| dc.contributor.advisor | Dhekne, PS | |
| dc.contributor.author | Hari, G. | |
| dc.date.accessioned | 2025-10-15T11:09:22Z | |
| dc.date.available | 2025-10-15T11:09:22Z | |
| dc.date.submitted | 1996 | |
| dc.identifier.uri | https://etd.iisc.ac.in/handle/2005/7195 | |
| dc.description.abstract | The basic hardware and software features of the ANUPAM system have been described
in this chapter. Communication within and outside the clusters in ANUPAM are also
described.
The communication bandwidth within an ANUPAM cluster is around 2 Mbytes/sec.
There are several software overheads with the UNIX communication drivers, which have
restricted the communication performance of the system. These overheads affect the speedup
of communication-intensive programs substantially. For example, the UNPACK benchmark
program (1000 × 1000 matrix size, double precision) for one CPU in ANUPAM gives around
7 Mflops whereas it yields just 18 Mflops with 8 processors. In the following chapters,
software and hardware modifications for improving the system communication performance
are suggested. | |
| dc.language.iso | en_US | |
| dc.relation.ispartofseries | T03935 | |
| dc.rights | I grant Indian Institute of Science the right to archive and to make available my thesis or dissertation in whole or in part in all forms of media, now hereafter known. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation | |
| dc.subject | Cluster Communication | |
| dc.subject | UNIX Communication Overheads | |
| dc.subject | Benchmark Performance | |
| dc.title | Design and implementation of a high speed interconnection network for ANUPAM | |
| dc.type | Thesis | |
| dc.degree.name | MSc Engg | |
| dc.degree.level | Masters | |
| dc.degree.grantor | Indian Institute of Science | |
| dc.degree.discipline | Engineering | |
