home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.parallel
- Path: sparky!uunet!gatech!hubcap!fpst
- From: grendel@fen.arc.nasa.gov (Raymond E. Suorsa)
- Subject: Re: Could I compete the low-end CM5 with multiple SPARCstations?
- Message-ID: <1992Jul23.174946.21554@hubcap.clemson.edu>
- Apparently-To: comp-parallel@ames.arc.nasa.gov
- Sender: usenet@news.arc.nasa.gov
- Reply-To: grendel@windchime.arc.nasa.gov
- Organization: NASA Ames
- References: <1992Jul23.134928.5014@hubcap.clemson.edu>
- Date: Thu, 23 Jul 1992 17:30:52 GMT
- Approved: parallel@hubcap.clemson.edu
- Lines: 29
-
- In article <1992Jul23.134928.5014@hubcap.clemson.edu>, lycmit@x1sun6.ccl.itri.org.tw (Yin-Chih Lin) writes:
-
- |> I am curious, if I connect 32 SPARCstation 10 model 30 (33MHz, SuperSPARC
- |> module with 86.1 MIPS - Sun Micro adverted!?) computers (loosely-coupled
- |> MIMD multicomputers?) either by Ethernet, ISDN or FDDI with the multi-
- |> thread OS and might use facilities from Linda-like distributed data
- |> structures. Is it possible for me to obtain the reasonable performance
- |> when compared with the 32-node TMC CM5 (SIMD machine)?
- |>
- more stuff
-
- I have use a network of SparcStation2's as a loosely-coupled MIMD
- machine. My application (obstacle detection and ranging using machine
- vision for guidance applications) would saturate the ethernet with 6
- to 8 machines (32 nodes just would slow it down). If your application
- doesn't need much or any internode communication AND it needs A LOT of
- computation time than a network of 32 Sparc10s would give good
- performance. As soon as the computation time nears the data transport
- time (which happens pretty fast with ethernet) the network will
- saturate.
-
- The price of a TMC CM5 is in its fat tree network... which is very
- fast compared to ethernet, or FDDI.
-
-
- __o Ray Suorsa,
- \<, grendel@windchime.arc.nasa.gov,
- ()/ () NASA/Ames (415)604-6334
-
-