Search arXiv⌕ Search
Home/Archive/arXiv

arXiv · 2502.16823

A Review of Memory Wall for Neuromorphic Computing

Dexter Le·Baran Arig·Murat Isik·I. Can Dikmen·Teoman Karadag
View original source ↗Read source document ↗Export metadata ↓

Abstract

This paper reviews memory technologies used in Field-Programmable Gate Arrays (FPGAs) for neuromorphic computing, a brain-inspired approach transforming artificial intelligence with improved efficiency and performance. It focuses on the essential role of memory in FPGA-based neuromorphic systems, evaluating memory types such as Static Random-Access Memory (SRAM), Dynamic Random-Access Memory (DRAM), High-Bandwidth Memory (HBM), and emerging non-volatile memories like Resistive RAM (ReRAM) and Phase-Change Memory (PCM). These technologies are analyzed based on latency, bandwidth, power consumption, density, and scalability to assess their suitability for storing and processing neural network models and synaptic weights. The review provides a comparative analysis of their strengths and limitations, supported by case studies illustrating real-world implementations and performance outcomes. This review offers insights to guide researchers and practitioners in selecting and optimizing memory technologies, enhancing the performance and energy efficiency of FPGA-based neuromorphic platforms, and advancing applications in artificial intelligence.

Explore related subjects

cs.AR

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Dexter Le, Baran Arig, Murat Isik, I. Can Dikmen, Teoman Karadag. 2025-02-24. A Review of Memory Wall for Neuromorphic Computing. https://arxiv.org/abs/2502.16823

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

arXiv2026

Exploiting Decompression Latency for Covert Channels in Inter-Line-Compressed LLCs

The recently proposed XOR cache is an inter-line-compressed last-level cache (LLC) that leverages the data-inclusion relationship between the private caches and the LLC, compressing two cache lines into one by XORing them. The architecture relies on the cache coherence protocol for data decompression. In this paper, we demonstrate that this mechanism - specifically the latency asymmetry between a cache hit on an uncompressed vs. compressed line - introduces microarchitectural vulnerabilities. Based on this observation, we propose a covert channel attack targeting the XOR cache. A colluding sender controls the receiver's access latency by triggering decompression through targeted write requests to partner cache lines. By exploiting the data-dependent compression behavior of the XOR cache, the sender and receiver establish the channel using pre-agreed data values. The channel achieves higher bandwidth than the Prime+Probe baseline for two reasons: first, each bit is encoded in the compression state of an individual line rather than the occupancy of a cache set, so a single set carries multiple bits; second, each bit is resolved by manipulating coherence-protocol state rather than forcing shared-cache evictions, so it costs fewer LLC accesses and demand misses than Prime+Probe. Full-system simulations show a bandwidth of 2.9 Mbps at an observed 0.98% bit-error rate (BER) over 50,000 transmitted bits, 13.1 times the bandwidth of Prime+Probe under the same sub-1%-BER selection rule.

cs.AR↗
arXiv2026

Mamba-Family State-Space Model Kernels on a Programmable CGLA

Edge and embedded inference is constrained by power and data movement. Mamba-family state-space models replace attention with sequence-linear recurrence, but their inference path combines dense projections, short-reduction SSD kernels, and recurrent-state updates. This paper maps these kernel groups onto IMAX, a programmable CPU-Grounded Linear Array (CGLA), and measures them from kernel execution to token-level integration. Projection kernels match the long-reduction IMAX pipeline, whereas SSD Step-1 is limited by short reductions and kernel-boundary overheads. Mamba-130M token-level integration identifies projection GEMV as the decode bottleneck. These results show that programmable CGLAs fit long-reduction projection kernels, while SSD and decode-time projection support require boundary reduction and persistent-weight execution.

cs.AR↗
arXiv2026

Energy-Oriented CGLA Mapping of a Memory-Polynomial Digital Predistortion Kernel

Memory-polynomial digital predistortion (DPD) evaluates a small fixed coefficient set over a sliding input history, so its reduction step is a complex-MAC workload with local reuse. We map this DPD reduction kernel onto In-Memory Accelerator eXtension (IMAX), a programmable CPU-Grounded Linear Array (CGLA) composed of a one-dimensional processing-element/local-memory pipeline. For a (P,M)=(5,5) odd-order memory-polynomial instance, the mapping keeps the 120 B coefficient set in local memory, advances the five-tap history over 1024-sample tiles, and realizes the 15 order-delay terms as a 33-stage streaming complex-MAC reduction. The evaluation measures kernel latency and modeled energy. All measured paths use the same single-precision complex workload of 32 sequences, each with 2048 complex samples, across an IMAX FPGA prototype, a CUDA implementation on an RTX 4090 system, and an ARM-NEON implementation on Jetson AGX Orin. With this 1024-sample tile configuration, the IMAX FPGA prototype reports 20.201 ms end-to-end latency and 1.948 ms kernel-only latency. Using the previously reported 28 nm IMAX frequency and power model, the projected IMAX configuration gives 3.14 ms end-to-end latency and 0.34 ms kernel-only latency. The RTX 4090 baseline has the lowest end-to-end latency at 0.484 ms. Under model-based platform power accounting and the stated power assumptions, the projected IMAX configuration gives 169.1 times smaller modeled end-to-end energy per batch than the RTX 4090 baseline. This value uses platform power assumptions rather than workload-dependent runtime power or a direct silicon power measurement. A controlled synthetic PA-model validation checks that the same 15-term form improves test-set NMSE by 26.1 dB and ACLR by 26.0 dB. These results characterize the mapped memory-polynomial DPD reduction on IMAX for the evaluated tile configuration and power model.

cs.AR↗