Performance difference when reading/writing many files with EPPlus versus Spreadsheet Gear - performance

Question

I've made a simple performance test between EPPlus and Spreadsheet Gear to see if there is any significant difference that would justify buying Spreadsheet Gear.
I am no expert at either application so it's possible the tests aren't written the most efficient way.
The test does the following:
1. Opens an existing Excel-file with 1000 rows and 3 columns. Saves the three values into an entity that is saved into a List<>.
2. Open a new Excel-object
3. Create a header row (bold) with the title of each column.
4. Write back the 1000 entities.
5. Save the new Excelfile.
If I run this test once EPPlus comes out the winner (approx times are EPPlus = 280ms, SG = 500ms). If I run the test 10 times in a row instead (a for-loop opening, copying, saving 10 seperate times) Spreadsheet Gear is faster instead (approx times per file: EPPlus = 165ms, SG = 95ms). For 20 tests the approx times are EPPlus = 160ms / file and SG = 60ms / file.
It seems like (to a certain extent at least) Spreadsheet Gears gets faster and faster the more files I create.
Could anyone explain why EPPlus is the slower one when running consecutive tests? And can I make changes to the code to change this?
EPPlus test function:
var timer = new Stopwatch();
timer.Start();
var data = new List<Item>();
using (var excelIn = new ExcelPackage(new FileInfo(folder + fileIn)))
{
var sheet = excelIn.Workbook.Worksheets[1];
var row = 2;
while (sheet.Cells[row, 1].Value != null)
{
data.Add(new Item()
{
Id = int.Parse(sheet.Cells[row, 1].Text),
Title = sheet.Cells[row, 2].Text,
Value = int.Parse(sheet.Cells[row, 3].Text)
});
row++;
}
}
using (var excelOut = new ExcelPackage())
{
var sheet = excelOut.Workbook.Worksheets.Add("Out");
sheet.Cells.LoadFromCollection(data);
sheet.InsertRow(1, 1);
sheet.Cells[1, 1, 1, 3].Style.Font.Bold = true;
sheet.Cells[1, 1].Value = "Id";
sheet.Cells[1, 2].Value = "Title";
sheet.Cells[1, 3].Value = "Value";
excelOut.SaveAs(new FileInfo(folder + "EPPlus_" + Guid.NewGuid() + ".xlsx"));
}
timer.Stop();
return timer.ElapsedMilliseconds;
Spreadsheet Gear:
var timer = new Stopwatch();
timer.Start();
var data = new List<Item>();
var excelIn = Factory.GetWorkbook(folder + fileIn);
var sheetIn = excelIn.Worksheets[0];
var rowIn = 1;
while (sheetIn.Cells[rowIn, 0].Value != null)
{
data.Add(new Item()
{
Id = int.Parse(sheetIn.Cells[rowIn, 0].Text),
Title = sheetIn.Cells[rowIn, 1].Text,
Value = int.Parse(sheetIn.Cells[rowIn, 2].Text)
});
rowIn++;
}
excelIn.Close();
var excelOut = Factory.GetWorkbook();
var sheetOut = excelOut.Worksheets.Add();
sheetOut.Name = "Out";
var rowOut = 0;
sheetOut.Cells[rowOut, 0, rowOut, 2].Font.Bold = true;
sheetOut.Cells[rowOut, 0].Value = "Id";
sheetOut.Cells[rowOut, 1].Value = "Title";
sheetOut.Cells[rowOut++, 2].Value = "Value";
foreach (var item in data)
{
sheetOut.Cells[rowOut, 0].Value = item.Id;
sheetOut.Cells[rowOut, 1].Value = item.Title;
sheetOut.Cells[rowOut++, 2].Value = item.Value;
}
excelOut.SaveAs(folder + "SpreadsheetGear_" + Guid.NewGuid() + ".xlsx", FileFormat.OpenXMLWorkbook);
excelOut.Close();
timer.Stop();
return timer.ElapsedMilliseconds;
Main function
var runs = 1;
var testerG = new TestSpreadsheetGear();
var testerE = new TestEpPlus();
var msE = 0.0;
var msG = 0.0;
var i = 0;
for (i = 0; i < runs; ++i)
{
msG += new TestSpreadsheetGear().Run(folder, originalFile);
}
for(i = 0; i < runs; ++i)
{
msE += new TestEpPlus().Run(folder, originalFile);
}
Console.WriteLine("Spreadsheet time: " + msG + ". Per file: " + msG / runs);
Console.WriteLine("EP Plus time: " + msE + ". Per file: " + msE / runs);
Console.ReadKey();

Answer 1

I believe that the reason for the results you are seeing is the fact that on the first run the .NET CLR must JIT the code. Since SpreadsheetGear is a complete spreadsheet engine under the hood (as opposed to a read / write library) there is more code to JIT - thus the first run is taking longer for SpreadsheetGear than EPPlus (I am speculating here but have a great deal of experience in benchmarking .NET code over the last 10 years).
I do not have EPPlus installed but I did write a test which tries to do the same thing you are doing. with SpreadsheetGear 2012 Since I don't have your starting workbook I first build the workbook. Then, I used more optimal SpreadsheetGear APIs. The first time I run I get 141 milliseconds for SpreadsheetGear 2012. After the first run I get 9 or 10 milliseconds for each run on an overclocked Core i7-980x running Win7 x86 and a release build run without debugger.
I have pasted my code below (just paste it into a .NET 4.0 C# console application).
One more thought I have is that this is a very small test case. To really see the performance of SpreadsheetGear 2012 try this with 100,000 rows or even 1 million rows.
Disclaimer: I own SpreadsheetGear LLC
using System;
using System.Collections.Generic;
using System.Diagnostics;
using SpreadsheetGear;
namespace SGvsEPPlus
{
class Program
{
internal struct Item
{
internal Item(int id, string title, int value)
{
Id = id;
Title = title;
Value = value;
}
internal int Id;
internal string Title;
internal int Value;
}
static void Test(int rows)
{
string filename = #"C:\tmp\MyWorkbook.xlsx";
Console.Write("Test({0})...", rows);
var timer = new Stopwatch();
// Create workbook since we don't have poster's original workbook.
timer.Restart();
var workbook = Factory.GetWorkbook();
var values = (SpreadsheetGear.Advanced.Cells.IValues)workbook.Worksheets[0];
for (int row = 1; row <= rows; row++)
{
values.SetNumber(row, 0, row);
values.SetText(row, 1, "Title " + row);
values.SetNumber(row, 2, row * 10);
}
Console.Write("Create workbook={0:0}...", timer.Elapsed.TotalMilliseconds);
// Save workbook
timer.Restart();
workbook.SaveAs(filename, FileFormat.OpenXMLWorkbook);
Console.Write("Save workbook={0:0}...", timer.Elapsed.TotalMilliseconds);
// Track total time of original test.
var totalTimer = Stopwatch.StartNew();
// Open workbook
timer.Restart();
var excelIn = Factory.GetWorkbook(filename);
Console.Write("Open excelIn={0:0}...", timer.Elapsed.TotalMilliseconds);
// Copy workbook to list
timer.Restart();
var sheetIn = excelIn.Worksheets[0];
var valuesIn = (SpreadsheetGear.Advanced.Cells.IValues)sheetIn;
var rowIn = 1;
var data = new List<Item>(rows);
while (valuesIn[rowIn, 0] != null)
{
data.Add(new Item(
(int)valuesIn[rowIn, 0].Number,
valuesIn[rowIn, 1].Text,
(int)valuesIn[rowIn, 2].Number));
rowIn++;
}
excelIn.Close(); // Not necessary but left for consistency.
Console.Write("excelIn->data={0:0}...", timer.Elapsed.TotalMilliseconds);
timer.Restart();
var excelOut = Factory.GetWorkbook();
var sheetOut = excelOut.Worksheets[0];
var valuesOut = (SpreadsheetGear.Advanced.Cells.IValues)sheetOut;
sheetOut.Name = "Out";
var rowOut = 0;
sheetOut.Cells[rowOut, 0, rowOut, 2].Font.Bold = true;
sheetOut.Cells[rowOut, 0].Value = "Id";
sheetOut.Cells[rowOut, 1].Value = "Title";
sheetOut.Cells[rowOut++, 2].Value = "Value";
foreach (var item in data)
{
valuesOut.SetNumber(rowOut, 0, item.Id);
valuesOut.SetText(rowOut, 1, item.Title);
valuesOut.SetNumber(rowOut, 2, item.Value);
rowOut++;
}
Console.Write("data->excelOut={0:0}...", timer.Elapsed.TotalMilliseconds);
timer.Restart();
excelOut.SaveAs(#"C:\tmp\SpreadsheetGear_" + Guid.NewGuid() + ".xlsx", FileFormat.OpenXMLWorkbook);
excelOut.Close(); // Again - not necessary.
Console.WriteLine("Save excelOut={0:0}...", timer.Elapsed.TotalMilliseconds);
Console.WriteLine(" Total={0:0}", totalTimer.Elapsed.TotalMilliseconds);
}
static void Main(string[] args)
{
// Do it three times with 1000 rows. Note that the first
// time takes longer because code must be JITted.
Test(1000);
Test(1000);
Test(1000);
}
}
}