NVIDIA NVPTX Port Added To GCC

[phoronix]

Phoronix: NVIDIA NVPTX Port Added To GCC

The NVPTX back-end has been committed to GCC 5 as part of the process for offloading support to NVIDIA graphics processors from the compiler...

http://www.phoronix.com/vr.php?view=MTgzNTM

[wikinevick]

Forcing binary blobs to developers.

There is no doubt things like this wouldn't have found their way into gcc had it not been for the fierce competition offered by LLVM. All in all, it is good for end-users but not good for freedom as in Stallman.

Of course since Stallman doesn't write code anymore his opinions are little more than advisory and it's probably better this way before the grumpy man rolls out GPLv4.

[vadix]

Please come faster OpenACC...

[mark45]

All in all, it is good for end-users.

How many devs plan to write for it? I guess practically zero.

[adakite]

How many devs plan to write for it? I guess practically zero.

You underestimate the potential of this in academia and scientific world, where people do not focus on optimization from low-level language but rather on solving complex systems. Devs will keep coding in OCL/Cuda for sure. But they are not alone that wants to take advantage of GPGPU and few are able to port their existing apps/codes into CUDA/OCL. Clearly, having OpenACC and derivative in CLang and/or GCC and call it like OpenMP is a really good news.