This is the first article in a series that I will write about on the topic of parallel programming and CUDA. In this guide I will explain how to install CUDA 6.0 for Mac OS X. CUDA is a proprietary programming language developed by NVIDIA for GPU programming, and in the last few years it has become the standard for GPU computing. GPU computing is a new branch of computer science and, more specifically, of parallel computing.

I will cover parallel computing in detail in later articles, but if I had to outline in a few words what parallel computing is I would say that it is the partition of repetitive (and therefore often time consuming) tasks into singular tasks, each of which is performed by a single core or processing unit.

To better explain the key concept behind parallel computing, let me give you a more 'realistic' example. You have to move to a new town and are looking to hire a furniture mover company. You have two choices: the first company can provide you with the strongest man alive driving the world's fastest van. The second one instead has 5 normal men available, each of them with a van of average speed. What would you prefer? Of course it depends of how much stuff you have to move and how much time you have to do it.

If you need the work done quickly and have to move a lot of furniture, of course you will chose the second company. Each of the men is much slower than the strong guy, but if they are able to transport your load concurrently they will have the time to complete the task, have a beer together and come back to their family before the strong guy has done half the work. GPU computing is like having hundreds and thousands of skinny guys with old vans, which can be pretty effective.

This is the main idea about parallel computing. What follows are the instructions to install the necessary software to use thousands of 'vans' to enhance the performance of your code.

The Mac 285 is the one I boot from and is attached to my main display - it is the PC card that is just cuda coprocessor and not driving a monitor most of the time, and that one stays up at 1.48GHz come what may. With only the Mac card loaded I got it up to just over 100,000 options per sec - there is clearly an overhead for the multi operation.

CUDA Installation On Mac OSX

In this guide, I explain how to install CUDA 6.0 (the latest release as I write) on your computer under OS X.

This Mac, click the More Info button, and then select Graphics/Displays under the Hardware list. There you will find the vendor name and model of your graphics card. If it is an NVIDIA card that is listed on the CUDA-supported GPUs page, your GPU is CUDA-capable. The Release Notes for the CUDA Toolkit also contain a list of supported. CUDA Driver for Mac is a very useful software package that provides support for a large collection of NVIDIA video cards. The CUDA Driver is designed for all NVIDIA products available on the Mac hardware platform and it will install an applet in System Preferences. Drivers; Graphics Cards; Nvidia CUDA Driver for Mac 396.148 The CUDA Driver is designed to bring support for all Nvidia graphic cards under Mac hardware. CUDA Mac Driver Latest Version: CUDA 418.163 driver for MAC Release Date: Previous Releases: CUDA 418.105 driver for MAC Release Date: CUDA 410.130 driver for MAC.

Finder menu 'GO'

General System Requirements

There are some general requirements which are common regardless of the operating system you use. The most important one is a CUDA-capable GPU, so you need a recent NVIDIA graphics card. A full list is available here. The other requirement is to have the CUDA Toolkit, which is free and provides you all you need to install and run CUDA code. By clicking on this link you will open the download page. Choose your operating system (in this case I presume Mac OS X) and then download the package by clicking on PKG.

CUDA 6.0 Downloads page

If you don't know the card/cards available on your Mac, then just go to Application > Utilities > System Information or search for System Information with Spotlight in this way and click on the icon.

Searching with Spotlight

Search for and download any torrent from the pirate bay using search query Splinter Cell: Blacklist. Direct download via magnet link. Tom Clancy's Splinter Cell Double Agent is a game more than worthy of the saga that gives it its name. We start with a very elaborate shell and argument, which has the user in tension, and according to the degree of trust we have with both organizations, we will enjoy interesting and varied consequences, which offers many hours of fun and surprises. Splinter cell for pc download.

Then look for Graphics/Displays and you will get information on your graphics card, including the name. In this case I have two of them and, fortunately, one of them is an NVIDIA.

System Information - Graphics/Displays

Mac OS X Requirements

If you want to install CUDA 6.0 on a Mac you need to have OS X 10.8 (Lion) or later running on your computer. If you don't, you can download Mavericks (OS X 10.9) from the App Store, which is free. In the situation where you cannot install OS X 10.8 or later, don't worry: you can install older CUDA releases. The installation is similar, but if some of you experience trouble, I will provide help in future articles.

In order to use CUDA 6.0 you need the GNU Compiler Collection (gcc) and clang on your Mac. They are compliers for the C-family of programming languages and CUDA is a library for them. Actually, you will code the 'serial' (non-parallel) parts of your code in C or C++. To get these two compilers you first need the Command Line Tools, which once again are free to download from the App Store.

Anyway, as you might have already installed the Command Line Tools without even noticing (especially if you already have XCode), I suggest you check its presence by going to the folder /Library/Developer/ by clicking on Go to Folder in the Finder menu or by using the shortcut Shift - Apple Key - G after clicking the Finder icon.

Go to Folder Window

If you find the folder named CommandLineTools you can then check for needed packages by entering it, opening the usr folder and then bin. If you already have the CommandLineTools folder you should very likely even have the clang and gcc packages. If you don't then you will need to download them.

Open your Terminal (which can be found in /Applications/Utilities or, as usual, by typing “Terminal” in Spotlight and then clicking on its icon) and typing into the terminal:

Terminal on OS X

Then a window will appear, asking you if you want to download and install this package. Click yes, and the installation will proceed. Do not close the Terminal as we will need it later.

CommandLineTools installation window

Installing CUDA Packages

In the previously downloaded CUDA package, you should find three distinct packages: CUDA Driver, CUDA Toolkit and CUDA Samples. Despite the fact that the only essential one is the CUDA Driver, I suggest you install all of them, as they are very useful.

Packages selection during CUDA 6.0 installation

As always installation of Mac OS X packages is straightforward. Just click 'Continue' on every window. At one point you will have to agree to terms and conditions by clicking on the Agree button when it appears) and then Install in the final one. Now you have CUDA 6.0 installed on your Mac, but it is not quite ready to be used. You have to carry out the final two steps. Open the Terminal (if you missed how to do this, just read a couple of paragraphs above) and then type or, more conveniently, copy and paste, the following:

Cuda Mac Os

Setting path variables on Terminal

This will set the environment path variables, which is necessary to use CUDA. If you have a Mac with a CUDA-capable graphic card, you probably actually have two graphic cards: an integrated one and a separate, discrete one. Since it is consumes a lot of batter power to always have the discrete card in use, OS X could prevent it by automatically switching between the two cards. So, when you want to use CUDA you have to be sure that your Mac won't 'decide' to use the integrated card while you run

Cuda Cards For Mac