# RowCove performance and comparison

## Published measurement

The public comparison uses one generated **20 GB synthetic CSV** with **39,062,499 records and 16 columns**. It is mock data, not a user dataset. Each accepted operation used three trials on Windows. Applications were already running and the operating-system file cache was not flushed, so this is a warm-cache engineering comparison rather than a universal hardware ranking.

The timing starts before the specified operation and includes source preparation, indexing or key preparation, result-index publication and the returned first page. It stops when data is returned; grid paint is excluded. A complete result was independently checked before publication.

| Task | RowCove Pro | EmEditor Professional | Modern CSV Free |
| --- | ---: | ---: | --- |
| Full-file access plus first 1,000 records × 16 fields | **3.48 s median** | FastView **5.51 s**; normal mode **10.77 s** | No accepted usable-open/page timing |
| Full-file contains filter plus first result page | **0.36 s median** | Normal mode **5.06 s median** | Not measured; installed Free build required Premium for column filtering |
| Open through complete numeric sort plus first result page | **10.14 s median**, including preparation | Not measured; sorting was cancelled at the unavailable-undo warning | Not measured; usable navigation was not confirmed |

RowCove's completed sorted indexes passed bounds, bijection, numeric ordering and stable source-order tie checks. Source and sorted page reads were checked at beginning, middle and end offsets. The public comparison does not claim a universal “fastest” title, because device, file shape, cache state, reader mode and operation semantics matter.

## How to interpret the comparison

- Full-file access is not the same as showing a small preview.
- A first grid appearance is not proof that navigation, filtering or sorting completed.
- EmEditor FastView and normal mode are separate workflows and are not combined.
- EmEditor sorting is explicitly unmeasured; no unavailable warning was bypassed.
- Modern CSV's small mock capability check confirmed numeric sorting, while its 20 GB usable workflow and Free filtering were not accepted.
- RowCove pages are batch reads; another application's per-cell automation calls are not a GUI paging equivalence.

## What RowCove optimizes

The engine streams source-row positions, projects requested columns, keeps UI rendering bounded, reuses valid indexes, writes query artifacts atomically, cleans cancelled partial work and preserves stable source provenance. Filter and sort preparation remains visible as progress rather than appearing to freeze.

These measurements support the claim that RowCove is highly competitive for the tested large-CSV workflows. They do not establish TB throughput, cold-disk speed, whole-machine I/O, or every format's performance. Use the app's progress and capability state for the actual source.