[Cython] cython.parallel tasks, single, master, critical, barriers

[mark florisson]

On 9 October 2011 13:18, Dag Sverre Seljebotn
<d.s.seljebotn at astro.uio.no> wrote:
>
> On 10/09/2011 02:11 PM, mark florisson wrote:
>>
>> Hey,
>>
>> So far people have been enthusiastic about the cython.parallel features,
>> I think we should introduce some new features. I propose the following,
>
> Great!!
>
> I only have time for a very short feedback now, perhaps more will follow.
>
>> assume parallel has been imported from cython:
>>
>> with parallel.master():
>>     this is executed in the master thread in a parallel (non-prange)
>> section
>>
>> with parallel.single():
>>    same as master, except any thread may do the execution
>>
>> An optional keyword argument 'nowait' specifies whether there will be a
>> barrier at the end. The default is to wait.
>>
>> with parallel.task():
>>     create a task to be executed by some thread in the team
>>     once a thread takes up the task it shall only be executed by that
>> thread and no other thread (so the task will be tied to the thread)
>>
>>     C variables will be firstprivate
>>     Python objects will be shared
>>
>> parallel.taskwait() # wait on any direct descendent tasks to finish
>
> Regarding tasks, I think this is mapping OpenMP too close to Python. Closures are excellent for the notion of a task, so I think something based on the futures API would work better. I realize that makes the mapping to OpenMP and implementation a bit more difficult, but I think it is worth it in the long run.

Hmm, that would be cool as well. Something like parallel.submit_task(myclosure)?

The problem I see with that is that parallel stuff can't have the GIL,
and you can only have 'def' closures at the moment. I realize that you
won't actually have to use closure support here though, and could just
transform the inner function to OpenMP task code. This would maybe
look inconsistent with other closures though, and you'd also have to
restrict the use of such a closure to parallel.submit_task().

Anyway, perhaps you have a concrete proposal that addresses these problems.

>>
>> with parallel.critical():
>>     this section of code is mutually exclusive with other critical sections
>>     optional keyword argument 'name' specifies a name for the critical
>> section,
>>     which means all sections with that name will exclude each other,
>> but not
>>     critical sections with different names
>>
>>     Note: all threads that encounter the section will execute it, just
>> not at the same time
>>
>> with parallel.barrier():
>>     all threads wait until everyone has reached the barrier
>>     either no one or everyone should encounter the barrier
>>     shared variables are flushed
>>
>> Unfortunately, gcc again manages to horribly break master and single
>> constructs in loops (versions 4.2 throughout 4.6), so I suppose I'll
>> first file a bug report. Other (better) compilers like Portland (and I'm
>> sure Intel) work fine. I suppose a warning in the documentation will
>> suffice there.
>>
>> If we at some point implement vector/SIMD operations we could also try
>> out the Fortran openmp workshare construct.
>
> I'm starting to learn myself OpenCL as part of a course. It's very neat for some kinds of parallelism. What I'm saying is that at least of the case of SIMD, we should not lock ourselves to Fortran+OpenMP thinking too early, but also look forward to coming architectures (e.g., AMD's GPU-and-CPU on same die design).

Oh, definitely. The good thing is that code generation backends
needn't be that hard. If you figure all semantics out in the Python
code you could based on a backend load a different utility template as
a string. It's probably not that easy, but the point is that as long
as your code semantics don't prevent other backends, you keep your
options open.

In the end I want to be able to write a parallel program almost
serially and have Cython compile it to OpenMP, MPI, GPU's or whatever
else I need. At the same time I need to stay in touch with reality, so
it's one step at a time :)

> Dag Sverre
> _______________________________________________
> cython-devel mailing list
> cython-devel at python.org
> http://mail.python.org/mailman/listinfo/cython-devel