March 2008 - Mpi-22 - mailman3.cels.anl.gov

2.2 at April meeting?
by unknown＠example.com 31 Mar '08

31 Mar '08

Bill -- Do we plan to have any time allocated in the April meeting to MPI 2.2, or are we still in a holding pattern waiting for 2.1? If we do discuss 2.2 in April, I'd like a few minutes to discuss the whole "const" issues surrounding the C++ bindings. A few people asked me last meeting to rehash the issues, etc. In short: my position is that *all* predefined handles in the C++ bindings should be "const" (reversing some of the [incorrect, I believe] errata) and fixing a few C++ API bindings to compensate. MPI::BOTTOM is the only exception, but we have discussed that one to death already. If we push all the 2.2 discussions out to the June meeting, no problem -- I just wanted to queue up the "const" discussion for whenever the 2.2 discussions occur. -- Jeff Squyres Cisco Systems

1 0

MPI 2.2 ballots
by unknown＠example.com 19 Mar '08

19 Mar '08

Bill -- Let us know when the first round of MPI 2.2 ballots will be coming up. Some of the C++ stuff (e.g., const) got moved to MPI 2.2. There was some confusion about it at the last Chicago meeting -- I couldn't remember enough context on the spot in Chicago to remember why "const" was Good for all C++ handles, for example. I'd like to bring that discussion forward again at the appropriate time. Thanks. -- Jeff Squyres Cisco Systems

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

You have to draw the line somewhere ;-) Especially since -- assuming the user uses MPI_BYTE to describe everything -- the MPI implementation cannot know if any of those datatypes needing adjustment are used. Of course, the application could know this if detailed information about the So, as soon as one of the predefined MPI datatypes is different I would flag the peer (or communicator) as heterogen(e)ous. For example, X86 vs X86_64 and gcc: All C types, except for pointers and long integers are the same; Still, I would say they are heterogen(e)ous without giving more details. Dries * Jeff Squyres <jsquyres_at_[hidden]> [2008-03-03 13:24:11]: > My point is that some datatypes may require translation while others > may not. It's not always just endian differences -- sometimes there's > floating point representation differences, or size differences (e.g., > sizeof(int) != sizeof(int)). *

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

The simple question seems to be: Can MPI_BYTE be used safely on this communicator? If even one datatype requires translation then the answer is "no". How about just calling the per communicator attribute something like "MPI_BYTE_SAFE" and leave other debates about the meaning of heterogeneous aside? There is clearly room to add complexity but is there a real need? It seems to me that communicator attribute "MPI_BYTE_SAFE" provides 99% of the value and is easy to do. Dick Treumann - MPI Team/TCEM IBM Systems & Technology Group Dept 0lva / MS P963 -- 2455 South Road -- Poughkeepsie, NY 12601 Tele (845) 433-7846 Fax (845) 433-8363 mpi-22-bounces_at_[hidden] wrote on 03/03/2008 01:24:11 PM: > On Mar 3, 2008, at 12:54 PM, Dries Kimpe wrote: > > >> Narrowing the scope to "whether translation functionality is > >> required" > >> is good. But note that whether the translation functionality is used > >> may be specific to a given (communicator,peer,datatype) tuple. > > > > You're right that this depends on the communicator and on the peer; > > ...and the datatype. > > > It would not be uncommon to have one 'different (memory representation > > wise)' node in the communicator. This could be a login node, a > > visualisation node, ... > > > > So, from a performance point of view (and also for the mpi > > implementation > > internally), it's actually better to record this information for > > (communicator, source, dest). However, my main reason for this > > proposal > > was to enable an application that doesn't care about datatypes > > (might be > > more common than you'd think, looking at some of the mpi-subset > > ideas) to > > abort execution if it detects that data conversion is needed when > > communicating to its peers. > > > My point is that some datatypes may require translation while others > may not. It's not always just endian differences -- sometimes there's > floating point representation differences, or size differences (e.g., > sizeof(int) != sizeof(int)). > > -- > Jeff Squyres > Cisco Systems > > _______________________________________________ > Mpi-22 mailing list > Mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 *

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

On Mar 3, 2008, at 12:54 PM, Dries Kimpe wrote: >> Narrowing the scope to "whether translation functionality is >> required" >> is good. But note that whether the translation functionality is used >> may be specific to a given (communicator,peer,datatype) tuple. > > You're right that this depends on the communicator and on the peer; ...and the datatype. > It would not be uncommon to have one 'different (memory representation > wise)' node in the communicator. This could be a login node, a > visualisation node, ... > > So, from a performance point of view (and also for the mpi > implementation > internally), it's actually better to record this information for > (communicator, source, dest). However, my main reason for this > proposal > was to enable an application that doesn't care about datatypes > (might be > more common than you'd think, looking at some of the mpi-subset > ideas) to > abort execution if it detects that data conversion is needed when > communicating to its peers. My point is that some datatypes may require translation while others may not. It's not always just endian differences -- sometimes there's floating point representation differences, or size differences (e.g., sizeof(int) != sizeof(int)). -- Jeff Squyres Cisco Systems

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

* Jeff Squyres <jsquyres_at_[hidden]> [2008-03-03 12:08:47]: > > I think the proposal has merit but we need to be specific that only > > data representation conversion for data transfer routines is > > involved in the meaning of "heterogeneous". > Narrowing the scope to "whether translation functionality is required" > is good. But note that whether the translation functionality is used > may be specific to a given (communicator,peer,datatype) tuple. You're right that this depends on the communicator and on the peer; It would not be uncommon to have one 'different (memory representation wise)' node in the communicator. This could be a login node, a visualisation node, ... So, from a performance point of view (and also for the mpi implementation internally), it's actually better to record this information for (communicator, source, dest). However, my main reason for this proposal was to enable an application that doesn't care about datatypes (might be more common than you'd think, looking at some of the mpi-subset ideas) to abort execution if it detects that data conversion is needed when communicating to its peers. Greetings, Dries *

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

* Richard Treumann <treumann_at_[hidden]> [2008-03-03 12:01:12]: > The real concern here is: > Does the MPI implementation need to provide data conversion services > between any pair of tasks in the communiator? If I have some tasks on a > slow node and other tasks on a fast one we could debate whether that is > heterogeneous. > I think the proposal has merit but we need to be specific that only data > representation conversion for data transfer routines is involved in the > meaning of "heterogeneous". This is what I meant: that's why I talked about not specifying the correct datatype, but instead using MPI_BYTE and relying on the fact that the in-memory representation on both ranks is the same. So, my definition of 'heterogenous': not having the same in-memory data representation. Greetings, Dries > > > Some applications are not written with heterogenous environments in > > > mind; > > > They do not (always) use correct datatype descriptions when sending or > > > receiving data, but instead treat the data as an array of bytes, > > > relying on all datatypes having the same memory representation on both > > > sender and receiver. Most often, this is done to avoid the added > > > complexity > > > of creating the correct datatypes. *

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

On Mar 3, 2008, at 12:01 PM, Richard Treumann wrote: > The real concern here is: > > Does the MPI implementation need to provide data conversion services > between any pair of tasks in the communiator? If I have some tasks > on a slow node and other tasks on a fast one we could debate whether > that is heterogeneous. > > I think the proposal has merit but we need to be specific that only > data representation conversion for data transfer routines is > involved in the meaning of "heterogeneous". > Narrowing the scope to "whether translation functionality is required" is good. But note that whether the translation functionality is used may be specific to a given (communicator,peer,datatype) tuple. -- Jeff Squyres Cisco Systems

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

The real concern here is: Does the MPI implementation need to provide data conversion services between any pair of tasks in the communiator? If I have some tasks on a slow node and other tasks on a fast one we could debate whether that is heterogeneous. I think the proposal has merit but we need to be specific that only data representation conversion for data transfer routines is involved in the meaning of "heterogeneous". Dick Treumann - MPI Team/TCEM IBM Systems & Technology Group Dept 0lva / MS P963 -- 2455 South Road -- Poughkeepsie, NY 12601 Tele (845) 433-7846 Fax (845) 433-8363 mpi-22-bounces_at_[hidden] wrote on 03/03/2008 11:09:13 AM: > I think that coming up with a precise definition for "heterogeneous" > could be problematic... > > > On Mar 2, 2008, at 7:01 AM, Dries Kimpe wrote: > > > Below: proposal for MPI-2.2 that adds support for querying if a > > communicator is heterogenous. > > > > ---------------------------------------------------------------------------- > > Proposal: > > ---------------------------------------------------------------------------- > > > > Provide some method to determine if a communicator is heterogenous. > > > > There are a number of different possibilities to provide this > > capability: > > > > 1) Provide a predefined integer valued attribute which can be used > > to query a > > communicator. (for example MPI_HETEROGENOUS) > > > > -- or -- > > > > 2) Create a seperate function: MPI_Comm_is_heterogenous (MPI_Comm > > comm, > > int * flag); > > > > ---------------------------------------------------------------------------- > > Rationale: > > ---------------------------------------------------------------------------- > > > > Currently, MPI-2.0 does not provide a portable way for an > > application to > > determine if it is running in a heterogenous environment. > > > > Some applications are not written with heterogenous environments in > > mind; > > They do not (always) use correct datatype descriptions when sending or > > receiving data, but instead treat the data as an array of bytes, > > relying on all datatypes having the same memory representation on both > > sender and receiver. Most often, this is done to avoid the added > > complexity > > of creating the correct datatypes. > > > > Although the pack/unpack functions provide an alternative, the do > > not have > > the same memory requirements (need an extra buffer to receive the > > data) or > > performance characteristics (need an extra copy). > > > > Adding a way for an application to test if it is currently running > > in a > > heterogenous environment enables it take appropriate action (aborting, > > switching to type-safe functions, ...) > > > > The MPI implementation -- if it supports heterogenous environments -- > > already needs to determine this information because it is > > responsible for > > performing type conversions in heterogenous communicators. > > > > Providing this information on a per-communicator base instead of > > returning > > it for the whole MPI_UNIVERSE / MPI_COMM_WORLD enables both the MPI > > implementation and the user application to avoid overhead > > in case of a homogenous communicator that is a subset of an > > inhomogenous > > MPI_COMM_WORLD. > > > > ---------------------------------------------------------------------------- > > Alternative ways to get the same information without modifying the > > standard: > > ---------------------------------------------------------------------------- > > > > * Store information about the architecture when compiling the > > application; > > Compare this information at runtime with all other members of the > > communicator. > > > > * (not 100% correct): calculate at run time the size of a number of > > elementary datatypes, compare information with the other ranks in the > > communicator. > > > > _______________________________________________ > > Mpi-22 mailing list > > Mpi-22_at_[hidden] > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 > > > -- > Jeff Squyres > Cisco Systems > > _______________________________________________ > Mpi-22 mailing list > Mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 *

1 0

Re: [Mpi-22] determine if running in a heterogenous environment
by unknown＠example.com 03 Mar '08

03 Mar '08

I think that coming up with a precise definition for "heterogeneous" could be problematic... On Mar 2, 2008, at 7:01 AM, Dries Kimpe wrote: > Below: proposal for MPI-2.2 that adds support for querying if a > communicator is heterogenous. > > ---------------------------------------------------------------------------- > Proposal: > ---------------------------------------------------------------------------- > > Provide some method to determine if a communicator is heterogenous. > > There are a number of different possibilities to provide this > capability: > > 1) Provide a predefined integer valued attribute which can be used > to query a > communicator. (for example MPI_HETEROGENOUS) > > -- or -- > > 2) Create a seperate function: MPI_Comm_is_heterogenous (MPI_Comm > comm, > int * flag); > > ---------------------------------------------------------------------------- > Rationale: > ---------------------------------------------------------------------------- > > Currently, MPI-2.0 does not provide a portable way for an > application to > determine if it is running in a heterogenous environment. > > Some applications are not written with heterogenous environments in > mind; > They do not (always) use correct datatype descriptions when sending or > receiving data, but instead treat the data as an array of bytes, > relying on all datatypes having the same memory representation on both > sender and receiver. Most often, this is done to avoid the added > complexity > of creating the correct datatypes. > > Although the pack/unpack functions provide an alternative, the do > not have > the same memory requirements (need an extra buffer to receive the > data) or > performance characteristics (need an extra copy). > > Adding a way for an application to test if it is currently running > in a > heterogenous environment enables it take appropriate action (aborting, > switching to type-safe functions, ...) > > The MPI implementation -- if it supports heterogenous environments -- > already needs to determine this information because it is > responsible for > performing type conversions in heterogenous communicators. > > Providing this information on a per-communicator base instead of > returning > it for the whole MPI_UNIVERSE / MPI_COMM_WORLD enables both the MPI > implementation and the user application to avoid overhead > in case of a homogenous communicator that is a subset of an > inhomogenous > MPI_COMM_WORLD. > > ---------------------------------------------------------------------------- > Alternative ways to get the same information without modifying the > standard: > ---------------------------------------------------------------------------- > > * Store information about the architecture when compiling the > application; > Compare this information at runtime with all other members of the > communicator. > > * (not 100% correct): calculate at run time the size of a number of > elementary datatypes, compare information with the other ranks in the > communicator. > > _______________________________________________ > Mpi-22 mailing list > Mpi-22_at_[hidden] > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-22 -- Jeff Squyres Cisco Systems

1 0