2. Installation

The VEGA package is available in different archives/setups:

where X.X.X and X.X.X.X are the release version of each archive/setup. You you can download only the package for your system, because each archive contains all the files needed, but not the source code that is provided in a separate archive.

2.1 Unix installation

This section of the manual shows the steps needed to install the VEGA package on Unix-like operating systems (e.g. IRIX, Linux, NetBSD, etc). If your operating system is not directly supported by Authors, you must download and compile the source code (Vega_X.X.X_Source.tar.gz archive).

2.1.1 Building the VEGA package

The portable source code allows to build the package virtually for any computer platform that has a standard ANSI C compiler. It is possible to find some minor compiling problems due to hardware differences. If you can't solve these problems, please contact the Authors.

As first step, you must unpack the Vega_XX_Source.tar.gz file, using the gzip command. If this command is not available in your system, you can download it from any GNU software archive (see: http://www.opensource.org). The correct syntax is:

gzip -d Vega_X.X.X_Source.tar.gz

the unpacked file (Vega_X.X.X_Source.tar) created by gzip must be dearchived with tar command:

tar -xvf Vega_X.X.X_Source.tar

A directory called Vega will be created.

Check if the libraries libbz2.a, loblocale.a, libxdrf.a, libz.a and libZ32.a are present in the directory ...Vega/Src/Vega/MySO where MySO is the directory of your operating system. If not, you must compile these external libraries using the sources placed respectively in ...Vega/Src/Bzip2, ...Vega/Src/LocaleLib, ...Vega/Src/XdrfLib, ...Vega/Src/Zlib and ...Vega/Src/Z32 editing the Makefile and running make. Each library must be copied in ...Vega/Src/Vega/MySO.

After this operation, change your current directory in ...Vega/Src/Vega/MySO (Amiga, Irix, Linux, Unix, Win32) and, if needed, edit the Makefile with your preferred program. Some remarks can help you in this operation. Please set the CC variable to the compiler name (usually cc or gcc) and the CFLAGS variable for the best optimization (e.g. -O -s).

At this point, run make and the VEGA executable is compiled for your system. The makefile was successfully tested with SGI cc and GNU C (gcc for Linux, gcc for AmigaDOS) compilers.

WARNING:
Starting from the 1.5.0 release, it was introduced the use of 64 bit integers to speed-up the manipulation of 8 character strings and thus to build VEGA, it's required a C compiler that supports that integer size. 

2.1.2 Setting-up your Unix system

If the downloaded package has been specifically developed for your operating system, you must unpack it using the following two commands:

gzip -d Vega_X.X.X_MyOS.tar.gz

tar -xvf Vega_X.X.X_MyOS.tar

If you have GNU tar, you can even do it in one step only:

tar -zxvf Vega_X.X.X_MyOS.tar.gz

Please note that the pathway where the archive has been unpacked, is the real installation path.

Test installation

Type the command (one time only):

chmod -R 755 Vega/Bin/*

and every time that you need VEGA:

cd Vega
source ./assign.sh

In this way, the more suitable set of binaries is automatically detected and its folder is added to the command search path.

Full installation
The next step is the editing of your shell start-up script (e.g. .cshrc for csh or tcsh, .bashrc for GNU bash) in order to set VEGADIR and the LD_LIBRARY_PATH environment variables to the installation path. For csh/tcsh shell, you must type:

setenv VEGADIR "<INSTALLATION_PATH>"
setenv LD_LIBRARY_PATH "$VEGADIR/Bin/<MyOS> $LD_LIBRARY_PATH"
setenv PATH "$VEGADIR:$PATH"

where <MyOS> is the operating system  (e.g. Linux_x64, Linux_x86, etc) and  <INSTALLATION_PATH> is the full installation path.

For sh/bash:

export VEGADIR="<INSTALLATION_PATH>"
export LD_LIBRARY_PATH="$VEGADIR $LD_LIBRARY_PATH"
export PATH="$VEGADIR/Bin/<MyOS>:$PATH"

The LD_LIBRARY_PATH is required to inform your system where are the dynamic libraries needed by VEGA. It's strongly recommended to add the installation directory pathway in the command search variable path, defined in shell start-up script.

For example, if you installed VEGA for Linux x64 in /usr/local/vega directory, you must set the environment variables (csh/tcsh):

setenv VEGADIR "/usr/local/vega"
setenv LD_LIBRARY_PATH "$VEGADIR $LD_LIBRARY_PATH"
setenv PATH "$VEGADIR/Bin/Linux_x64:$PATH"

or (sh/bash):

export VEGADIR="/usr/local/vega"
export LD_LIBRARY_PATH="$VEGADIR $LD_LIBRARY_PATH"
export PATH="$VEGADIR/Bin/Linux_x64:$PATH"

Finally, you must change the file permissions:

chmod -R 755 $VEGADIR/Bin

To set the localization language of VEGA program, you can edit the <INSTALLATION_PATH>/Config/prefs file: find the <LANGUAGE> item and select your preferred language (at this time, two languages are supported only: english and italian). The automatic language selection isn't supported by Unix operating systems. The installation can be completed enabling the OpenCL acceleration, if the host operating system is Linux x86/x64.

2.1.2 Linux and OpenCL

If your system is equipped with an OpenCL-enabled device (a GPU or an accelerator) and its driver is correctly installed, VEGA recognizes and uses it automatically. For AMD GPUs (HD4000 series and above) and for systems not equipped with a OpenCL-ready GPU, you must install the ATI Stream SDK. In the second case, the OpenCL acceleration is available even if a compatible GPU is not installed, by the CPU that must support SSE2 instruction set to emulate the OpenCL environment. Although the performance aren't comparable to a real OpenCL device, you can obtain a speed-up due to the massive use of the SIMD instructions by the real-time OpenCL compiler.

To install the ATI Stream SDK:

• go to ATI Stream SDK download page and choose the most appropriate version for your system (Linux 32 or 64 bit);
• with the mministrative rights, download and unpack it in the directory in which it will be installed:

tar -xvzf ati-stream-sdk-vX.X-lnxYY.tgz

where X.X is the SDK version and YY could be 32 or 64 on the basis of the version chosen by you. Remember that the examples and the C includes aren't required by VEGA and if you don't want to develop OpenCL application, you can remove the docs, include, make and samples directories to shrink the installation.

• in your .bashrc file set the following environment variables:

export ATISTREAMSDKROOT=<location where the SDK is installed>
export ATISTREAMSDKSAMPLESROOT=$ATISTREAMSDKROOT # Required if you want build the examples

For 32-bit systems:

export LD_LIBRARY_PATH=$ATISTREAMSDKROOT/lib/x86:$LD_LIBRARY_PATH

For 64-bit systems:

export LD_LIBRARY_PATH=$ATISTREAMSDKROOT/lib/x86_64:$LD_LIBRARY_PATH

• If you installed an AMD OpenCL compatible GPU and you want to use both GPU and CPU (not at the same time), you must register the OpenCL ICD driver. To do it, download the package icd-registration.tgz and type in the command shell (with the amministrative rights):

cd /
tar -xvzf icd-registration.tgz

If you have a Nvidia OpenCL enabled graphic card (GeForce 8 series and above with at least 256 Mb of ram), download and install the latest CUDA Toolkit for Linux.

If you find any problem using the OpenCL acceleration, see the prefs file in the Data directory to enable/disable or to select devices.

2.2 Windows installation

This release contains two versions: the console (command line) and the ZZ (OpenGL) versions. The first one is a true Win32 console application. It supports long filenames and works fine with Windows 2000, XP, Server 2003, Vista, Server 2008 and 7 operating systems (x86 and x64 editions). The package has been compiled using the standard Pentium Pro ® (686) instruction set. The second version is a powerful application with an enhanced graphic interface. For more information, go to VEGA ZZ section.
To install the Vega_ZZ_XX_Setup.exe package, you must execute this file and follow the simple installation wizard. You must remember that you must have the administrator rights to complete the installation.
If your system has a software firewall, you must configure it granting the network access to REBOL.exe, otherwise the scripting system doesn't work because it uses the standard TCP/IP communication ports.
After the installation, run VEGA ZZ and the activation procedure starts. The setup can be completed installing the optional components and enabling the OpenCL acceleration.

2.2.1 Windows and OpenCL

As explained in the Unix/Linux installation section, if your system is equipped with an OpenCL-enabled device (a GPU or an accelerator) and its driver is correctly installed, VEGA and VEGA ZZ recognize and use it automatically. For AMD GPUs (HD4000 series and above) the OpenCL driver is included in the standard driver package. For the systems not equipped with a OpenCL-ready GPU, you can install ATI Stream SDK which provides the OpenCL acceleration emulated by the CPU that must support SSE2 instruction set. Although the performance aren't comparable to a real OpenCL device, you can obtain a speed-up due to the massive use of the SIMD instructions by the real-time OpenCL compiler.

To install the ATI Stream SDK:

• go to ATI Stream SDK download page and choose the most appropriate version for your system (Windows XP 32/64 bit, Windows Vista/7 32/64 bit);
• run ati-stream-sdk-vX.X-OS.exe where X.X is the SDK version and OS is the chosen operating system. The SDK will be unzipped in the setup directory that you selected;
• if you don't intend to develop OpenCL application, you could perform a custom installation, unchecking all components keeping checked ATI Stream SDK v2 Developer only.

If you have a Nvidia OpenCL enabled graphic card (GeForce 8 series and above with at least 256 Mb of ram), download and install the latest CUDA Toolkit for Windows.

If you find any problem using the OpenCL acceleration, see the Preferences window to enable/disable or to select devices.

2.3 AmigaDOS installation

System requirements needed to run the AmigaDOS version of VEGA:

• AmigaOS 2.04 or greater (3.0 recommended);
• Ixemul.library. If this library is not installed, you can find it in the Aminet archive or Geek Gadgets archive;
• PowerPacker.library (optional for data decompression);
• FPU is strongly recommended, but not required.

To install this version of VEGA package, you must unpack the distribution archive using lha command from the command shell:

lha x Vega_XX_Amiga.lha

A directory called Vega will be automatically created. If you don't have lha, you can download it from Aminet. As for Unix systems, you must add in the user-startup file (placed in s directory of your boot disk) the following line:

SetEnv VEGADIR <INSTALLATION_PATH>

Instead of the user-startup modification, you can create a new text file in ENVARC: directory containing a single line with the installation pathway. Starting from 1.2 release, this installation step is needed only if VEGA executable is placed in a directory that is not the installation folder.

If you want to add VEGA to your standard command pathway, you can add in the user-startup the following line:

Path <INSTALLATION_PATH> Add

As final step, reboot your computer. Please note that VEGA for Amiga can accept AmigaDOS and Unix-like pathway specification. Use the specific VEGA version for the CPU installed in your system:

The PowerPC CPUs aren't supported.