Week 10: August 9 - 13

<-- Prev Week

This Week's Goals:

  1. Finalize and document code:

    1. Remove any debugging print statements and variables that are no longer needed and are cluttering up the code and output. Particularly look at place_and_route.c. Also, check for informal comments and variable names, and places I may have forgotten to attach my name to a modification.

    2. Explain (in README file) how to use VPR - which modes are supported/unsupported/untested, what command line information is now required, what the input format is for new inputs, and how to get the graphics to display properly (this is a little tricky).

    3. Add a list of improvements I've made to the README file.

  2. Complete experimentation with VPR:

    1. Complete goal 1 from week 9: increase the bounding box size to get the larger benchmark circuits to route.

    2. Finish running experiments using a single hard macro

    3. Integrate the code that was written a couple of weeks ago to handle multiple hard macros into the most recent version of VPR. Add error checks to make sure hard macros and widened channels do not intersect.

    4. Run a trial on one benchmark with multiple hard macros.

    5. Examination of data suggests that increasing the sizes of channels around the hard macro by adding 10 tracks might be better than doubling the number of tracks in wide channels as was done in previous trials. Run experiments this way, using a single seed to see if this approach yields better data.

  3. Analyze Data and Prepare Final Presentation:

    1. Make graphs of data:

      1. Make one graph per benchmark showing total tracks in each experiment run with that benchmark

      2. Make 3 additional graphs:

        1. tracks per benchmark

        2. wirelength per benchmark

        3. wide channel utilization per benchmark

      3. Make a graph describing the results of the multiple hard macro trial

      4. Make a graph describing the results of the add ten trial

    2. Make slides for my 30 minute final presentation. Include graphs of the data collected in these slides.

  4. Complete These Tasks when Others are Finished - these items will most likely be completed after I leave UCIrvine.

    1. Add code to get information about how long VPR takes to run. Use this feature during experimentation to see how much longer the modified VPR takes than the original VPR did.

    2. Expand on my mid-mentorship progress report to make a final report. Add data, recent VPR improvements, and explanation of how the multiple hard macro features work.

    3. Finalize this website

Progress Toward Goals:

  1. Completed Monday:

    1. Completed Monday

    2. Completed Monday

    3. Completed Monday

  2. Completed Tuesday:

    1. Completed Sunday - All of the circuits that failled to route before routed successfully. Experiments were run Saturday night with placer seed 3 (because this seed had more failures than the other seeds used), and today I will finish trials with the other two seeds.

    2. Completed Monday - ran overnight starting on Sunday

    3. Completed Monday

    4. Completed Tuesday - ran overnight starting on Monday

    5. Completed Tuesday - the data generated this way was no better than the data gained by doubling the channels.

  3. Completed Thursday:

    1. Completed Wednesday:

      1. Completed Tuesday

      2. Completed Tuesday

        1. Completed Tuesday

        2. Completed Tuesday

        3. Completed Tuesday

      3. Completed Wednesday

      4. Completed Wednesday

    2. Completed Thursday

  4. Completed After Leaving UCIrvine:

    1. Abandoned Tuesday - because I don't have much time left here, my mentor and I decided that this wasn't necessary.

      In running my trials, I did see that the largest circuit, with the largest expanded bounding box used in my experiments, the placement and routing finished in less than 5 hours - so 5 hours should be an upper limit on the time anything I've been working with needs. (No one else was logged in to the computer I used for this, so the timing information should be accurate.)

    2. Completed After Leaving UCIrvine

    3. Completed After Leaving UCIrvine

About My Project | home