Nvidia goes to Ada

I don't know what's worse: the fact that you linked a URL from your hard drive, or the fact that it's a Windows file URL.

Yep, it is -- but I popped the correct link.


Agreed -- I remember being *SEVERELY* dissapointed with CUDA because it was essentially manual insertion of GPU primitives, not significantly different from inline-assembly, rather than using Ada and taking advantage of Task and Protected types/objects.

I mean, it would have been really nice to use a implementation-pragma like:
Pragma CUDA( Some_Task );

And have the compiler pop the task onto the GPU automattically or issue a warning that it can't and use the normal CPU -- you also get the advantage of the programs being portable to other Ada compilers (though w/o the GPU support, obviously) AND when you get better GPU/compiler tech you can increase the capabilities w/o impacting extant code-bases.

*shrug*
At work I have a Windows box, a Solaris box, a VxWorks box and [in storage] a Mac.

Write in Forth, homos.

That's one thing that I really hate about C: it's *SO* fragile on those compiler-switches/build-environment. It really makes me appriciate when I (eg) took a non-trivial 30 year old Ada program written for a different compiler on a different archetecture with about a half-dozen changes (mostly because I was using an Ada 2012 compiler, and the source had used as an identifire what became new keywords).


I love Forth -- I am thinking about using it as the backend/emition for a compiler just to leverage Forth's porting power when bootstrapping the compiler.

Attached: 2c4aa2ebc35d40cafcc68603eb407702bbdd2dc29efc87d06a9c48c4dffb311f.png (500x584, 123.73K)

Remember that Paki rape gang taximan wanting his victims to call him like that?

Yes, we know.


No, and I don't want to.

Attached: 1490905674746.jpg (1280x720, 195.49K)

No. Why would I be frustrated?