cc_staff
782
edits
OpenACC: Difference between revisions
Jump to navigation
Jump to search
Forgot to change the section title
(Removed incomplete but still duplicated information) |
(Forgot to change the section title) |
||
| Line 3: | Line 3: | ||
OpenACC makes it relatively easy to offload vectorized code to accelerators such as GPUs, for example. Unlike [[CUDA]] and OpenCL where kernels need to be coded explicitly, OpenACC minimizes the amount of modifications to do on a serial or [[OpenMP]] code. The compiler converts the OpenACC code into a binary executable that can make use of accelerators. The performance of OpenACC codes can be similar to the one of a [[CUDA]] code, except that OpenACC requires less code development. | OpenACC makes it relatively easy to offload vectorized code to accelerators such as GPUs, for example. Unlike [[CUDA]] and OpenCL where kernels need to be coded explicitly, OpenACC minimizes the amount of modifications to do on a serial or [[OpenMP]] code. The compiler converts the OpenACC code into a binary executable that can make use of accelerators. The performance of OpenACC codes can be similar to the one of a [[CUDA]] code, except that OpenACC requires less code development. | ||
= OpenACC | = OpenACC directives = | ||
Similar to [[OpenMP]], OpenACC can convert a for-loop into parallel code that would run on an accelerator. This can be achieved with compiler instructions <code>#pragma acc ...</code> before a for-loop. All <code>pragma</code> directives are described in the [https://www.openacc.org/specification OpenACC specification]. | Similar to [[OpenMP]], OpenACC can convert a for-loop into parallel code that would run on an accelerator. This can be achieved with compiler instructions <code>#pragma acc ...</code> before a structured block of code like a for-loop. All <code>pragma</code> directives are described in the [https://www.openacc.org/specification OpenACC specification]. | ||
= Code examples = | = Code examples = | ||