jon@cs.washington.edu (Jon Jacky) (07/19/90)
Here are excerpts from R&D magazine (a controlled-circulation trade magazine), July 1990, p. 68, in an article "Five Steps to Successful Software Development" by Roger Woodward of Woodward Design Associates, San Francisco: "...For instrument software, specifically real-time microprocessor software, my firm has developed a historical price-estimation scheme that has proven useful and accurate. Based on a large sample of our own projects and another sampling of projects completed at instrument firms that cooperated in our survey, you can figure on $2.75 per byte at 1990 prices. Thus a project that must fit in 64K should cost, start to finish, about $180,000. Our evidence suggests that, for a project of this size, cost estimates below $100,000 and above $300,000 are both suspect." This seems roughly consistent with estimates of $10.00 per source line that I can vaguely remember hearing somewhere. Jon Jacky, University of Washington, Seattle jon@gaffer.rad.washington.edu
mcmahan@netcom.UUCP (Dave Mc Mahan) (07/19/90)
In a previous article, jon@cs.washington.edu (Jon Jacky) writes: >Here are excerpts from R&D magazine (a controlled-circulation trade magazine), >July 1990, p. 68, in an article "Five Steps to Successful Software Development" >by Roger Woodward of Woodward Design Associates, San Francisco: > >"...For instrument software, specifically real-time microprocessor software, >my firm has developed a historical price-estimation scheme that has proven >useful and accurate. > >Based on a large sample of our own projects and another sampling of >projects completed at instrument firms that cooperated in our survey, you can >figure on $2.75 per byte at 1990 prices. > >Thus a project that must fit in 64K should cost, start to finish, about >$180,000. > >Our evidence suggests that, for a project of this size, cost estimates >below $100,000 and above $300,000 are both suspect." > >This seems roughly consistent with estimates of $10.00 per source line that >I can vaguely remember hearing somewhere. FYI, Just for kicks, I once calculated the costs of a radio-telescope I worked on. It had two computers one using BASIC (ugh!) and one using 'C'. Lots of hardware, IEEE devices, etc. I figured it cost about $.10 per character of source (even spaces and comments) and about $3.45 per line of source. If you do the math, you can see that works out to about 34.5 chars/line. This isn't as bad as it sounds (yes, I do indent rigourously) if you count tabs as one character. The whole project had about 16,000 lines of 'C' and about 9,500 lines of BASIC and took me the better part of a year to do. I worked as a contractor for that time and kept all me invoices. By the way, about 75 pages of technical documentation are include in that time frame, but are not factored into the cost/char of source code. The time also included about 5 months in Japan as installation occured. This cost counts the work time I spent over there, but not the per Diem I charged for living expenses. >Jon Jacky, University of Washington, Seattle jon@gaffer.rad.washington.edu -dave
warren@eecs.cs.pdx.edu (Warren Harrison) (07/19/90)
In article <12577@june.cs.washington.edu> jon@cs.washington.edu (Jon Jacky) writes: >Here are excerpts from R&D magazine (a controlled-circulation trade magazine), >July 1990, p. 68, in an article "Five Steps to Successful Software Development" >by Roger Woodward of Woodward Design Associates, San Francisco: ... a description of how they obtained their data >Thus a project that must fit in 64K should cost, start to finish, about >$180,000. > Does this mean that a project which needs to fit in 16K should cost $44,000? I guess the best way to reduce the cost of that 64K system is to get rid of some ROM :-) It seems it would make more sense to include the expected functionality in the estimation procedure - (IMHO) Warren ========================================================================== Warren Harrison warren@cs.pdx.edu Department of Computer Science 503/725-3108 Portland State University