home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ukma!usenet.ins.cwru.edu!agate!dog.ee.lbl.gov!ucbvax!ACCESS.DIGEX.COM!tdarcos
- From: tdarcos@ACCESS.DIGEX.COM (Paul Robinson)
- Newsgroups: comp.os.vms
- Subject: [TDR] Large Fortran Arrays
- Message-ID: <Pine.3.05.9301271045.C3981-b100000@access>
- Date: 27 Jan 93 15:21:45 GMT
- Sender: daemon@ucbvax.BERKELEY.EDU
- Distribution: world
- Organization: The Internet
- Lines: 34
-
- KN>I have a friend who wants to run a program that has large arrays.
- >He hasn't been able to get beyond linking the darn thing. His
- >declaration section reads as follows:
-
- KN> COMMON X(2400,2400),A1(2400,2400),A2(2400,2400),WT(5,2400),
- > & XT(2400,2400),D(2400),XMEAN(2400),SUM(2400),SD(2400),DT(2400)
- > CHARACTER*15 IN(12),OUT1,OUT2
-
- Let's see: a 2400 x 1000 element record would be 2.4 million words, and
- that's floating-point elements. Times 2400 it would be 5,760,000 words
- *EACH* for X, A1, A2 and XT. It would be another minor 20K for the WT
- and other arrays.
-
- That's 23,040,000 *words* of virtual memory!
-
- Assuming one can set up a process this large (someone else on INFO-VAX
- will probably be able to say) he would almost certainly have to raise some
- limits on resource consumption.
-
- Why does he need to set up such huge arrays? Can he use a sparse matrix
- or something else? Can he create files on disk? This is probably just
- too big given the current limits that are imposed on him.
-
- Reminds me of a story. This one university wanted to do some tests on the
- effects of certain conducting capabilities using very short high-energy
- bursts, of, oh say, 10 gigawatts for three seconds twice a day. Well,
- suffice to say this was more than the entire power capacity for the
- electric company for that area! The Utility informed the university that
- they were not interested in their business for this application and were
- free to find electricity from any other source that was willing to supply
- it in those kinds of quantities.
-
-
-
-