Skip to content
demuddler

Performance

The time and memory that the importer needed to open and check CSV files of 1 million to 15 million rows on one laptop.

The importer has no fixed row limit for opening a file. The memory of the device and the browser sets the largest file that opens. A plan limits only the rows that an export or a delivery can contain. See Plans and limits.

This page lists what we measured on one laptop on 8 November 2026. Your times and memory use can differ.

Each run opened the demo, chose the file and waited for each step. Each case ran once.

  • Open is the time from choosing the file until the parse-complete event. It includes the start of the hosted frame.
  • Checked is the time from choosing the file until the validate event, when the first check of every value finished. The importer made the schema from the file. No host schema or validation hook ran.
  • Peak memory is the highest total memory of all the browser processes. It includes the browser itself, the page and the hosted frame.
  • Parsed data is the memory that the parsed file used in the importer, from importer.stats.heapBytes.

Each file was a generated CSV file with 25 columns of customer records. It had problems that real exports often have. It also had accented names and text in other scripts.

The laptop had an 8-core processor from 2021 and 16 GB of memory, and ran Linux.

1 GB is 1,000 MB, and 1 MB is 1,000,000 bytes.

Rows File size Browser Open Checked Peak memory Parsed data
1,000,000 133 MB Chromium 3.7 s 3.9 s 0.54 GB 151 MB
1,000,000 133 MB Firefox 4.2 s 4.5 s 1.01 GB 151 MB
1,000,000 133 MB WebKit 4.5 s 4.8 s 0.84 GB 151 MB
5,000,000 686 MB Chromium 17.6 s 18.4 s 1.50 GB 764 MB
5,000,000 686 MB Firefox 20.1 s 21.1 s 1.97 GB 764 MB
5,000,000 686 MB WebKit 22.1 s 22.8 s 1.74 GB 764 MB
15,000,000 2.09 GB Chromium 54.1 s 56.2 s 3.69 GB 2.31 GB
15,000,000 2.09 GB Firefox 62.2 s 64.5 s 3.95 GB 2.31 GB
15,000,000 2.09 GB WebKit 66.0 s 67.8 s 4.06 GB 2.31 GB

A CSV export of a 1,000,000-row file with no problems took 10.9 s in Chromium, with a peak memory of 0.95 GB.

  • Time and memory grew in a straight line with the number of rows.
  • The parsed data used about 1.1 times the size of the CSV file. The browser needed more in total: 1.50 to 1.97 GB for the 686 MB file and 3.69 to 4.06 GB for the 2.09 GB file.
  • The parsed data must fit in the memory of the importer, which has a limit of about 4 GB. The 2.09 GB file used 2.31 GB of it. A file with longer values or more columns per row reaches the limit sooner.
  • We measured one laptop, one run for each case. We did not measure phones, tablets, Windows or macOS.
  • A device with less memory or an older processor is slower. It can run out of memory with a smaller file. A phone or tablet usually gives a page much less memory than a desktop browser.
  • Other open tabs and programs share the device’s memory.
  • We did not measure Excel files, a host schema with many rules, validation hooks, edits, fixes, or an export of the larger files.
  • Your files differ from ours. Long text values, many columns and many problems change the time and memory.

Open your file in the demo on the devices that your users have. The demo shows the row count while it reads the file.

In your own application, the parse-complete event carries the parse statistics, and validate follows when the check finishes. See Events and React callbacks.

importer.addEventListener("parse-complete", (event) => {
const { stats } = event.detail;
console.log(stats.schema.n_rows, "rows in", Math.round(stats.elapsedMs), "ms");
});

The Chrome task manager and the Firefox about:processes page show the memory of the page and the importer frame. If a file does not open, see A large file exceeds browser memory.