what parallel C language does MIPS Pro C Compiler support?

[ramyach]

Hi friends,
I need to write a parallel code in 'C' on the server that is
running SGI Irix 6.5. This server supports MIPS Pro C compiler. I don't
have any idea of parallel C languages. I looked into few posts in this
group. I could make out that there are several languages for parallel
programming and parallel C is one of them. I need to know if this is
supported by MIPS Pro C Compiler. Or are there any other parallel C
languages that have this feature?
It would be more helpful if someone explains the differences
among mpC, paralle C, parallel C in OpenMP and MPI. To which language
does the following directives belong to.
#pragma parallel
#pragma pfor
#pragma synchronize

Thanks in advance
Ramya

 

[Walter Roberson]

I need to write a parallel code in 'C' on the server that is
running SGI Irix 6.5.

You should probably take this question to comp.sys.sgi.misc

This server supports MIPS Pro C compiler. I don't
have any idea of parallel C languages. I looked into few posts in this
group. I could make out that there are several languages for parallel
programming and parallel C is one of them. I need to know if this is
supported by MIPS Pro C Compiler.

No, the MIPSpro compilers do not support the Parallel C language.

Or are there any other parallel C
languages that have this feature?

[OT]

Here are some SGI links for you:

"MIPSpro Auto-Parallelizing Option Programmer's Guide"

http://techpubs.sgi.com/library/tpl...&db=bks&cmd=toc&pth=/SGI_Developer/MPro_AP_PG

You probably don't want to use that, though, as it is noticable
extra cost. It's useful when you first start out, but once you see
what kind of transformations it makes to your code, it is usually
easier to put in the directives manually.


"C Language Reference Manual"

"Chapter 10. Multiprocessing Directives"
http://techpubs.sgi.com/library/tpl...SGI_Developer/CLanguageRef/sgi_html/ch10.html

"Chapter 11. Multiprocessing Advanced Features"
http://techpubs.sgi.com/library/tpl...SGI_Developer/CLanguageRef/sgi_html/ch11.html

In other words, you can put the directives in manually.
Includes #pragma parallel, #pragma pfor, and #pragma synchronize .
These are not from different languages: they are all pragmas
that SGI uses to mark different aspects of parallel programming.

These are supported by SGI's C, and C++. SGI's F77 (Fortran 77)
multiprocessing directive support is documented at
http://techpubs.sgi.com/library/tpl...e=/SGI_Developer/MproF77_PG/sgi_html/apb.html

and F77's OpenMP directive support is documented at
http://techpubs.sgi.com/library/tpl...=/SGI_Developer/MproF77_PG/sgi_html/ch05.html

The F90 (Fortran 90) OpenMP support is documented at
http://techpubs.sgi.com/library/tpl...GI_Developer/MPro7F90CD_RM/sgi_html/ch04.html


"Message Passing Toolkit: MPI Programmer's Manual"

http://techpubs.sgi.com/library/tpl...&db=bks&cmd=toc&pth=/SGI_Developer/MPT_MPI_PM


"Message Passing Toolkit: PVM Programmer's Guide"

http://techpubs.sgi.com/library/tpl...&db=bks&cmd=toc&pth=/SGI_Developer/MPT_PVM_PM


Various SGI system manual pages:

mpconf - multiprocessing configuration
http://techpubs.sgi.com/library/tpl...&fname=/usr/share/catman/p_man/cat3c/mpconf.z

sysmp - multiprocessing (and realtime) controls
http://techpubs.sgi.com/library/tpl.../usr/share/catman/p_man/cat2/standard/sysmp.z

pthreads - introduction to POSIX threads
http://techpubs.sgi.com/library/tpl/cgi-bin/getdoc.cgi?cmd=getdoc&coll=0650&db=man&fname=5 pthreads

 

[Peter Nilsson]

Hi friends,
I need to write a parallel code in 'C' on the server that is
running SGI Irix 6.5. This server supports MIPS Pro C compiler. I don't
have any idea of parallel C languages.

Neither does standard C.

http://www.ungerhu.com/jxh/clc.welcome.txt

 

[Jordan Abel]

Neither does standard C.

http://www.ungerhu.com/jxh/clc.welcome.txt

That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

If clc doesn't have a charter, don't you think it needs one?

 

[Simon Biber]

http://www.ungerhu.com/jxh/clc.welcome.txt

That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

If clc doesn't have a charter, don't you think it needs one?

Indeed, clc predates the introduction of newsgroup charters. It has
always got along fine without one. The regular posters define what is
allowable and not allowable.

 

[Keith Thompson]

That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

There is none.

[...]

If clc doesn't have a charter, don't you think it needs one?

I'm not sure that it would be practical to create one and have it
accepted by -- well, by whoever would need to accept it for it to
become "official". We seem to have a general (though not universal)
consensus about what's topical.

 

[Mark McIntyre]

That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

There isn't one - CLC predates the existence of the charter scheme.
The welcome message, FAQ, other regularly posted notes and topicality
guidelines provided by regulars constitute the equivalent of a
charter.

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

This is merely the group description.

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

You're mistaken. The nonexistence of a written consitution doesn't
prevent the group having one, any more than it prevented the UK from
having one for at least a millenium longer than CLC.

If clc doesn't have a charter, don't you think it needs one?

No.

 

[Jordan Abel]

http://www.ungerhu.com/jxh/clc.welcome.txt

That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

If clc doesn't have a charter, don't you think it needs one?

Indeed, clc predates the introduction of newsgroup charters.

And also that of Standard C.

Google's earliest clc message is dated 5 November 1986 - Though
ironically, it quotes a draft version of the standard, 86-017 to be
precise.

My question is, _why_ is only standard C on-topic. I got to thinking
about this because a recent crosspost between clc and comp.std.c seemed
to have the c.s.c-ers thinking that something [which is _clearly_
off-topic here] would be on-topic for clc, and, oddly, there was at
least one voice for the reverse [i.e. that the same thread would be
on-topic for csc and not clc]

which leaves open the question of just _WHERE_ such a thing _WOULD_ be
on-topic. I believe the specific issue was printf extensions.

It has always got along fine without one. The regular posters define
what is allowable and not allowable.

The problem is that without a _clear_ and _agreed-on_ definition of what
is allowed, you'll always have dissent. And is it really _right_ That
only "ISO Certified 98.99% Pure" topics should be discussed here? Who
decided that? Was it a consensus that people actually agreed on, or did
one person just make the claim and people assumed he was right, and it
has developed into a convention?

 

[Kenny McCormack]

The problem is that without a _clear_ and _agreed-on_ definition of what
is allowed, you'll always have dissent. And is it really _right_ That
only "ISO Certified 98.99% Pure" topics should be discussed here? Who
decided that? Was it a consensus that people actually agreed on, or did
one person just make the claim and people assumed he was right, and it
has developed into a convention?

Who just wandered into a mosque and asked "Why is Christianity O/T here?" ?

 

[jacob navia]

http://www.ungerhu.com/jxh/clc.welcome.txt


That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

If clc doesn't have a charter, don't you think it needs one?

Indeed, clc predates the introduction of newsgroup charters.

And also that of Standard C.

Google's earliest clc message is dated 5 November 1986 - Though
ironically, it quotes a draft version of the standard, 86-017 to be
precise.

My question is, _why_ is only standard C on-topic. I got to thinking
about this because a recent crosspost between clc and comp.std.c seemed
to have the c.s.c-ers thinking that something [which is _clearly_
off-topic here] would be on-topic for clc, and, oddly, there was at
least one voice for the reverse [i.e. that the same thread would be
on-topic for csc and not clc]

which leaves open the question of just _WHERE_ such a thing _WOULD_ be
on-topic. I believe the specific issue was printf extensions.

It has always got along fine without one. The regular posters define
what is allowable and not allowable.

The problem is that without a _clear_ and _agreed-on_ definition of what
is allowed, you'll always have dissent. And is it really _right_ That
only "ISO Certified 98.99% Pure" topics should be discussed here? Who
decided that? Was it a consensus that people actually agreed on, or did
one person just make the claim and people assumed he was right, and it
has developed into a convention?

I am a regular poster here and I have a different view.

This group should discuss the C language, not a C ISO 89 subset.
This means that questions like extensions, new developments,
critics of the language, are on topic here.

 

[Jordan Abel]

Jordan Abel wrote:


http://www.ungerhu.com/jxh/clc.welcome.txt


That document does not claim to have the 'legal' status of a newsgroup
charter. Anyone know where the comp.lang.c charter can be found?

The best I can find online is
<ftp://ftp.uu.net/usenet/control/comp/comp.lang.c>
which contains a single-line description, "Discussion about C"

[When i look on google it looks like i'm opening a whole can of worms by
asking this, but it has to be said. Claims on a newsgroup that something
is forbidden as off-topic make an implicit claim that the charter says
so, which seems at best misleading and at worst dishonest.]

If clc doesn't have a charter, don't you think it needs one?

Indeed, clc predates the introduction of newsgroup charters.

And also that of Standard C.

Google's earliest clc message is dated 5 November 1986 - Though
ironically, it quotes a draft version of the standard, 86-017 to be
precise.

My question is, _why_ is only standard C on-topic. I got to thinking
about this because a recent crosspost between clc and comp.std.c seemed
to have the c.s.c-ers thinking that something [which is _clearly_
off-topic here] would be on-topic for clc, and, oddly, there was at
least one voice for the reverse [i.e. that the same thread would be
on-topic for csc and not clc]

which leaves open the question of just _WHERE_ such a thing _WOULD_ be
on-topic. I believe the specific issue was printf extensions.

It has always got along fine without one. The regular posters define
what is allowable and not allowable.

The problem is that without a _clear_ and _agreed-on_ definition of what
is allowed, you'll always have dissent. And is it really _right_ That
only "ISO Certified 98.99% Pure" topics should be discussed here? Who
decided that? Was it a consensus that people actually agreed on, or did
one person just make the claim and people assumed he was right, and it
has developed into a convention?

I am a regular poster here and I have a different view.

This group should discuss the C language, not a C ISO 89 subset.
This means that questions like extensions, new developments,
critics of the language, are on topic here.

I move that we have a vote. Someone post a RFD or something?

 

[Walter Roberson]

I move that we have a vote. Someone post a RFD or something?

This is a really bad time to run an RFD. The entire RFD / CFV mechanism
is undergoing a -considerable- shakeup, and no RFDs are being
accepted for now (and possibly not for quite a number of months.)

The quick summary is that the people who administer the official
mechanisms at the moment want out and have announced their firm
commitment to resign, and are pushing for a complete rework of the
structure in order to have a viable organization to hand the reigns
over to. news.groups has the discussions.

 

[Mark McIntyre]

And also that of Standard C.

My question is, _why_ is only standard C on-topic.

Its an interesting question. The answer is very simple - because that
what it is. There's no better answer, over the last 2 decades it has
generally been agreed that the purpose of CLC is to discuss the use of
standard C.

I got to thinking
about this because a recent crosspost between clc and comp.std.c seemed
to have the c.s.c-ers thinking that something [which is _clearly_
off-topic here] would be on-topic for clc, and, oddly, there was at
least one voice for the reverse [i.e. that the same thread would be
on-topic for csc and not clc]

Yes, there are sometimes things which do fall into that interesting
camp.

The problem is that without a _clear_ and _agreed-on_ definition of what
is allowed, you'll always have dissent.

The point you're missing is that it /does/ have a clear and agreed on
defintion. Its just not written down.

This isn't uncommon - I mentioned the constitution of the UK earlier.
We seem to have managed ok for several hundred years.

And is it really _right_ That
only "ISO Certified 98.99% Pure" topics should be discussed here? Who
decided that? Was it a consensus that people actually agreed on, or did
one person just make the claim and people assumed he was right, and it
has developed into a convention?

Why do you actually need to know? The point is, the topic /is/
defined. If you want to change it, feel free to try.

 

[Mark McIntyre]

to the best of my knowledge, Jacob is in a minority of at best one, at
worst a handful, amongst the regulars.

I move that we have a vote. Someone post a RFD or something?

Not interested.

If you want to change the topic, go form your own group, stop
buggering about with ones that have got along quite happily for
decades.

 

[Malcolm]

I move that we have a vote. Someone post a RFD or something?

For a long time it made perfect sense to discuss only ANSI C.
Now we've got two developments. It has become obvious that C99 is unlikely
to ever be widely implemented, which changes the status of ANSI vis a vis
the language, and parallel programming is going to become a lot more
important in the near future.

Parallel programming will probably rejuvenate C because of the problems
inherent in trying to pass objects and other high-level data structures
between processes. That is something the newsgroup will probably welcome.

However exactly how to redefine topicality is a good question - there is no
point regs engaging in flame wars with each other about what is and what
isn't on-topic.

 

[Mark McIntyre]

However exactly how to redefine topicality is a good question - there is no
point regs engaging in flame wars with each other about what is and what
isn't on-topic.

The simplest solution is to create a new group comp.lang.nonstandard-c
or something like that. If it proves useful and informative, people
will move over to it, and CLC will die away. If it proves useless, and
uninformative people will return to CLC.

 

[=?ISO-8859-1?Q?Bj=F8rn_Augestad?=]

The simplest solution is to create a new group comp.lang.nonstandard-c
or something like that.

Good idea, how about comp.lang.posix.c?

Bjørn

[snip]

 

[pete]

Good idea, how about comp.lang.posix.c?

I think comp.unix.programmer
would be a better name to call that one.

http://www.ungerhu.com/jxh/clc.welcome.txt

 

[Randy Howard]

Malcolm wrote
(in article

For a long time it made perfect sense to discuss only ANSI C.

Or ISO C 90.

Now we've got two developments. It has become obvious that C99 is unlikely
to ever be widely implemented,

That's mostly because it offers almost nothing of value to
convince people to do the work to adopt it, and it directly
conflicts with widespread extensions, most notably those of gcc,
which means it was stillborn the day it was published, whether
they realized it at the time or not.

which changes the status of ANSI vis a vis
the language,

Forget about ANSI, it's not even in the picture anymore. Blame
the current state of affairs on ISO and lack of understanding of
how reality meets the dream of a standards group.

and parallel programming is going to become a lot more
important in the near future.

It already is, and has been for quite some time.

Parallel programming will probably rejuvenate C because of the problems
inherent in trying to pass objects and other high-level data structures
between processes. That is something the newsgroup will probably welcome.

Parallel programming today pretty much ignores passing between
processes (unless on different systems over the wire), and
focuses on data use (and sharing) within a single process and
multiple threads. C with extensions is one way this is done
commonly, however if that is your bag, then you should be
discussing it in a forum filled with people that focus on it
actively, such as those in comp.programming.threads.

There is no reason to drag a group into that sort of discussion
when a better one already exists. Just as there is no reason to
discuss high end graphics programming here when there are better
forums.

Just as monolithic design is bad in most software engineering,
it's also bad in Usenet group scope. I suspect that more often
than not this desire to lump everything into clc is due to not
wanting to read multiple newsgroups rather than some well
thought out rational reason.

there is no
point regs engaging in flame wars with each other about what is and what
isn't on-topic.

That much is certainly true.

 

[Kenny McCormack]

Not interested.

If you want to change the topic, go form your own group, stop
buggering about with ones that have got along quite happily for
decades.

Do you even come close to realizing how much like either a spoiled child
or, (worse) a religious fundie, you sound like?