Scholarlink

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® 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.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Benchmarking Deep Learning Convolutions on Energy-Constrained CPUs

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Enrique Galvez,
  • Adrien Cassagne,
  • Alix Munier,
  • Manuel Bouyer

摘要

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® 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.