I've got a scientific code running on my mac (I'm running snow leopard) , a code written in Fortran. I've been trying to get a profile of the run time of the code, but am having little success. It appears that gprof doesn't return any actual timings, just a count on how many times a subroutine was called. I looked around, this appears to be a common problem.
I tried Saturn, but it doesn't do 64 bit and I'm using external libraries, so that's a no go.
I also tried Shark, but the output of that is not really what I'm looking for, or at least I can't decipher it...
Anyways, just wondering if anyone knows of a profiler for the mac that works with Fortran? I'm using Intel's ifort to compile.
You should be able to use the Time Profiler in Instruments (part of the Developers tools). I've used it before for Fortran programs and been very impressed.
One workaround could be to convert the code to C (Convert Fortran to C or C++) and then to profile it
Related
Short version of question: How do I get started with C programming? Note that I am not asking for a tutorial on learning C language (I can learn that easy enough). I need to setup the environment (I hope I'm asking this question clearly). Here's what I mean:
For my math thesis, I need to write a program in C on Gentoo Linux, using a library called CVODE/SUNDIALS. There is nobody (it seems) in my department who can help me set this up - my professor has left the computer work 100% to me because I have some programming background and he's a math geek. But my experience is with scripting languages (think VBA) and not full, powerful programming languages where you have to link the compiler and libraries, etc. like C.
There is no development environment on the Linux cluster - or at least not that's friendly, and has a debugger - that I've found. So, what I need to figure out how to setup a C programming environment with CVODE library on my PC (Win 7 x64, at little to no cost.
I have found plenty of tutorials on programming in C. I looked up Eclipse, which I have a little bit of experience with, as a development environment, but it's instructions say you need to install a compiler, too.
What I would like is someone to tell me, in simple language that I can understand (which might be the most difficult part of this question) the big picture of what I need and what to do (and maybe even links to where I can find what I need) to set up a C environment with CVODE. If the information is Windows/Gentoo Linux cross platform, even better.
Thank you.
P.S. I did search the site and saw lots of "How do I setup" quesitons, but no C one. Because I know someone will yell at me for that. Also, I don't want to have a convo about whether to use C#, C++, Java, etc. That just complicates the issue - and I need to get this done.
Edit: I have learned a little more since this question and now realize that I left out a key part of the question. The CVODE library and Linux cluster at school use MPI - parallel programming - which is not available on your average, run-of-the-mill PC. So all development must be done directly on the cluster.
Linux: Simple way is to install gcc or g++.
You can write your code in your plain text editor (nano, vim, gedit, kwrite, etc)
Save your file in .c or .cpp extention and type in terminal
gcc filename.c
or
g++ filename.cpp
You said that you want to write c code on Gentoo Linux, as i understand you're not familiar with Linux? The best choice in this case is to:
Install virtualbox in your windows machine (https://www.virtualbox.org/), it's a free software that let you emulate in your desktop another systems like Linux...
Install Gentoo linux on virtualbox, there are a lot of tutorials on the net, for example this video: http://www.youtube.com/watch?v=DUf_1wAPeyA
When you install Gentoo Linux on virtualbox you have all you need to develop C (gcc compiler, gdb debugger...)
Now you can download your library, and decompress it
In general all (Good) Linux libraries come with a 'README' file that contain all instructions for installing the library.
I think you need to do this:
./configure --prefix=/DIRECTORY_YOU_WANT_TO_INSTALL_THE_LIBRARY
make
make install
You can now play with C and you new library, like this:
suppose you create a new file test_lib_ CVODE.c you can compile it like this:
gcc -Wall test_lib_ CVODE.c -o test_lib_ CVODE -lcvode
I assume that the installed library is named libcvode.so
If you have any questions, you can always get help here.
Regards.
I think you should use Code::Block in Linux, it is very similar to Window's Code::Block and it is very easy to debug and other things.
These were all useful answers. I tried pursuing each of them at least a little bit. However, the only reasonable solution seems to be to use emacs on a terminal window. This is because I'm using MPI - yes, I know I didn't mention that in the OP - which can only be done on a cluster.
I am new to this environment and was not aware of the MPI or the affect it would have on my attempt to develop.
I believe I can do better than this if I can figure out X/Windows using Cygwin. But I am a long way from that.
Thanks all for your effort and sorry I can't really award a best answer (I guess).
I have the following problem: I want to simulate some control engineering system. As it is quite complicated I used the computer to derive some (complicated) equations which can only be exported to fortran77 code. On my development pc this is no problem (linux machine).
No I want to do it in hardware and here I have to use a windows OS. Unfortunately matlab does not support gfortran on windows. Therefore I can not compile the fortran files on that box.
My idea was now to translate to C as a C compiler is available at matlab on win. Any other (better) options?
I have several SUBROUTINES in the manner of the following.
SUBROUTINE sys_rhs (x, v, dx, param)
REAL*8 x(6)
C code is coming here
RETURN
END SUBROUTINE
Is it good practice and does it work to use the function sys_rhs__ in the interface function written in C?
The problem is, that I can not test it as I do not have massive access to the windows machine. So it should work if I try it and I should not need to experiment a lot.
Also what lib's do I need on windows? Where can I get them?
Thanks
Christian
Intel sells a pretty good fortran compiler (http://software.intel.com/en-us/articles/fortran-compilers/) which they used to give away for free for testing ... ;-)
PS No, Im not linked to this company in any way!
You might wanna check the gnumex project, which allows to compile MEX-files using the GNU GCC compiler suite (using either MinGW or Cygwin), with the ability to use g77 to compile Fortran programs.
The answer was quite simple: I had some syntax errors in my fortran code. gfortran does no such strict syntax checking and therefor interpreted the fortran code correctly. After removing the typos I succceded transcoding with f2c.
Thanks
there is a better solution to all this
use your linux machine or your windows machine to implement the function in c/c++ and then generate a dll file.
In matlab you can load the DLL and make a call to the functions inside it.
here is a simple tutorial that show how to call the dll with matlab.
http://www.mathworks.com/matlabcentral/fileexchange/15065-example-of-loading-c-dll-and-performance-comparison
The title is fairly self explanatory... isn't anywhere to be found in the version I have (the latest), and all searches on the web have turned up little to nothing in the way of real solutions. There is only a pdf file which explains that omp flags must be marked in the C compiler I use, but I can't even get that far because I can't find the library anywhere!
Is there just a way to pull the library from somewhere and load it in? Or do I need an entirely new gcc file for my IDE to pull from, and if so, where do I get it?
EDIT: Still no luck searching. I'm also willing to work with a different IDE similar to NetBeans... preferably one that uses the Cygwin compiler set as well. If anyone has any suggestions on what I could use that would support OMP, I would be more than willing to entertain them.
Oracle Solaris Studio on Solaris (SPARC and x86) and Linux (x86) provides an IDE with C/C++/Fortran compilers with OpenMP 3.0 enabled, and a debugger and performance analyzer that understand OpenMP.
http://www.oracle.com/technetwork/server-storage/solarisstudio/index.html
And, it's a free download and free to use.
Has anybody experienced using the Invensense IMU3000 with some microcontroller?
I am trying to build the IMU library for a PIC but I am stuck with the dependencies.. any other experience with others microcontrollers will be nice as well!
Basically I don't get whether it is better to take the Visual Studio 2005 project and make the changes there, adding the PIC dependencies (I get stuck..) or compile the whole library in the PIC environment..
Any hint, even with other platforms, would help!
Thank you all!
PC and PIC programming are so very different... Also there are so many pIC variants, they are hugely different from each other, it's hard to answer such an open ended question. However, basically you're writing mathematical algorithms. So write these as ansi c functions, hosted with a load of PC things (dialogs etc) and once they're working, you can move just the math functions over to the PIC - having already got a framework running on the PIC, ready to receive the algorithms. BUT - take care with memory. You have bags of it on the PC, you have to be mean with memory once you work on a PIC. Good luck, enjoy!
While it can be helpful to write code on the PC that will eventually move to the PIC, you will need to make sure that all code that will move has been written with portability in mind. That is, you cannot assume that code that compiles and works perfectly under Visual Studio will work without modification on any other platform.
To run in the PIC, all of the code must be compiled with cross development tools that are designed to target the PIC.
That said, I often develop algorithms and detailed processing code in the PC where a test suite can be easily used to verify its operation, and then recompile it for my target platform.
Incidentally, Google tells me that the IMU-3000 is a MEMS Gyro. It would probably be helpful to include at least the link to its data sheet in the question.
I want to run a simple hello world, written in c, app.
on my at91sam9rl-ek.
Is it possible without an os?
And (if it is) how do I have to compile it?
-right now I try using g++ lite for creating arm code
(In general which programms can the board start without OS,
assembler, arm code?)
Sure, no problem running without an operating system, I do that kind of thing daily...
http://sam7stuff.blogspot.com/
You programs are, at least at first, not going to resemble desktop applications, I would avoid any libraries C libraries, no printfs or strcmps or things like that until you get the feel for it and find the right tools. No floating point as well. add some numbers do some shifting blink some leds.
codesourcery lite is probably the fastest way to get started, the gnueabi one I believe is the one you want.
This winarm site has a compiler and tons of non-os projects for seems like every arm based microcontroller.
http://www.siwawi.arubi.uni-kl.de/avr_projects/arm_projects/
Atmel is very very good about information, no doubt they have example programs you can try as well on the eval board.
emdebian is another cross compiler that is somewhat up to date and has binaries. building a gcc from scratch for cross compiling is not bad at all. The C library is another story though, and even the gcc library for that matter. I find it easier to do without either library.
It is possible get a C library working and run a great many kinds of programs. Depends on what you are looking to do. Ahh, just looked at the specs, that is a pretty serious eval board, plenty of power for an operating system should you choose to run one. You can certainly run programs that use the display as a user interface. read/write sd cards, usb, basically everything on the board, without an os, if you choose.