Reuse Distance-based Copy-backs of Clean Cache Lines to Lower-level Caches

Rui Wang; Chundong Wang; Chongnan Ye

Home/Archive/arXiv

arXiv · 2105.14442

Reuse Distance-based Copy-backs of Clean Cache Lines to Lower-level Caches

Rui Wang·Chundong Wang·Chongnan Ye

View original source ↗Read source document ↗Export metadata ↓

Abstract

Cache plays a critical role in reducing the performance gap between CPU and main memory. A modern multi-core CPU generally employs a multi-level hierarchy of caches, through which the most recently and frequently used data are maintained in each core's local private caches while all cores share the last-level cache (LLC). For inclusive caches, clean cache lines replaced in higher-level caches are not necessarily copied back to lower levels, as the inclusiveness implies their existences in lower levels. For exclusive and non-inclusive caches that are widely utilized by Intel, AMD, and ARM today, either indiscriminately copying back all or none of replaced clean cache lines to lower levels raises no violation to exclusiveness and non-inclusiveness definitions. We have conducted a quantitative study and found that, copying back all or none of clean cache lines to lower-level cache of exclusive caches entails suboptimal performance. The reason is that only a part of cache lines would be reused and others turn to be dead in a long run. This observation motivates us to selectively copy back some clean cache lines to LLC in an architecture of exclusive or non-inclusive caches. We revisit the concept of reuse distance of cache lines. In a nutshell, a clean cache line with a shorter reuse distance is copied back to lower-level cache as it is likely to be re-referenced in the near future, while cache lines with much longer reuse distances would be discarded or sent to memory if they are dirty. We have implemented and evaluated our proposal with non-volatile (STT-MRAM) LLC. Experimental results with gem5 and SPEC CPU 2017 benchmarks show that on average our proposal yields up to 12.8% higher throughput of IPC (instructions per cycle) than the least-recently-used (LRU) replacement policy with copying back all clean cache lines for STT-MRAM LLC.

Explore related subjects

cs.AR

Keep this discovery

BibTeX RIS

Rui Wang, Chundong Wang, Chongnan Ye. 2021-05-30. Reuse Distance-based Copy-backs of Clean Cache Lines to Lower-level Caches. https://arxiv.org/abs/2105.14442

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

KEEP EXPLORING

Related papers

arXiv2026

Fengshui: Demystifying Chiplet Ecosystem and Bespoke Neural Network Accelerator Codesign

Modern ML workloads, with stringent latency and energy constraints, are increasingly hard to run efficiently on homogeneous commodity hardware. We argue that operator-level disaggregation--tailoring microarchitecture, batching, and memory hierarchy to each operator--is essential to overcome these limitations, though the resulting highly bespoke accelerators incur prohibitive Non-Recurring Engineering (NRE) costs. Chiplet-based integration amortizes NRE across applications, but choosing which chiplets to build and how to compose them into accelerators is circularly dependent--a chiplet pool's value depends on the constructed accelerators, while accelerator quality is constrained by available chiplets. This paper introduces Fengshui, a chiplet ecosystem and accelerator co-design framework that jointly optimizes chiplet pool composition and bespoke application-specific integrated circuit (BASIC) design. Fengshui constructs BASICs through operator-level disaggregation, co-exploring chiplet and memory heterogeneity, tensor fusion, and pipeline/tensor/expert parallelism with place-and-route validation for physical implementability. With just 8 strategically selected chiplets, encompassing network switches, processing-in-memory units, and accelerators with diverse microarchitectures, Fengshui-generated BASICs achieve 48.5%, 88.1%, 93.0%, and 97.8% reductions in energy, energy-cost product (EC), energy-delay product (EDP), and energy-delay-cost product (EDPC) over homogeneous accelerators, while scoring within 4.1% of unconstrained heterogeneous designs across diverse neural networks. For datacenter MoE and dense LLM serving, Fengshui reduces prefill energy and EC by up to 16.8% and 28.7%, respectively; for edge autonomous vehicle perception, it achieves 12.0% energy and 23.6% EC reductions under real-time latency constraints.

cs.AR↗

arXiv2026

BEACON: A Versatile Accelerator for Computational Pathology Applications

While accelerators for AI have seen great commercial success, it is challenging to replicate that success for other specialized domains due to a number of factors. We make the case that barriers for new accelerators can be lowered by starting with a baseline AI accelerator, and adding minimal logic to support new operators demanded by new specialized domains. This leads to a versatile chip that can be manufactured at high volume and deployed for a range of popular applications. We refer to this as the AI+X approach. This paper explores its potential for the emerging domain of Computational Pathology, which involves analysis of large whole-slide tissue images with a multi-stage pipeline. The pipeline requires support for a number of different kernels and operators - early stages perform segmentation and feature extraction, followed by graph creation with k nearest neighbor (kNN) algorithms, and finally inference with an iterative graph convolutional network (GCN) that alternates between Aggregation and Combination. We show that these stages execute inefficiently on a range of baseline CPU, GPU, AI, and GCN accelerators. That inefficiency is addressed with a combination of software re-structuring and small modifications to a baseline systolic AI accelerator. Many of the above kernels can be mapped to a systolic accelerator by offering a flexible datapath between processing elements and register access mechanisms. We add support for feature aggregation, load balanced execution, Euclidean distance calculation, binning, and counter aggregation. This additional flexibility and logic grows the area of a baseline AI chiplet by 1.1x, but by avoiding the memory wall and offering high parallelism, the proposed accelerator BEACON yields over an order of magnitude higher throughput for Computational Pathology than baseline CPU and GPU platforms.

cs.AR↗

arXiv2026

Bio-inspired Learning and Decision-Making with Probabilistic In-Memory Computing Hardware: Part 2

This report extends our previous work (Part 1), which introduced an energy-based model for learning and decision-making under uncertainty. The model leverages stochastic Langevin dynamics to continuously evolve approximate probability distributions over neuron states and model weights. However, as noted in Part 1 and confirmed through GPU-based implementations, large-scale probabilistic energy-based models of this nature face significant scalability challenges due to excessive execution latency. This latency stems from a fundamental mismatch: massively parallel models with low arithmetic intensity (such as energy-based models) are being executed on processor architectures like GPUs that rely on high-bandwidth memory (HBM) interfaces. The HBM imposes brutally sequential execution constraints on inherently parallelizable models, creating the false impression that such models are unscalable. In reality, it is the GPU architecture itself, with its dependence on HBM interfaces, that is not a scalable processor architecture for this class of AI model. In this report, we demonstrate using a detailed transaction-level model (TLM) of a probabilistic analogue in-memory computing (AIMC) processor that the same energy-based model can execute well over 1000x faster than data-center-grade hardware by eliminating the HBM interface and performing computation directly within on-chip memory.

cs.AR↗