Return to BSD News archive
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!uwm.edu!lll-winken.llnl.gov!ames!newsfeed.gsfc.nasa.gov!news!kstailey From: kstailey@leidecker.gsfc.nasa.gov (Kenneth Stailey) Newsgroups: comp.unix.bsd Subject: Re: bsd process sticks in 'D' state Date: 14 Nov 1994 03:58:07 GMT Organization: NASA Goddard Space Flight Center -- Greenbelt, Maryland USA Lines: 24 Message-ID: <KSTAILEY.94Nov13225808@leidecker.gsfc.nasa.gov> References: <39v5fv$47e@usenet.INS.CWRU.Edu> <flipk.784601361@soclab.soc.iastate.edu> NNTP-Posting-Host: leidecker.gsfc.nasa.gov Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit In-reply-to: flipk@iastate.edu's message of 12 Nov 94 00:49:21 GMT >>The problem is the machine does not appear to be executing my program >>efficiently. The symptom is the process state is almost always 'D' >>rather than 'R' or 'S' like it is on my other programs that I believe >>are executing much more efficiently: > >> USER PID %CPU %MEM SZ RSS TT STAT START TIME COMMAND >> bx970 9575 16.8 59.449088 7620 p0 D Nov 33395:00 cdmbv > >>P.S. The program is a learning algorithm on a large text corpus and it >>does need 49Mb of space to represent the data. I decided to read in all >>my data on startup since unix allows a process to consume essentially >>all the virtual memory that's available when it starts up, and the use >>of virtual memory in unix seems to generally be quite efficient >>(e.g. loading a large array - like 49 Mb - with a binary disk file of >>data). > >So in answer to your question, there isn't a whole heck of a lot you can >do, aside from adding more ram .. :( Try "man 2 madvise" for details on how to clue the kernel. More RAM will probably do more to boost the performance than madvise(2) though. Ken