fortress-technologies.com · web viewmicrosoft visual studio 2010 tools for office runtime should...

4
Who needs Excelbooster Microsoft Excel is the most popular spreadsheet software throughout the world regardless industry. However, one critical drawback is slow speed especially when you simulate lots of data. Therefore, you had to rely on a special solution or expensive and cumbersome CPU Grids. Excel Booster is for you if you use Excel and VBA to simulate Variable Product pricing, CTE reserve calculation, Greeks calculation for dynamic hedging, modeling policyholders for IFRS4 Phase2 requirement, pricing derivative securities, modeling catastrophe (cyclone, earthquake, …) etc. For an example, you can simulate 2 million variable policyholders with 1000 scenarios for 100 years in 1 hour with 8 of Nvidia Titan X GPU cards. Convert/Compile with C++ Excel Booster converts all numbers in Excel cell and VBA code to native C++ codes. This conversion process requires least efforts and will let you execute your program more than 10 times faster depends on environments. Convert/Compile with Microsoft C++ AMP C++ AMP (Accelerated Massive Parallelism) accelerates execution of your program by taking advantage of data parallel hardware such as a GPU (Graphics Processing Unit) card. This process requires some modification of your VBA codes and will let you execute your program 100 times 5,000 times faster depends on the GPU cards you are using. You can use any (AMD, Intel, Nvidia) GPU card with this option. Convert/Compile with Cuda C This process requires technical modification of your VBA codes and will let you execute your program 100 times 5,000 times faster depends on the GPU cards you are using. You can use only Nvidia GPU card with this option. System Requirements Excelbooster requires Microsoft Excel 2007 or higher version and also requires Visual Studio 2013 Express Windows Desktop version or Community version which can be downloaded free of cost. Microsoft C++ AMP requires DirectX driver. Cuda C requires CUDA SDK 7.0 version. Install Excelbooster Excelbooster installation file can be downloaded from following web-site: http://www.excelbooster.com/eb/ eb_installer.exe For any security reason, if you can’t download Excelbooster from the above link, please download zip file from following link and unzip to install. http://www.excelbooster.com/eb/ eb_installer.zip Microsoft Visual Studio 2010 Tools for

Upload: others

Post on 07-Apr-2020

8 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: fortress-technologies.com · Web viewMicrosoft Visual Studio 2010 Tools for Office Runtime should be installed first; however, if it has not been installed already, a pop-up screen

Who needs Excelbooster

Microsoft Excel is the most popular spreadsheet software throughout the world regardless industry. However, one critical drawback is slow speed especially when you simulate lots of data. Therefore, you had to rely on a special solution or expensive and cumbersome CPU Grids.

Excel Booster is for you if you use Excel and VBA to simulate Variable Product pricing, CTE reserve calculation, Greeks calculation for dynamic hedging, modeling policyholders for IFRS4 Phase2 requirement, pricing derivative securities, modeling catastrophe (cyclone, earthquake, …) etc. For an example, you can simulate 2 million variable policyholders with 1000 scenarios for 100 years in 1 hour with 8 of Nvidia Titan X GPU cards.

Convert/Compile with C++

Excel Booster converts all numbers in Excel cell and VBA code to native C++ codes. This conversion process requires least efforts and will let you execute your program more than 10 times faster depends on environments.

Convert/Compile with Microsoft C++ AMP

C++ AMP (Accelerated Massive Parallelism) accelerates execution of your program by taking advantage of data parallel hardware such as a GPU (Graphics Processing Unit) card. This process requires some modification of your VBA codes and will let you execute your program 100 times 5,000 times faster depends on the GPU cards you are using. You can use any (AMD, Intel, Nvidia) GPU card with this option.

Convert/Compile with Cuda C

This process requires technical modification of your VBA codes and will let you execute your program 100 times 5,000 times faster depends on the GPU cards you are using. You can use only Nvidia GPU card with this option.

System Requirements

Excelbooster requires Microsoft Excel 2007 or

higher version and also requires Visual Studio 2013 Express Windows Desktop version or Community version which can be downloaded free of cost. Microsoft C++ AMP requires DirectX driver. Cuda C requires CUDA SDK 7.0 version.

Install Excelbooster

Excelbooster installation file can be downloaded from following web-site:http://www.excelbooster.com/eb/eb_installer.exe

For any security reason, if you can’t download Excelbooster from the above link, please download zip file from following link and unzip to install.http://www.excelbooster.com/eb/eb_installer.zip

Microsoft Visual Studio 2010 Tools for Office Runtime should be installed first; however, if it has not been installed already, a pop-up screen will appear automatically.

Page 2: fortress-technologies.com · Web viewMicrosoft Visual Studio 2010 Tools for Office Runtime should be installed first; however, if it has not been installed already, a pop-up screen

Activate Excelbooster

Activate Excelbooster with activation key

Excel Trust Center Settings

Please select “More Commands…” menu from drop down toolbar menu at the top of the left corner.

Select “Trust Center” menu and click “Trust Center Settings…” button.

Select “Add-ins” menu and undo all checkboxes.

Microsoft Visual Studio 2013 Express

Please download Visual Studio 2013 Express version from following link.http://www.microsoft.com/en-us/download/details.aspx?id=44914

Congratulation! You are ready to use Excelbooster!

Run Sample Program

1. Save a blank workbook with xlsm extension:

- Excel starts with a blank workbook (book1) with 3 Sheets. Save this blank

Page 3: fortress-technologies.com · Web viewMicrosoft Visual Studio 2010 Tools for Office Runtime should be installed first; however, if it has not been installed already, a pop-up screen

workbook with xlsm extension for using VBA macros.

- Excelbooster works with only one workbook; therefore, please close all other workbook except one.

2. Click “Add Module” Menu to create a sample VBA code in module:

3. Click “VBA Editor” to review the sample VBA codes in module 1:

- The psuedo code ‘’START defines where to start.

- The pseudo code ‘’IGN{ and ‘’IGN} defines the codes to ignore which work only with Excel.

- The pseudo code ‘’FMP:FOR(Idx,1,1000) defines the code which will be execute with GPU parallel process.

4. Save Project by using “Project” menu

- This sample workbook consists with 3 worksheets and 1 module. You need to save project with only module1 which contains all sample codes to convert.

- You don’t need to export any worksheet since the sample program does not use any cell.

5. Convert/Compile with MS C++ option or MS C++ AMP option by using “Build” menu:

6. Excute by using “Local Exec.” menu:

- This sample program is executed in 1.99 seconds with Nvidia 540m laptop GPU card which contains 96 cuda cores.

- The result is 500599999.999984

7. Execute the program with Excel by clicking “Run Sub Start” menu:

Page 4: fortress-technologies.com · Web viewMicrosoft Visual Studio 2010 Tools for Office Runtime should be installed first; however, if it has not been installed already, a pop-up screen

- It will take 123 seconds with Excel.- The result is 500599999.999984.

Exactly same as GPU result.

Parallel Process Coding

Public Sub Calc() Dim Idx As Integer ''FMP:FOR(Idx,1,1000) For Idx = 1 To 1000 Dim Idx2 As Integer, Idx3 As Integer odata(Idx) = idata For Idx2 = 1 To 1000 For Idx3 = 1 To 1000 Dim x As Double Call CalcDiv(Idx2, Idx3, x) odata(Idx) = odata(Idx) + x Next Idx3 Next Idx2 Next IdxEnd Sub

- Sequential variables “Idx” should be declared with integer or long dimension.

- Iteration number “1 to 1000” should be declared with constant number.

- Step feature is not working since the For Loop will executed at the same time .

- The local variables (“Idx2”, “Idx3”) inside the For Loop will be executed simultaneously with GPU cores.