arXiv subjects

Manuel Bouyer

Publications and source records attributed to Manuel Bouyer.

3 recordsLinked to original sources

arXiv2025

Benchmarking Deep Learning Convolutions on Energy-constrained CPUs

This work evaluates State-of-the-Art convolution algorithms for CPU-based CNN inference. Although most prior studies focus on GPUs or NPUs, CPU implementations remain comparatively under-optimized. Our first contribution is to provide fair benchmarking for embedded CPU inference. We evaluate direct, GEMM-based, and Winograd convolutions across modern CPUs from ARM, Intel, AMD, and NVIDIA vendors, considering both latency and energy efficiency. To the best of our knowledge, this is the first study to present a fair, cross-vendor comparison of CPU energy consumption using a high-resolution socket-level measurement platform. To validate our methodology, we further compare socket-level power measurements with estimates derived from model-specific registers (MSRs), finding that MSRs underestimate the power consumption of convolution inference by 10--30%. Our results show that the ARM\R Cortex-A78AE CPU combined with an implicit GEMM convolution implementation offers the best trade-off between latency and power consumption, achieving ResNet50v1.5 inference in 102 ms with an average power of 25.3 W, corresponding to 2.58 J.

cs.AI↗

arXiv2025

Dalek: An Unconventional and Energy-Aware Heterogeneous Cluster

Dalek is an experimental compute cluster designed to evaluate the performance of heterogeneous, consumer-grade hardware for software design, prototyping, and algorithm development. In contrast to traditional computing centers that rely on costly, server-class components, Dalek integrates CPUs and GPUs typically found in mini-PCs, laptops, and gaming desktops, providing a cost-effective yet versatile platform. This document details the cluster's architecture and software stack, and presents results from synthetic benchmarks. Furthermore, it introduces a custom energy monitoring platform capable of delivering 1000 averaged samples per second with milliwatt-level resolution. This high-precision monitoring capability enables a wide range of energy-aware research experiments in applied Computer Science.

cs.DC↗

arXiv2025

Energy-Aware Scheduling Strategies for Partially-Replicable Task Chains on Heterogeneous Processors

The arrival of heterogeneous (or hybrid) multicore architectures has brought new performance trade-offs for applications, and efficiency opportunities to systems. They have also increased the challenges related to thread scheduling, as tasks' execution times will vary depending if they are placed on big (performance) cores or little (efficient) ones. In this paper, we focus on the challenges heterogeneous multicore processors bring to partially-replicable task chains, such as the ones that implement digital communication standards in Software-Defined Radio (SDR). Our objective is to maximize the throughput of these task chains while also minimizing their power consumption. We model this problem as a pipelined workflow scheduling problem using pipelined and replicated parallelism on two types of resources whose objectives are to minimize the period and to use as many little cores as necessary. We propose two greedy heuristics (FERTAC and 2CATAC) and one optimal dynamic programming (HeRAD) solution to the problem. We study an open source implementation of the DVB-S2 communication standard based on the StreamPU runtime. Leading processor vendors are covered with ARM, Apple, AMD, and Intel platforms. Both the achieved throughput and the energy consumption are evaluated. Our results demonstrate the benefits and drawbacks of the different proposed solutions. On average, FERTAC and 2CATAC achieve near-optimal solutions, with periods that are less than 10% worse than the optimal (HeRAD). These three scheduling strategies now enable programmers and users of StreamPU to transparently make use of heterogeneous multicore processors and achieve a throughput that differs from its theoretical maximum by less than 6% on average. On the DVB-S2 receiver, it is also shown that the heterogeneous solutions outperform the best homogeneous ones in terms of energy efficiency by 8% on average.

cs.DC↗