Is there a language that can take advantage of multicore right off the bat without too much bullshit? No CUDA bullshit plox
Is there a language that can take advantage of multicore right off the bat without too much bullshit...
Angel Rogers
Other urls found in this thread:
Nathaniel Green
Rust
Haskell
C with OpenMP
C with OpenCL
Jonathan Collins
user has a decent list, but also consider C/C++ with pthreads or MPI if you need parallelism in a cluster.
Nathaniel Adams
>C with MPI
>not too much bullshit
Tyler Jones
What do you mean with "right off the bat"?
You'll have to write your program with multiple threads which ever language you use.
Ryder Rogers
Cuda isnt real cores
Nicholas Hall
Sorry, no easy way out of this.
There isn't a framework that will adapt heavy parallelism to your brain, *you* have to adapt your brain to heavy parallelism.
Then of course there are various languages/utilities that can abstract/automate part of the boring aspects.
Christian Stewart
Go with go routines
C with OpenMP is cool for shared mem
C with MPI is cool for distributed mem
Austin Carter
Those and most BEAM VM languages. However, applies.
Aiden Kelly
Erlang, Go, Rust, C (openMP), pthreads etc. etc. there is alot of choices out there.
David Johnson
I mean that I want to write a function and then tell it to use 100% of my resources until the problem is solved.
Andrew Harris
Pseudo code:
function some_operation(x) {
return x^2;
}
huge_data_set_or_event_loop = [...]
cpu.start(some_operation, huge_data_set_or_event_loop).full_power()
Noah Edwards
>mfw it's 2018 and Erlang are literally removing Uniprocessor support from BEAM
>and Nodefags want to tell me about muh concurrency
Landon Wright
Still, it needs to be a problem that can be divided over however many cores you have available. Can you process every value in the set independently?
It's probably easiest to use a high-level language. I know in C# you can stick everything in a Task object and all the resource management is taken care of. some_operation in its current form isn't intensive enough to make the overhead of Tasks worth it, but you could divide huge_data_set into as many parts as you have cores and feed those into Tasks. That still won't reach 100% efficiency, but it's close enough.
Wyatt Thompson
Most languages have second-class support for parallelism, in the sense that it has to be explicit, even if certain languages make it easier than others.
However, there are a few languages with first-class support for parallelism, for example Chapel:
chapel-lang.org
Luke James
What you want is Java actually
list.stream().parallel().map(item -> doSomeShit(item)).collect(Collectors.toList());
Wyatt Rodriguez
if you would like a copy of Parallel and Concurrent Programming in Haskell I have epub, mobi, or pdf I can post.
Easton Gray
If bash is acceptable, xargs and parallel are a good start
Chase Rodriguez
Could you do it? I'd be interested :)
Christopher Richardson
it's part of the functional programming humble bundle, would you like just that book form it or should i post the whole bundle? I think I've got a few of the recent programming bundles hosted online in tar files
Brayden Smith
Is it easy to make my whole program run multicore/multithreaded in rust? Want to make a game with it and obviously try to squeeze out as much performance as I can
Chase Hill
Erlang is literally made for that