chasm-interop-devel Mailing List for chasm-interop

Hello

Does chasm support 64 bit code compiled with gfortran 4.4? I started looking at the chasm source code, and its not entirely clear if it does or not.

Would you know if the array descriptor changed between 32 and 64 bit versions.

thanks

Begin forwarded message:

> From: "SourceForge.net" <no...@so...>
> Date: December 17, 2007 11:47:01 PM PST
> To: no...@so...
> Subject: [chasm-interop - Help] 2D array descriptor
>
>
> Read and respond to this message at:
> https://sourceforge.net/forum/message.php?msg_id=4680454
> By: ank_psu
>
>
> Hello,
>
> I was using the syntax as given in driver.cc example  to create a 1D  
> array
> descriptor. That worked fine. However, I am facing problems in  
> creating a descriptor
> for a 2D array. Could some one please provide me with a very simple  
> example
> on creation of an array descriptor for a 2D array. I am not looking  
> for a full-blown
> example; all I need is just an idea of how that can be done.
>
> Thanks in advance !!
>
> Regards,
> Ankur
>
> ______________________________________________________________________
> You are receiving this email because you elected to monitor this  
> forum.
> To stop monitoring this forum, login to SourceForge.net and visit:
> https://sourceforge.net/forum/unmonitor.php?forum_id=174700

Sorry for the delay - I was trapped in a different project this week, and 
let this in my to-reply-to pile.

> Since Matt is on the TASCS payroll, I wonder if he would be willing to
> be the "point man" in making sure Chasm issues get addressed?  That
> doesn't necessarily mean fixing everything himself, just making sure
> that anyone who submits an issue gets some kind of response, even if
> it is just "we're on it" and that the _someone_ really does address
> the issue appropriately.

I will be willing to monitor the chasm mailing list and web site to make 
sure submissions for issues get responded to and forwarded to this list. 
The biggest issue is making sure that bug reports that are submitted to 
the chasm bug tracker don't fall into a black hole.  I'll do my best to 
babysit the trackers and forward the issues to this list to make sure they 
are at least made known to people who are willing to work on them.  As 
people who work with me know, I can nag pretty well.  :-)

Whether or not this will lead to issues being addressed is another 
question -- I only have access to a small subset of Fortran compilers 
(that number decreased greatly when I left LANL -- UO doesn't have nearly 
the number and diversity of licenses as LANL did), so I'm limited in how 
much I can do myself.  If people don't respond to my nagging, there's not 
much I can promise.

-m

Craig, thanks for your input and help.

Since TASCS/Babel are dependent on Chasm, I consider it in our
interest to insure that support requests receive appropriate
responses. The tracker issues seem to be primarily from Tom, but I
remember Victor Eijkhout recently saying that he had asked a Chasm
question (where?) and not gotten any response, so he redirected his
question to the Babel community.

Since Matt is on the TASCS payroll, I wonder if he would be willing to
be the "point man" in making sure Chasm issues get addressed?  That
doesn't necessarily mean fixing everything himself, just making sure
that anyone who submits an issue gets some kind of response, even if
it is just "we're on it" and that the _someone_ really does address
the issue appropriately.

Also, would it be useful to setup automatic periodic testing in a few
places in order to try to catch updates to as many of the supported
compilers as we can?  Jim Kohl is nominally working on testing for the
CCA tool chain and if the tests have already been written, I'm happy
to add automation and deployment of the process to Jim's plate.

It would be great to have a prioritized list of compilers, and then we
could look for places to run the tests.  I might be able to put some
on cca-forum.org if we need to. But I think we're better off, wherever
possible, trying to do it at sites with more organized maintenance of
the system software environment.
--
David E. Bernholdt                   |   Email: ber...@or... 
Oak Ridge National Laboratory        |   Phone: +1 (865) 574 3147    
http://www.csm.ornl.gov/~bernhold/   |   Fax:   +1 (865) 576 5491

On Dec 11, 2007, at 4:46 PM, Tom Epperly wrote:

>
>> 3) Assuming the F2008 stuff is an appropriate replacement, Babel  
>> needs
>>    to extended to support it -- testing the compiler for support in
>>    autoconf, and generating the appropriate code.
>>
> This assumption is completely and utterly incorrect. The  
> interoperability provided by the F2003/F2008 standards (as they  
> currently exist) is insufficient for (at least) the reasons  
> enumerated above. To be utterly clear, we cannot provide Babel  
> users the level of Fortran interoperability that they're accustomed  
> to in Babel's F90/95 binding using the features available in F2003/ 
> F2008 standard as they exist today.

To clarify the F2008 C-interoperability standard, it currently is  
just F2003 as Tome says.  The new Chasm-like additions have been  
voted into a TR and will come out later (the exact details of what it  
will look like are being debated).  However, this is all a moot point  
as F2008 won't appear in compilers for years.

Cheers,
Craig

No to F2008 and no to users outside of Babel that are a concern.

I would suggest splitting up the resources.  Perhaps Babel could help  
with the autocon stuff as it seems best to make Chasm and Babel  
builds more consistent.  Since I'm the only one who knows about the  
descriptor stuff, I can handle updating that.

Regards,
Craig


On Dec 11, 2007, at 3:52 PM, ber...@or... wrote:

> Okay, now that I can finally post here, I can suggest we try to shift
> the discussion about Chasm and standardized array interop back to this
> list, where the interested parties can see it, and where it gets
> archived for posterity.
>
> To summarize recent discussions amongst CCAers about this topic:
>
> 1) Chasm has fallen behind in support for the latest compiler versions
>    on a number of platforms, especially 64-bit.  This is an immediate
>    issue that needs to be addressed.
>
> 2) The Babel folks need to understand whether what's been proposed for
>    F2008 really covers everything they've been getting from Chasm thus
>    far. (Also, I guess the previous draft stuff that some vendors have
>    already implemented.)
>
> 3) Assuming the F2008 stuff is an appropriate replacement, Babel needs
>    to extended to support it -- testing the compiler for support in
>    autoconf, and generating the appropriate code.
>
> 4) We need for Chasm to be maintained until users upgrade to compiler
>    versions supporting the F2008 features.  Just because a site has a
>    compiler version that does support the needed features, doesn't
>    mean the users are actually using that version.  Many sites,
>    especially large centers, maintain multiple compiler versions
>    simultaneously.
>
> So I think we need to figure out an approach that covers these bases.
>
> Are there any other major users of Chasm, besides Babel/CCA, who we
> should try to draw into this discussion?
> --
> David E. Bernholdt                   |   Email: ber...@or...
> Oak Ridge National Laboratory        |   Phone: +1 (865) 574 3147
> http://www.csm.ornl.gov/~bernhold/   |   Fax:   +1 (865) 576 5491
>
> ---------------------------------------------------------------------- 
> ---
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> Chasm-interop-devel mailing list
> Cha...@li...
> https://lists.sourceforge.net/lists/listinfo/chasm-interop-devel

ber...@or... wrote:
> 2) The Babel folks need to understand whether what's been proposed for
>    F2008 really covers everything they've been getting from Chasm thus
>    far. (Also, I guess the previous draft stuff that some vendors have
>    already implemented.)
>   
I've downloaded the latest draft for the next Fortran standard 
(2007-Sep-27), and I've read the relevant section on C interoperability 
(15). It seems largely like what I read in the F2003 standard. *The 
Fortran 2008 standard as currently proposed DOES NOT provide the level 
of interoperability that Babel requires and that Chasm provides.* Even a 
complete implementation of the F2003 or F2008 standard would not be 
sufficient to provide the level of interoperability that Babel needs. In 
particular, the existing F2003 standard *and *the proposed F2008 
standard do not provide sufficient capabilities to:

   1. Allow passing array slices without copying data. For example, if I
      take a 2-D slice of a 3-D array in Babel, I can end up with an
      array where no dimension is stride one. With CHASM and Babel, I
      can pass that array back and forth between Fortran 90/95 and any
      other supported language without requiring copying. As far as I
      can see, the bind C approach requires arrays that are passed
      between Fortran and C to be dense and column-major ordered. Babel
      needs access to the internals of the array descriptor to pass
      other kinds of arrays without copying.
   2. The C_F_POINTER subroutine forces every array lower bound to be 1
      regardless of what the Babel array's lower bound is. In general,
      the bind C feature doesn't give us control of the lower and upper
      bounds like CHASM does.
   3. Note 15.17 says "A polymorphic, allocatable, or pointer array is
      never interoperable."  This seems to be a problem too because
      that's basically what Babel arrays are.


> 3) Assuming the F2008 stuff is an appropriate replacement, Babel needs
>    to extended to support it -- testing the compiler for support in
>    autoconf, and generating the appropriate code.
>   
This assumption is completely and utterly incorrect. The 
interoperability provided by the F2003/F2008 standards (as they 
currently exist) is insufficient for (at least) the reasons enumerated 
above. To be utterly clear, we cannot provide Babel users the level of 
Fortran interoperability that they're accustomed to in Babel's F90/95 
binding using the features available in F2003/F2008 standard as they 
exist today.
> 4) We need for Chasm to be maintained until users upgrade to compiler
>    versions supporting the F2008 features.  Just because a site has a
>    compiler version that does support the needed features, doesn't
>    mean the users are actually using that version.  Many sites,
>    especially large centers, maintain multiple compiler versions
>    simultaneously.
>   
We need a library, such as Chasm, to provide access to internals of the 
Fortran dope vectors, so we have direct access to the lower bound, upper 
bound, and stride information about arrays. Until a Fortran standard 
comes out and is implemented to provide that level of access to the 
internals of Fortran arrays, we will need the capabilities that Chasm 
currently provides.

Tom Epperly
Babel Architect
https://computation.llnl.gov/casc/components/

Okay, now that I can finally post here, I can suggest we try to shift
the discussion about Chasm and standardized array interop back to this
list, where the interested parties can see it, and where it gets
archived for posterity.

To summarize recent discussions amongst CCAers about this topic:

1) Chasm has fallen behind in support for the latest compiler versions
   on a number of platforms, especially 64-bit.  This is an immediate
   issue that needs to be addressed.

2) The Babel folks need to understand whether what's been proposed for
   F2008 really covers everything they've been getting from Chasm thus
   far. (Also, I guess the previous draft stuff that some vendors have
   already implemented.)

3) Assuming the F2008 stuff is an appropriate replacement, Babel needs
   to extended to support it -- testing the compiler for support in
   autoconf, and generating the appropriate code.

4) We need for Chasm to be maintained until users upgrade to compiler
   versions supporting the F2008 features.  Just because a site has a
   compiler version that does support the needed features, doesn't
   mean the users are actually using that version.  Many sites,
   especially large centers, maintain multiple compiler versions
   simultaneously.

So I think we need to figure out an approach that covers these bases.

Are there any other major users of Chasm, besides Babel/CCA, who we
should try to draw into this discussion?
--
David E. Bernholdt                   |   Email: ber...@or... 
Oak Ridge National Laboratory        |   Phone: +1 (865) 574 3147    
http://www.csm.ornl.gov/~bernhold/   |   Fax:   +1 (865) 576 5491

Sorry for the junk mail, but due to overzealous spam protectors at
SourceForge, I had been unable to post to this list.  I think it is 
fixed now.  Matt, if you receive this you please let me know?  Thanks
--
David E. Bernholdt                   |   Email: ber...@or... 
Oak Ridge National Laboratory        |   Phone: +1 (865) 574 3147    
http://www.csm.ornl.gov/~bernhold/   |   Fax:   +1 (865) 576 5491

On Fri, 19 Oct 2007, Gary Kumfert wrote:
[snip]
>
> Can we agree Chasm is entering the endgame of its lifecycle
> and that a gradual drawdown best serves Chasm's customers?
>

No response?  Okay... I try again.  One thing that could be
a big help is autoconf macros to detect if Chasm is really needed.
That way, users who install Chasm could get a warning that
Chasm-like support is already "built in" to the compiler.

Imagine how motivated us Babelers become when all our customers
say "why are you using Chasm, when Chasm itself says its redundant?"

Hint, hint.  ;)

Gary

Craig,

I think we're all agreed that:

  1.  You made important contributions to the Fortran standard
      that we all will benefit from.  Thank you!

  2.  Chasm should not be needed in the future and we should
      begin planning for a graceful retirement.

I understand if you're less interested in supporting Chasm and
would rather move on to new things.  *I* just don't want
to lose any customers by dropping Chasm like a hot potato.

Can we agree Chasm is entering the endgame of its lifecycle
and that a gradual drawdown best serves Chasm's customers?

Gary


On Fri, 19 Oct 2007, Craig Rasmussen wrote:

>
> On Oct 19, 2007, at 11:51 AM, Tom Epperly wrote:
>
>> A new F2003 binding will start using F2003 capabilities. When will retire 
>> the F77 and F90 bindings when people stop using those languages.
>
> Unlike C and C++, there is one language Fortran and as far as I know, all of 
> the major vendors have only one compiler.  What people mean when they say F77 
> is implicit interfaces with fixed format input.  Usually, F90 is used to 
> imply explicit interfaces with free format input.  One can choose to program 
> in any of these styles (including mixing implicit and explicit interfaces 
> with either fixed or free form) using any modern compiler.
>
> As I understood from Matt, the discussion was whether the Chasm hacks (which 
> seem to work well) could be dropped in favor language standards supported by 
> the compiler vendors.  I'm not certain, but I'm pretty sure that the C 
> interoperability standard with work quite well for Babel users.  I believe 
> the following is correct:
>
> 1. Support for implicit interfaces (i.e., F77 bindings) would remain the 
> same.  This should work with any compiler and compiler version.
>
> 2. Support for explicit interfaces (i.e,. F90 bindings) "could" be switched 
> to C interop with no impact on users, thus retiring Chasm.
>
> The primary reason I joined the Fortran standards committee was to get Chasm 
> capabilities into the language.  I've nearly succeeded but time will tell.
>
> Cheers,
> Craig
>

Craig Rasmussen wrote:
>
> On Oct 19, 2007, at 11:51 AM, Tom Epperly wrote:
>
>> A new F2003 binding will start using F2003 capabilities. When will 
>> retire the F77 and F90 bindings when people stop using those languages.
>>
>
> Unlike C and C++, there is one language Fortran and as far as I know, 
> all of the major vendors have only one compiler.  What people mean 
> when they say F77 is implicit interfaces with fixed format input.  
> Usually, F90 is used to imply explicit interfaces with free format 
> input.  One can choose to program in any of these styles (including 
> mixing implicit and explicit interfaces with either fixed or free 
> form) using any modern compiler. 
We have had this argument before.  When a new Fortran standard comes 
into existence, the old Fortran standards do not cease to exist. There 
are still people writing code to the old standards, and there are still 
compilers in use that only process source written the old standards. For 
example, the g77 compiler can still be compiled and installed on many 
platforms. Older versions of gfortran (that don't support F2003) will 
remain installed on machines for years to come, and people will use it.

I just got a new Linux desktop machine, and it's running  Red Hat 
Enterprise Linux WS release 4 (Nahant Update 5). CASC is just now 
rolling out RHEL WS4. Most people in CASC are still running RHEL WS3. 
The C & C++ compilers are both gcc 3.4.6 (20060404), and the Fortran 
compiler that comes preinstalled is g77 version 3.4.6 (not gfortran).

Because people still code to old standards and still use tools that only 
support old standards, we cannot stop supporting F77 and F90 (as it 
exists yesterday and today).
>
> As I understood from Matt, the discussion was whether the Chasm hacks 
> (which seem to work well) could be dropped in favor language standards 
> supported by the compiler vendors.  I'm not certain, but I'm pretty 
> sure that the C interoperability standard with work quite well for 
> Babel users.  I believe the following is correct:
>
> 1. Support for implicit interfaces (i.e., F77 bindings) would remain 
> the same.  This should work with any compiler and compiler version.
>
> 2. Support for explicit interfaces (i.e,. F90 bindings) "could" be 
> switched to C interop with no impact on users, thus retiring Chasm.
When we talked about a year or more ago, I thought we came to the 
conclusion that we couldn't use solely the C interop standard because it 
still didn't provide enough access to the details of the array 
descriptors. Babel needs the ability to make/read Fortran pointers to 
arrays with arbitrarily strided data.  Does the standard allow me to do 
that?

The way I would say this is that people could start choosing Babel's 
F2003 binding instead of F90 binding.
>
> The primary reason I joined the Fortran standards committee was to get 
> Chasm capabilities into the language.  I've nearly succeeded but time 
> will tell.
>
> Cheers,
> Craig
>

On Oct 19, 2007, at 11:51 AM, Tom Epperly wrote:

> A new F2003 binding will start using F2003 capabilities. When will  
> retire the F77 and F90 bindings when people stop using those  
> languages.

Unlike C and C++, there is one language Fortran and as far as I know,  
all of the major vendors have only one compiler.  What people mean  
when they say F77 is implicit interfaces with fixed format input.   
Usually, F90 is used to imply explicit interfaces with free format  
input.  One can choose to program in any of these styles (including  
mixing implicit and explicit interfaces with either fixed or free  
form) using any modern compiler.

As I understood from Matt, the discussion was whether the Chasm hacks  
(which seem to work well) could be dropped in favor language  
standards supported by the compiler vendors.  I'm not certain, but  
I'm pretty sure that the C interoperability standard with work quite  
well for Babel users.  I believe the following is correct:

1. Support for implicit interfaces (i.e., F77 bindings) would remain  
the same.  This should work with any compiler and compiler version.

2. Support for explicit interfaces (i.e,. F90 bindings) "could" be  
switched to C interop with no impact on users, thus retiring Chasm.

The primary reason I joined the Fortran standards committee was to  
get Chasm capabilities into the language.  I've nearly succeeded but  
time will tell.

Cheers,
Craig

Craig Rasmussen wrote:
> Matt mentioned an item in the discussion that came up that I didn't  
> understand.  He said that CCA had to support Fortran 90 codes so the  
> interoperability features now in the Intel, IBM, Cray, gfortran  
> (soon), ... compilers couldn't be used.
>   
There was still at least one person at the CCA meeting who asked us not 
to remove the FORTRAN 77 binding because people are still using it with 
old compilers (i.e., g77).

My sense of things is that there will probably be people using F90/95 
with old compilers for years to come as well.  Hence, the F90 binding 
should be targeted to todays and yesterday's F90/95 compilers rather 
than ones that have already started implementing the F2003 standard.

A new F2003 binding will start using F2003 capabilities. When will 
retire the F77 and F90 bindings when people stop using those languages.

Tom
> I know of no features that were removed by any compiler vendor in the  
> latest releases of their compilers.  If fact they all have strongly  
> stated that they will NEVER remove any feature, even if something  
> were to be removed from the standard.
>
> If I'm misunderstanding the issue, perhaps someone could kindly clue  
> me in.
>
> Cheers,
> Craig
>   

Matt mentioned an item in the discussion that came up that I didn't  
understand.  He said that CCA had to support Fortran 90 codes so the  
interoperability features now in the Intel, IBM, Cray, gfortran  
(soon), ... compilers couldn't be used.

I know of no features that were removed by any compiler vendor in the  
latest releases of their compilers.  If fact they all have strongly  
stated that they will NEVER remove any feature, even if something  
were to be removed from the standard.

If I'm misunderstanding the issue, perhaps someone could kindly clue  
me in.

Cheers,
Craig

I added everyone who had some interest in the continued life of Chasm to 
this mailing list.  If you don't want to be on the list, feel free to 
unsubscribe.  Please follow instructions in the "welcome" email you just 
got if you want to unsubscribe.  (If I see an "unsubscribe" message sent 
to the list, I will not unsubscribe you, but invite the group to make fun 
of you for not following instructions.)  Please pass on any additional 
people that you believe should be added to this list -- the current 
subscribers are: Craig, Chris, Boyana, Gary, Tom, Jarek, David, me.

This list should be relatively low traffic.  I would recommend that any 
questions that remain after the CCA meeting (such as how BIND(C) relates 
to Chasm, and what it fails to help with) be discussed sooner than later. 
Craig is on this list, so he can field anything I couldn't handle 
sufficiently in San Francisco.

-m