跳到正文
研踪PaperTrail
← 研踪 PaperTrail

arXiv:2609.06635 · v2

From JPEG to CUDA Batch: Measuring RGB Input Pipeline Throughput

首次提交 2026-09-06 22:32
版本更新 2026-11-16 10:42
雷达观测 2026-11-17 17:12
arXiv ↗PDF ↗

摘要

We measure how many ready batches per second eight image input pipelines deliver to an NVIDIA L4. Each pipeline starts with RGB JPEG files and delivers synchronized CUDA float16 batches of 256 images. The benchmark contains 1314 measurements across 73 recipes. On the 34 recipes shared by all eight paths, DALI delivered a median of 19.7 batches/s and AlbumentationsX (AX) delivered 17.1. DALI was fastest on 34 of those recipes. Median peak process GPU memory was 1,872 MiB for DALI and 404 MiB for AX.

作者备注:31 pages, 6 figures. Expanded the benchmark to 73 recipes and eight input paths, revised the comparisons, and updated the results appendix. Code and data: https://github.com/albumentations-team/benchmark/tree/154baa0da1880013941daad1b6bd35c58097f969

AI 论文解读

基于摘要 · 非全文审阅

此版本尚未生成 AI 解读。解读由后台任务生成并按论文版本保存,登录后共享已有结果。

版本差异

本站首次收录为 v2,暂无更早版本可对照。

观测到的版本

相邻记录可能跳过中间版本;首次发现 v2 及以上也会标为版本更新。

v2 · 2026-11-16 10:42 · 最新观测版本

From JPEG to CUDA Batch: Measuring RGB Input Pipeline Throughput

We measure how many ready batches per second eight image input pipelines deliver to an NVIDIA L4. Each pipeline starts with RGB JPEG files and delivers synchronized CUDA float16 batches of 256 images. The benchmark contains 1314 measurements across 73 recipes. On the 34 recipes shared by all eight paths, DALI delivered a median of 19.7 batches/s and AlbumentationsX (AX) delivered 17.1. DALI was fastest on 34 of those recipes. Median peak process GPU memory was 1,872 MiB for DALI and 404 MiB for AX.

查看此版本解读 →查看此版本原文 ↗