GPU utilization in AWS's S3 data-loading test with sequential tar shards (4-256 MB); random per-object reads were ingestion-bound until the worker count was raised (more than 16 workers for near saturation)
~100%observed
| Value kind | observed — Reported measurements, counts, and specifications keep the precision and scope stated by their source; an exact specification is not treated as a range. |
|---|---|
| Scope | One AWS micro-benchmark on its own S3 clients, not a general result. AWS recommends shards of about 100 MB-1 GB; its benchmark tested 4-256 MB. |
| As of | 2025-12 |
| Source | AWS Machine Learning Blog, 'Applying data loading best practices for ML training with Amazon S3 clients' |
| Review | checking…review by 2027-06-15 · fast cadence |
| Recorded changes | last 2026-10-16 · 3 revisions tracked |
| Claim id | gpu-utilization-naive-csv-small-file-loading-vs |
Where the guide uses it
← Full numbers register — every date-stamped figure in the guide, with revision history.