SHMEM Technical Note for Fortran, Cray Research, Oct. 1994 Home/Search Document Not in Database Summary Related Articles Check |
This paper is cited in the following contexts:
Automatic Parallelization for Non-cache Coherent Multiprocessors - Paek, Padua (1996) (Correct)
....time the program references them. This significantly increases average memory latency and network contention. To avoid these inefficiencies it is necessary to explicitly control caching and data transfer. We plan to do so in future versions of the translator. To this end we will use the SHMEM [21, 22] communication library which contains various single sided communication primitives, such as PUT and GET, and explicit cache control routines. For example, SHMEM enables message aggregation which is not possible in ordinary CRAFT programs. The PUT GET communication [19] allows asynchronous ....
SHMEM Technical Note for Fortran, Cray Research, Oct. 1994
Access Regions: Toward a Powerful Parallelizing Compiler - Paek, Hoeflinger, Padua (1996) (Correct)
....the Region Processor would report that it is unique under the predicate N 2 . The region test can then produce code which will check the predicate at run time, choosing between a parallel version of the loop and a serial version. 4. 2 Communication Analysis Single sided communication protocols [18, 19, 23, 27] in the form of PUT GET primitives have been rapidly gaining wide acceptance. A great advantage of PUT GET primitives is that their use of asynchronous data communication works well with the shared memory programming paradigm, which is also assumed by Polaris. PUT GETs are useful for removing ....
SHMEM Technical Note for Fortran, Cray Research, Oct. 1994
A Software-level Cray T3D Emulation Package for SGI.. - Sawdey, O'Keefe (1995) (Correct)
....do strided access an scatter gather operations) mechanisms to create Fortran pointers to remote objects, barrier and eureka synchronizations, cache management routines, and routines to take advantage of the T3D s atomic swap and prefetch hardware. The library we emulate now is called libsma (see [2]) which is a standard system library on the Cray T3D that provides many of the functions in libfminus. 3 Implementation We have implemented a subset of the libsma library. The SGI implementation consists of two pieces: a driver program that creates multiple processes and sets up the address ....
"SHMEM Technical Note for Fortran," SN-2516 2.3, Cray Research, Inc., Eagan, MN, 1994.
Access Regions: - Toward Powerful Parallelizing (Correct)
No context found.
SHMEM Technical Note for Fortran, Cray Research, Oct. 1994
Access Regions: Toward a Powerful Parallelizing Compiler - Paek, Hoeflinger, Padua (1996) (Correct)
No context found.
SHMEM Technical Note for Fortran, Cray Research, Oct. 1994
Using the parallel MICOM on SGI multiprocessors and the Cray T3D. - Sawdey (1995) (Correct)
No context found.
"SHMEM User's Guide," SH-2516 1.1, Cray Research, Inc., Eagan, MN, 1994.
An Implementation of - Libfminus For Sgi (Correct)
No context found.
"SHMEM User's Guide," SH-2516 1.1, Cray Research, Inc., Eagan, MN, 1994.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC