Question on co-arrays

[Steven_L_Intel1]

Moved to separate thread

[Yuri1]

Dear Steve, when Intel will include in Fortran syntax coarray?

[Steven_L_Intel1]

Co-arrays are at the top of our development list, currently, but it's too early to predict when they'll appear in the product. On what type of system would you envision using co-arrays? Single-processor multicore system, multiprocessor system, and/or cluster?

[Yuri1]

Dear Steve,

excuse me for a delay with the answer ("has lost" thread).

Now I use coarray in cluster.

If there will be an opportunity to get multicore a computer there will be a system multicore and cluster.

Unfortunately, COarray CONsol (cocon-author Andrew Vaught, the project g95 - www.g95.org),

which I use, does not realize the opportunities offered by authors coarray (R.Numrich and J.Reid).

How much I know, standard Fortran 2008 (I hope, together with coarray) are going to accept in 2010. Coarray is in development.

Soon there should be 2-nd version coarray (CAF-2) at university Rice.

I the beginner in Linux, but shall try to collect necessary libraries if CAF-2 will appear in free access.

I am very glad, that coarray - at the top of development Intel.

Prompt, who in Intel has an operational experience with coarray for cluster?

Yours faithfully, Yuri

[Steven_L_Intel1]

Yuri,

Intel does not yet have a coarray implementation. We do have extensive cluster experience.

[Yuri1]

Thanks for attention!

[ypapadimitriou]

Hello everyone,

Does anyone has an update on Intel's plans and timeline for adding coarrays to Visual Development Studio? From what I read, Fortran 2008 is practically finished and only some minor changes (e.g. typos) are pending.

Thank you in advance,

Yanni.

[Steven_L_Intel1]

We are currently implementing the coarray feature from F2008, along with a few other F2008 features (plus most (but not all) of the F2003 features we are currently missing.) Our next major release is planned for late 2010.

[msbriggs]

May we suggest priorities for other features?

A F2003 feature, specifying lower bounds on pointer assignment, such as:
p(0:,0:) => a
This will be particularly useful with c_f_pointer, which doesn't allow specification of the bounds of the Fortran pointer, only the shape.

A F2008 feature, "newunit":
integer :: AutoUnitNum
open (newunit = AutoUnitNum, .....
It is archaic to have to manage unit numbers.

[Steven_L_Intel1]

I expect both of those to be in the late-2010 release.