Printable View
I apologize if this is not the appropriate place to ask this question, or if my query is too elementary for this board, yet I am having difficulty determining the number of floating point operations that are required to run a program I wrote in MATLAB and compiled to C.
Is there a simple way to determine the number of flops required to run my code?
Thank you so much for any help and guidance.
mollyann
Well, you can tell your compile to spit out the assembly code for the program and then you can comb through that. For example, if you're using gcc as your compiler you just give it the -S switch and then it will create a <your program name>.s file that contains the assembly code.
The only way I can think of to do this would be to examine the assembly code of your program (either by setting your compiler to output an assembly listing or by viewing the assembly window of your debugger) and then looking up each floating point instruction in the documentation of your processor to see how many clock cycles that instruction takes. There may be some profilling software available to automate this process.
[edit]itsme86 you beat me too it[/edit]
Except a static count of the number of instructions takes no account of the run-time behaviour (say instructions inside loops).
Thank you all for taking the time to respond.
Salem, true. Do you have any suggestions other than going over the assembly code? A program "Crystal Flow" was recommended to me to hlep determine the number of flops within loops, but I have not tried it yet and know nothing about it.
I didn't mean to just count the number of assembly instructions. Loops and such usually depend on variable data that's determined at runtime. So if she just steps through the program for a given set of data as if she was processing each instruction himself, she could just keep a simple tally.Quote:
Originally Posted by Salem