUDA-capable Graphic Processing Units (GPUs) for Coulomb, Landau and maximally Abelian gauge fixing in 3+1 dimensional SU(3) lattice gauge field theories. The local overrelaxation algorithm is perfectly suited for highly parallel…
Pipeline parallelism is an essential technique in the training of large-scale Transformer models. However, it suffers from imbalanced memory consumption, leading to insufficient memory utilization. The BPipe technique was proposed to…
Large Language Models (LLMs) such as LLaMA and DeepSeek, are built on transformer architectures, which have become a standard model for achieving state-of-the-art performance in natural language processing tasks. Recently, there has been…
The new Cell processor represents a turning point for computing intensive applications. Here, I show that for molecular dynamics it is possible to reach an impressive sustained performance in excess of 30 Gflops with a peak of 45 Gflops for…
High Bandwidth Memory with Processing-in-Memory (HBM-PIM) offers an opportunity to reduce data movement by executing computation directly inside memory, but current commercial platforms expose limited instruction sets and require…
Recently DRAM-based PIMs (processing-in-memories) with unmodified cell arrays have demonstrated impressive performance for accelerating AI applications. However, due to the very restrictive hardware constraints, PIM remains an accelerator…
This study presents a novel computer architecture where a last level cache and a SIMD accelerator are replaced by an Associative Processor. Associative Processor combines data storage and data processing and provides parallel computational…
While Large Language Models (LLMs) have shown remarkable abilities, they are hindered by significant resource consumption and considerable latency due to autoregressive processing. In this study, we introduce Adaptive N-gram Parallel…
The state of the art in electromagnetic Finite Element Particle-in-Cell (EM-FEMPIC) has advanced significantly in the last few years. These have included understanding function spaces that must be used to represent sources and fields…
The Liquid Argon Time Projection Chamber (LArTPC) technology plays an essential role in many current and future neutrino experiments. Accurate and fast simulation is critical to developing efficient analysis algorithms and precise physics…
We propose LIGHTNE 2.0, a cost-effective, scalable, automated, and high-quality network embedding system that scales to graphs with hundreds of billions of edges on a single machine. In contrast to the mainstream belief that distributed…
We report on the development of a novel pixel charge readout system, Grid Activated Multi-scale pixel readout (GAMPix), which is under development for use in the GammaTPC gamma ray instrument concept. GammaTPC is being developed to optimize…
Hardware accelerators, in particular accelerators for tensor processing, have many potential application domains. However, they currently lack the software infrastructure to support the majority of domains outside of deep learning.…
The Biham-Middleton-Levine (BML) traffic model is a simple two-dimensional, discrete Cellular Automaton (CA) that has been used to study self-organization and phase transitions arising in traffic flows. From the computational point of view,…
QPACE is a novel parallel computer which has been developed to be primarily used for lattice QCD simulations. The compute power is provided by the IBM PowerXCell 8i processor, an enhanced version of the Cell processor that is used in the…
High-performance computing underpins modern artificial intelligence (AI), enabling foundation models, real-time inference and perception in autonomous systems, and data-intensive scientific simulations. Recent advances in quantization…
The evolution of distributed architectures and programming paradigms for performance-oriented program development, challenge the state-of-the-art technology for performance tools. The area of high performance computing is rapidly expanding…
Infinite projected entangled-pair states (iPEPS) have been introduced to accurately describe many-body wave functions on two-dimensional lattices. In this context, two aspects are crucial: the systematic improvement of the {\it Ansatz} by…
Large model inference is shifting from cloud to edge due to concerns about the privacy of user interaction data. However, edge devices often struggle with limited computing power, memory, and bandwidth, requiring collaboration across…