Aggregate & Group: Validated at Scale. 200 Million Rows / 10.4 GB.
Full 19-function advanced mode on a 10.4 GB / 200M-row fixture (1M groups), streamed through OPFS — never uploaded.
Throughput by File Size — Simple vs Advanced Mode
| File Size | Simple Mode (5 functions) | Advanced Mode (19 functions) | Notes |
|---|---|---|---|
| 200M rows · 10.4 GBGate 3 | ~235K rows/sec | ~270K rows/sec | Gate 3 (May 2026) — 1M groups, OPFS external sort, peak heap 308–605 MB, ~8.2 GB OPFS disk, both configs 0 group failures. I/O-bound at this scale, so aggregation config has only minor effect on throughput. |
Advanced mode: All 19 functions including MEDIAN, P25–P95, MODE, STDDEV, VARIANCE, FIRST, LAST, COUNT_DISTINCT, CONCAT_UNIQUE — requires value-storage arrays per group for accurate computation.
All benchmarks: Chrome, Windows 11, dev machine (Alder Lake class), 64GB RAM, Gate 3, May 2026. 10.4 GB / 200M-row fixture (1M groups), OPFS external sort. Results vary by hardware, browser, and configuration.
Rows Per Second — Visual Comparison
Simple Mode — 5 Functions (COUNT, SUM, AVG, MIN, MAX)
Higher values indicate better performance (faster processing)
Advanced Mode — 19 Functions (includes MEDIAN, Percentiles, STDDEV)
Higher values indicate better performance (faster processing)
Performance improves with larger files due to 10MB chunk batching amortizing startup overhead. Results vary ±15–20% by hardware, browser, and function selection.
Benchmark Methodology & Test Configuration
Why Browser-Based Aggregation Outperforms Excel on Large Files
What Excel Pivot Tables Do
What SplitForge Does
The architectural result: Excel's performance ceiling is determined by worksheet materialization cost — rendering 1M+ rows into a grid exhausts memory before aggregation even begins. SplitForge's ceiling is determined by HashMap cardinality (number of unique group combinations) and function selection. Aggregating 200M rows with 1M unique groups stores ~1M aggregate records — not 200 million rows. That's why SplitForge validated a 10.4 GB / 200M-row aggregation (Gate 3, May 2026) while Excel cannot open the file at all.
Time Savings Calculator
Baseline: Excel Pivot on large file ≈ 20 min per analysis. SplitForge ≈ 2 min.
20-minute baseline derived from internal timing of Excel Pivot Table setup and processing on 500K–1M row exports (field collapse, recalculate, filter copy). Results vary by file size and analyst experience.
Function Performance Characteristics
Online algorithms — maintain running state without storing values. No additional RAM per group beyond the running total. Performance is essentially constant regardless of group size.
STDDEV/VARIANCE use Welford's algorithm — O(1) memory but O(n) computation. COUNT_DISTINCT and CONCAT_UNIQUE maintain a Set per group. FIRST/LAST require tracking insertion order.
Require storing all values per group to compute the true median or percentile. At 10M rows with 500 groups, this means potentially millions of stored values. Sorted at finalization — O(n log n) per group. Select only when you need true distributional statistics.
Honest Limitations: Where Falls Short
No tool is perfect for every use case. Here's where might be a better choice, and the real limitations of our browser-based architecture.
Browser-Based Processing
Performance depends on your device's RAM and CPU. Modern laptops (2022+) handle 10M+ rows easily, but older devices may struggle with very large files.
No Offline Mode (Initial Load)
Requires internet connection to load the tool initially. Processing happens offline in your browser after loading.
Browser Tab Memory Limits
Most browsers limit individual tabs to 2-4GB RAM. This is the practical ceiling for file size.
Questions about limitations? Check our FAQ section below or contact us via the feedback button.
Frequently Asked Questions
How was the 10.4 GB / 200M-row benchmark measured?
Why is advanced mode slower than simple mode?
Why does performance improve with larger files (1K rows is slower than 10M)?
How does this compare to Excel Pivot Tables at scale?
What happens with high-cardinality GROUP BY columns?
Does the HAVING filter affect processing speed?
What's the maximum file size SplitForge can handle?
What browser gives the best performance?
Does SplitForge transmit any file data for analytics?
Benchmarks last updated: May 2026 (Gate 3) · Conducted by the SplitForge engineering team · Aggregation Engine (OPFS external sort) · Recorded at 10.4 GB / 200M rows (1M groups) · Privacy verified: no outbound file payload requests (network inspector confirmed)