Bureaucrats, cc_docs_admin, cc_staff
2,879
edits
CUDA: Difference between revisions
Marked this version for translation
(created, copied from cgeroux's work at CUDA tutorial) |
(Marked this version for translation) |
||
| Line 3: | Line 3: | ||
<translate> | <translate> | ||
<!--T:37--> | |||
"CUDA® is a parallel computing platform and programming model developed by NVIDIA for general computing on graphical processing units (GPUs)."<ref>[http://www.nvidia.com/object/cuda_home_new.html Nvidia CUDA Home Page]. CUDA is a registered trademark of NVIDIA.</ref> | "CUDA® is a parallel computing platform and programming model developed by NVIDIA for general computing on graphical processing units (GPUs)."<ref>[http://www.nvidia.com/object/cuda_home_new.html Nvidia CUDA Home Page]. CUDA is a registered trademark of NVIDIA.</ref> | ||
<!--T:38--> | |||
It is reasonable to think of CUDA as a set of libraries and associated C, C++, and Fortran compilers that enable you to write code for GPUs. See [[OpenACC Tutorial]] for another set of GPU programming tools. | It is reasonable to think of CUDA as a set of libraries and associated C, C++, and Fortran compilers that enable you to write code for GPUs. See [[OpenACC Tutorial]] for another set of GPU programming tools. | ||
== Quick start guide == | == Quick start guide == <!--T:39--> | ||
<!--T:40--> | |||
Here we show a simple example of how to use the CUDA C/C++ language compiler, <code>nvcc</code>, and run code created with it. For a longer tutorial in CUDA programming, see [[CUDA tutorial]]. | Here we show a simple example of how to use the CUDA C/C++ language compiler, <code>nvcc</code>, and run code created with it. For a longer tutorial in CUDA programming, see [[CUDA tutorial]]. | ||
<!--T:41--> | |||
First, load a CUDA [[Utiliser des modules/en|module]]. | First, load a CUDA [[Utiliser des modules/en|module]]. | ||
<source lang="console"> | <source lang="console"> | ||
| Line 17: | Line 21: | ||
</source> | </source> | ||
<!--T:42--> | |||
The following program will add two numbers together on a GPU. Save the file as <code>add.cu</code>. ''The <code>cu</code> file extension is important!''. | The following program will add two numbers together on a GPU. Save the file as <code>add.cu</code>. ''The <code>cu</code> file extension is important!''. | ||
| Line 57: | Line 62: | ||
<translate> | <translate> | ||
<!--T:43--> | |||
Compile the program with <code>nvcc</code> to create an executable named <code>add</code>. | Compile the program with <code>nvcc</code> to create an executable named <code>add</code>. | ||
<source lang="console"> | <source lang="console"> | ||
| Line 62: | Line 68: | ||
</source> | </source> | ||
<!--T:44--> | |||
To run the program, create a Slurm job script as shown below. Be sure to replace <code>def-someuser</code> with your specific account (see [[Running_jobs#Accounts_and_projects|accounts and projects]]). For options relating to scheduling jobs with GPUs see [[Using GPUs with Slurm]]. | To run the program, create a Slurm job script as shown below. Be sure to replace <code>def-someuser</code> with your specific account (see [[Running_jobs#Accounts_and_projects|accounts and projects]]). For options relating to scheduling jobs with GPUs see [[Using GPUs with Slurm]]. | ||
{{File | {{File | ||
| Line 75: | Line 82: | ||
}} | }} | ||
<!--T:45--> | |||
Submit your GPU job to the scheduler with this command. | Submit your GPU job to the scheduler with this command. | ||
<source lang="console"> | <source lang="console"> | ||
| Line 81: | Line 89: | ||
</source>For more information about the <code>sbatch</code> command and running and monitoring jobs see [[Running jobs]]. | </source>For more information about the <code>sbatch</code> command and running and monitoring jobs see [[Running jobs]]. | ||
<!--T:46--> | |||
Once your job has finished you should see an output file similar to this. | Once your job has finished you should see an output file similar to this. | ||
<source lang="console"> | <source lang="console"> | ||
| Line 88: | Line 97: | ||
If you run this without a GPU present you might see output like <code>2+7=0</code>. | If you run this without a GPU present you might see output like <code>2+7=0</code>. | ||
<!--T:47--> | |||
To learn more about how the above program works and how to make the use of a GPUs parallelism see [[CUDA tutorial]]. | To learn more about how the above program works and how to make the use of a GPUs parallelism see [[CUDA tutorial]]. | ||
</translate> | </translate> | ||