Sheetz Benchmarks v1.0.1
Side-by-side code comparison and JMH performance benchmarks of Sheetz (v1.0.1) vs the 4 major Java Excel libraries.
Libraries Compared
| Library |
Version |
Type |
| Sheetz |
1.0.1 |
Annotation-based, zero boilerplate |
| Apache POI |
5.2.5 |
Manual cell-level API |
| EasyExcel |
4.0.3 |
Annotation-based, listener pattern |
| FastExcel |
0.19.0 |
Lightweight cell-level API |
| Poiji |
5.2.0 |
Read-only, annotation-based |
Code Simplicity
| Library |
Write |
Read |
Total lines |
| Sheetz |
1 line |
1 line |
~15 |
| Apache POI |
~25 lines |
~20 lines |
~60 |
| EasyExcel |
~3 lines |
~12 lines |
~30 |
| FastExcel |
~18 lines |
~15 lines |
~45 |
| Poiji |
N/A |
~1 line |
~15 (read-only) |
Write Performance (ms/op, lower is better)
| Library |
1K rows |
10K rows |
100K rows |
| FastExcel |
6.48 |
31.95 |
309.70 |
| EasyExcel |
11.44 |
58.60 |
542.84 |
| Sheetz |
23.15 |
232.51 |
423.75 |
| Apache POI |
22.46 |
217.17 |
2,453.35 |
At 100K rows, Sheetz is 5.8x faster than Apache POI thanks to automatic SXSSF streaming.
Read Performance (ms/op, lower is better)
| Library |
1K rows |
10K rows |
100K rows |
| FastExcel |
2.43 |
24.88 |
210.17 |
| EasyExcel |
4.91 |
42.66 |
334.17 |
| Apache POI |
10.86 |
106.02 |
1,097.20 |
| Poiji |
12.26 |
114.92 |
1,042.16 |
| Sheetz |
13.18 |
128.35 |
1,285.89 |
How to Run
# Code comparison demos
mvn compile exec:java -Dexec.mainClass="io.github.chitralabs.sheetz.benchmarks.comparison.SheetzComparison"
# JMH benchmarks (full run ~30 min)
mvn clean package
java -jar target/benchmarks.jar
# Quick smoke test
java -jar target/benchmarks.jar -f 1 -wi 1 -i 1
Methodology
- JMH 1.37, JDK 11.0.30 (OpenJDK), Apple Silicon
- 2 forks, 3 warmup iterations, 5 measurement iterations
- All libraries read/write identical data with fixed random seed
Links