arXiv subjects

Ahmed J. Abdelmaksoud

Publications and source records attributed to Ahmed J. Abdelmaksoud.

4 recordsLinked to original sources

arXiv2026

D-Legion: A Scalable Many-Core Architecture for Accelerating Matrix Multiplication in Quantized LLMs

The performance gains obtained by large language models (LLMs) are closely linked to their substantial computational and memory requirements. Quantized LLMs offer significant advantages with extremely quantized models, motivating the development of specialized architectures to accelerate their workloads. This paper proposes D-Legion, a novel scalable many-core architecture, designed using many adaptive-precision systolic array cores, to accelerate matrix multiplication in quantized LLMs. The proposed architecture consists of a set of Legions where each Legion has a group of adaptive-precision systolic arrays. D-Legion supports multiple computation modes, including quantized sparse and dense matrix multiplications. The block structured sparsity is exploited within a fully-sparse, or partially-sparse windows. In addition, memory accesses of partial summations (psums) are spatially reduced through parallel accumulators. Furthermore, data reuse is maximized through optimized scheduling techniques by multicasting matrix tiles across the Legions. A comprehensive design space exploration is performed in terms of Legion/core granularity to determine the optimal Legion configuration. Moreover, D-Legion is evaluated on attention workloads from two BitNet models, delivering up to 8.2$\times$ lower latency, up to 3.8$\times$ higher memory savings, and up to 3$\times$ higher psum memory savings compared to state-of-the-art work. D-Legion, with eight Legions and 64 total cores, achieves a peak throughput of 135.68 TOPS at a frequency of 1 GHz. A scaled version of D-Legion, with 32 Legions, is compared to Google TPUv4i, achieving up to 2.5$\times$ lower total latency, up to 2.3$\times$ higher total throughput, and up to 2.7$\times$ higher total memory savings.

cs.AR↗

arXiv2026

ADiP: Adaptive-Precision Systolic Array for Matrix Multiplication Acceleration

Transformers are at the core of modern AI nowadays. They rely heavily on matrix multiplication and require efficient acceleration due to their substantial memory and computational requirements. Quantization plays a vital role in reducing memory usage, and can be exploited for computations by designing reconfigurable architectures that enhance matrix multiplication by dynamically adjusting the precision. This paper proposes ADiP, a novel adaptive-precision systolic array architecture designed for efficient matrix multiplication acceleration. The proposed architecture consists of $N$ $\times$ $N$ reconfigurable processing elements (PEs), along with shared shifters and accumulators. ADiP supports multiple computation modes, including symmetric single-matrix multiplication as well as asymmetric multi-matrix multiplication with a shared input matrix, thereby improving data reuse and PE utilization. By adapting to different precisions, ADiP achieves up to 4$\times$ higher throughput and up to 4$\times$ higher memory efficiency. Analytical models are developed for ADiP architecture, including latency and throughput for different architecture configurations. A comprehensive hardware design space exploration is demonstrated using commercial 22nm technology. Furthermore, ADiP is evaluated on different Transformer-based workloads from GPT-2 medium, BERT large, and BitNet-1.58B models, delivering total latency improvement up to 53.6%, and total energy improvement up to 24.4% for attention workloads in BitNet-1.58B model. At a 64$\times$64 size with reconfigurable 4,096 PEs, ADiP achieves a peak throughput of 8.192 TOPS, 16.384 TOPS, and 32.768 TOPS for 8bit$\times$8bit, 8bit$\times$4bit, and 8bit$\times$2bit operations, respectively.

cs.AR↗

arXiv2025

DiP: A Scalable, Energy-Efficient Systolic Array for Matrix Multiplication Acceleration

Transformers are gaining increasing attention across Natural Language Processing (NLP) application domains due to their outstanding accuracy. However, these data-intensive models add significant performance demands to the existing computing architectures. Systolic array architectures, adopted by commercial AI computing platforms like Google TPUs, offer energy-efficient data reuse but face throughput and energy penalties due to input-output synchronization via First-In-First-Out (FIFO) buffers. This paper proposes a novel scalable systolic array architecture featuring Diagonal-Input and Permutated weight stationary (DiP) dataflow for matrix multiplication acceleration. The proposed architecture eliminates the synchronization FIFOs required by state-of-the-art weight stationary systolic arrays. Beyond the area, power, and energy savings achieved by eliminating these FIFOs, DiP architecture maximizes the computational resource utilization, achieving up to 50\% throughput improvement over conventional weight stationary architectures. Analytical models are developed for both weight stationary and DiP architectures, including latency, throughput, time to full PEs utilization (TFPU), and FIFOs overhead. A comprehensive hardware design space exploration using 22nm commercial technology demonstrates DiP's scalability advantages, achieving up to a 2.02x improvement in energy efficiency per area. Furthermore, DiP outperforms TPU-like architectures on transformer workloads from widely-used models, delivering energy improvement up to 1.81x and latency improvement up to 1.49x. At a 64x64 size with 4096 PEs, DiP achieves a peak throughput of 8.192 TOPS with energy efficiency 9.548 TOPS/W.

cs.AR↗

arXiv2025

3D-TrIM: A Memory-Efficient Spatial Computing Architecture for Convolution Workloads

The Von Neumann bottleneck, which relates to the energy cost of moving data from memory to on-chip core and vice versa, is a serious challenge in state-of-the-art AI architectures, like Convolutional Neural Networks' (CNNs) accelerators. Systolic arrays exploit distributed processing elements that exchange data with each other, thus mitigating the memory cost. However, when involved in convolutions, data redundancy must be carefully managed to avoid significant memory access overhead. To overcome this problem, TrIM has been recently proposed. It features a systolic array based on an innovative dataflow, where input feature map (ifmap) activations are locally reused through a triangular movement. However, ifmaps still suffer from memory accesses overhead. This work proposes 3D-TrIM, an upgraded version of TrIM that addresses the memory access overhead through few extra shadow registers. In addition, due to a change in the architectural orientation, the local shift register buffers are now shared between different slices, thus improving area and energy efficiency. An architecture of 576 processing elements is implemented on commercial 22 nm technology and achieves an area efficiency of 4.47 TOPS/mm$^2$ and an energy efficiency of 4.54 TOPS/W. Finally, 3D-TrIM outperforms TrIM by up to $3.37\times$ in terms of operations per memory access considering CNN topologies like VGG-16 and AlexNet.

cs.AR↗