llvm dev - Jun 2016 - [Openmp-dev] [cfe-dev] RFC: Proposing an LLVM subproject for parallelism runtime and support libraries

On Wed, Jun 1, 2016 at 11:21 PM, Mehdi Amini <mehdi.amini at apple.com>
wrote:
>
>> On Jun 1, 2016, at 7:42 AM, C Bergström via llvm-dev <llvm-dev at
lists.llvm.org> wrote:
>>
>> I still don't see why
>> they can't fork it on github or create a project to let it bake and
>> get some users or traction.
>
> The intent may be that instead of creating a separate community of users,
it'd create its community inside the llvm projects community. That looks
like a nice "feature" long term.
>
> Cost/benefit: cost is not really existent, potential benefits are
interesting.
Not true - the cost isn't zero.

Of the top of my head...

#1 The project will mix and blend with other programming models - The
shared runtime should cover all the popular stuff and "this"

#2 User confusion - Promoting this over other more established
projects and models is just another turd in the pot

Lots of things are interesting.. lets take Legion as an example.. that
little heard of project at *least* has a bunch of examples and more
docs than SE..
https://github.com/StanfordLegion/legion/tree/stable/examples

Why do I feel so strongly - because I've been dealing with
here-today-gone-tomorrow programming models for 8 years now... It's
boring and wastes a lot of time. Research projects should incubate
outside the tree before being included with production parts.

> On Jun 1, 2016, at 8:29 AM, C Bergström <cbergstrom at pathscale.com>
wrote:
> 
> On Wed, Jun 1, 2016 at 11:21 PM, Mehdi Amini <mehdi.amini at
apple.com> wrote:
>> 
>>> On Jun 1, 2016, at 7:42 AM, C Bergström via llvm-dev <llvm-dev
at lists.llvm.org> wrote:
>>> 
>>> I still don't see why
>>> they can't fork it on github or create a project to let it bake
and
>>> get some users or traction.
>> 
>> The intent may be that instead of creating a separate community of
users, it'd create its community inside the llvm projects community. That
looks like a nice "feature" long term.
>> 
>> Cost/benefit: cost is not really existent, potential benefits are
interesting.
> 
> Not true - the cost isn't zero.
> 
> Of the top of my head...
> 
> #1 The project will mix and blend with other programming models - The
> shared runtime should cover all the popular stuff and "this"
True, but I see it as a positive things: the runtime will be designed in a way
more generic/capable.
Also, it means that we gets the SE authors/maintainers to contribute to the
runtime (otherwise the projects is dead and can be ditched), making the runtime
more "alive".

-- 
Mehdi


> 
> #2 User confusion - Promoting this over other more established
> projects and models is just another turd in the pot
> 
> Lots of things are interesting.. lets take Legion as an example.. that
> little heard of project at *least* has a bunch of examples and more
> docs than SE..
> https://github.com/StanfordLegion/legion/tree/stable/examples
> 
> Why do I feel so strongly - because I've been dealing with
> here-today-gone-tomorrow programming models for 8 years now... It's
> boring and wastes a lot of time. Research projects should incubate
> outside the tree before being included with production parts.

On Thu, Jun 2, 2016 at 12:04 AM, Mehdi Amini <mehdi.amini at apple.com>
wrote:
>
>> On Jun 1, 2016, at 8:29 AM, C Bergström <cbergstrom at
pathscale.com> wrote:
>>
>> On Wed, Jun 1, 2016 at 11:21 PM, Mehdi Amini <mehdi.amini at
apple.com> wrote:
>>>
>>>> On Jun 1, 2016, at 7:42 AM, C Bergström via llvm-dev
<llvm-dev at lists.llvm.org> wrote:
>>>>
>>>> I still don't see why
>>>> they can't fork it on github or create a project to let it
bake and
>>>> get some users or traction.
>>>
>>> The intent may be that instead of creating a separate community of
users, it'd create its community inside the llvm projects community. That
looks like a nice "feature" long term.
>>>
>>> Cost/benefit: cost is not really existent, potential benefits are
interesting.
>>
>> Not true - the cost isn't zero.
>>
>> Of the top of my head...
>>
>> #1 The project will mix and blend with other programming models - The
>> shared runtime should cover all the popular stuff and "this"
>
> True, but I see it as a positive things: the runtime will be designed in a
way more generic/capable.
> Also, it means that we gets the SE authors/maintainers to contribute to the
runtime (otherwise the projects is dead and can be ditched), making the runtime
more "alive".
What runtime? So far I haven't seen anything tangible in terms of them
designing things to be portable. It's all glued to what they have now.
They should care about this regardless of if the project is accepted,
right? They can go give feedback to OpenMP offloading now.. I did..