Example fbench

From wiki.emacinc.com
Revision as of 04:16, 4 December 2013 by Mdean (talk | contribs) (Marked buggy and annotated.)
Jump to: navigation, search
TODO: {{#todo:Review(12.03.13-14:18->JG+);(12.04.13-03:15->MD-)|Jgreene|oe 4,oe 5,jg,md,Buggy}}

NOTE ---- As I said before, we do not put program names, keywords from programming languages, shell commands, etc. in bold. We put them in Courier using the code tag. I did the first one for you. You need to fix the rest of them.

This is a guide to the fbench C example project included in the EMAC OE SDK.

NOTE ---- Use of the second person tense is not acceptable for professional writing. This needs to be worded differently. How fast does your machine execute floating point operations? How accurate are those operations? This project is a floating point benchmark and accuracy testing application that utilizes ray tracing algorithms. It's also a good example of a method of processor performance comparison and compiler optimization testing. It's an excerpt from the fbench project by John Walker of Fourmilab. See John Walker's Floating Point Benchmarks project homepage for more information.

The fbench project builds two executables: fbench and ffbench.

NOTE ---- "it's actually" doesn't sound very professional. "It was derived from," "It is based on," and "It utilizes" are all more professional. If you're going to tell them this, then you should also give them a quick summary of what the algorithm is for rather than just pointing them at some non-EMAC webpage. fbench is a floating point benchmark. It's actually a complete optical design raytracing algorithm based on the Marinchip Interactive Lens Design System. See John Walker's fbench project homepage for more information.

NOTE ---- Be careful about capitalization. Remember that Fast Fourier Transform, in its entirety, is a proper name. Therefore, it must always be capitalized fully, and always the same way. ffbench is a Fast Fourier Transform benchmark. It loops through a fast Fourier transform of a square matrix of complex numbers, reverses the transform and then checks the results. See John Walker's ffbench project homepage for more information.

Opening, Building and Uploading the Project Files

1. Open the C/C++ editing perspective.


2. Open the fbench project files.


3. Build the fbench project.


4. Upload the fbench and ffbench executables to the target machine.


Usage and Behavior

Hardware Requirements

The fbench project is intended for use on C implementations that define int as 32 bits or longer and permit allocation and direct addressing of arrays larger than one megabyte. If your target does not meet these requirements then you're going to have a bad time. NOTE ---- again, 2nd person, and again, too informal.

Using fbench

The fbench program is executed from the console. It takes a single optional parameter.

./fbench <itercount>

Where <itercount> specifies the number of iterations to be performed, with 1,000 being the default.
For archival purposes you'll want to use a value slightly higher than 1,000.

Usage Example

NOTE ---- Make sure to look at how the text is displayed. In this case, using console provided no benefit over using text, but had the drawback that half the text was printed in blue. The Internal Wiki page provides a link to a page which tells all about the syntax highlighting of the plugin we use. You should read over that page to see the different languages available. The text "language" is often what we want for the console, since what we type on the console doesn't get syntax highlighted as we type (in bash, at least). The console type is more for pasting scripts into a wiki page.

root@som9g20:/tmp# ./fbench 2000
Ready to begin John Walker's floating point accuracy
and performance benchmark.  2000 iterations will be made.

Measured run time in seconds should be divided by 2
to normalise for reporting results.  For archival results,
adjust iteration count so the benchmark runs about five minutes.

Press return to begin benchmark:

After fbench is finished it prompts us to stop the timer (by pressing return). HEY MIKE it successfully runs through all iterations of it's process (I tested it with some debug code) but should it really just hang at the end waiting for your keypress? I doubt it. Further investigation is called for. NOTE: ----- looking in the code, you will find this bit:

#ifndef ACCURACY

        printf("Stop the timer:\007");
        fgets(tbfr, sizeof tbfr, stdin);

NOTE: ----- As you can see above, the preprocessor define, ACCURACY, can be used to remove the wait. You should document both ways of building it. gcc has an argument that can be used to set preprocessor defines without having to modify the c program.

Stop the timer:

So we press return... NOTE: ----- again, 2nd person

No errors in results.

...and fbench reports that no errors were found in our floating point operations. Hurrah! NOTE: ----- Hurrah? Really?

Using ffbench

The ffbench program is executed from the console. It takes no parameters.


Usage Example

root@som9g20:/tmp# ./ffbench
20 passes.  No errors in results.

NOTE: ----- The important thing to note about how long it takes is that this example is included so that customers can determine the relative speed of our different boards. The transformation is also not a "dance." It runs until it is finished (30 seconds or so, depending on the speed of your machine). It performed 20 iterations of our Fast Fourier Transform dance and discovered no errors.


The fbench floating point benchmark C example tests the speed and accuracy of your floating point operations. It also provides an excellent example of code addressing processor performance comparison and compiler optimization. I hope you enjoyed it as much as I did. NOTE: ----- look at the last sentence.