hogstedt@ae.chalmers.se (PER HOGSTEDT) (10/28/90)
I'm sorry if this question has already been covered here, but I was not aware of comp.dsp until this evening :-( We are currently looking for a suitable software package for signal acquisition and analysis on a PC platform in our biomagnetism lab. We would like to use the programs for signal acquisition, data editing, fairly advanced signal processing, control of analouge and digital I/O and would most of all like to see this as a set of callable subroutines to include in our own software. After scanning most of the market (I think) we seems to be left with two candidates: ASYST and MatLab-386. None of these packages allow us to use then as a library, rather, we will have to use these programs as our platform, calling special routines of our own from them :-(. Questions: If you have used one or both of these programs, what is your general opinion? ASYST offers possibilities to create a good, graphics oriented, user interface, but what about speed? Is it slow in signal processing/analysis operations compared to MatLab? Benchmarks? Programming MatLab seems to be quite straight forward, but ASYST's FORTH- looking code makes me a bit doubtful. Do you write understandable, main- tainable code in ASYST? Can MatLab support mouse driven, menu/icon based user interfaces? Any problems interfacing to routines written in e.g. MS-C 6.0, Turbo C++ or Fortran77 (probably Laheigh)? Since we are planning to interface the software to a couple of hardware based (TMS320C25) adaptive filters, it would be interesting to hear of any such experiances. If you respond directly by E-mail, I'll post a summary to the net provided there is enough interest. BTW - are there any listservers/mailservers on signal processing? Regards, Per Hogstedt Chalmers University of Technology Gothenburg, Sweden Internet: hogstedt@plab.se OR hogstedt@ae.chalmers.se
mcmahan@netcom.UUCP (Dave Mc Mahan) (10/30/90)
In a previous article, hogstedt@ae.chalmers.se (PER HOGSTEDT) writes: >We are currently looking for a suitable software package for signal >acquisition and analysis on a PC platform in our biomagnetism lab. > >We would like to use the programs for signal acquisition, data editing, >fairly advanced signal processing, control of analouge and digital I/O >and would most of all like to see this as a set of callable subroutines >to include in our own software. > >After scanning most of the market (I think) we seems to be left with two >candidates: ASYST and MatLab-386. > >None of these packages allow us to use then as a library, rather, we will >have to use these programs as our platform, calling special routines >of our own from them :-(. > >Questions: >If you have used one or both of these programs, what is your general >opinion? I have used matlab (an older version, dated 1984) for doing simulations of digital filters. It works quite well. It has a built-in function that actually implements FIR and IIR filter structures within the program, and so is fairly quick. I don't know how well it would work for real-time data acquisition, filtering, and control. I am not aware of any way to call your own external program from within matlab, but this may have changed since my version was released. The version I have only works in floating point and requires a math coprocessor to work. If you plan to do anything quickly, you should use a fast CPU host. I think you may have trouble doing sustained realtime I/O and control from within this program. >Programming MatLab seems to be quite straight forward, but ASYST's FORTH- >looking code makes me a bit doubtful. Do you write understandable, main- >tainable code in ASYST? I haven't used ASYST, but the MatLab programming language isn't bad. It is not as structured as I would like to see it, but it does work. It has trouble dealing with very large data sets (dimensions of 3000 or more) but seems to work fine for data that is smaller. It isn't really oriented for making decisions and bit-control of I/O (there aren't any operators that really allow this easily) and doesn't really understand about true integers. >Can MatLab support mouse driven, menu/icon based user interfaces? The version I have doesn't support this mode, but this may have changed since 1984. If you really need speed or a menu-type of user interface, you will probably have to create the desired functions within your own code after verifying the simulations on matlab. >Per Hogstedt >Chalmers University of Technology >Internet: hogstedt@plab.se OR hogstedt@ae.chalmers.se -dave