log_plc

PLC Log

09:00-11:00

Revising schedule. Syncing wiki pages about Evaluation with tex-code. Looking at performance graphs.
11:00-11:59

Syncing wiki, looking at ggplot2 for R.
13:00-13:59

Hacking up Criterion to print report samples. Testing sample printing.
14:00-14:39

Sample printing done.
14:40-16:00

Filled out Accelerate evaluation in new schema.

09:30-14:30

Counting reverse dependencies. Made a small program to weed out false positives. Refined scripts, counting is very effective now.
14:30-16:00

Performed the actual counting of reverse dependencies. We now have counts for: Accelerate,Nikola,Repa and Vector.

10:00-10:59

Reading our report section on the survey. It might even make sense (maybe not immediately) to report on the languages that we chose to exclude, because knowing what we couldnt even get working is still valuable information. Investigating Criterion output possibilities. Criterion easily supports outputting all samples through report-template mechanism. Started counting reverse dependencies.
11:00-11:59

Wrote a crude script for collection of reverse dependencies, accepting false positives.
13:00-13:59

Checking reverse dependencies. Adding Accelerate,Nikola,Repa to ohloh.net to see what their automatic code analysis yields.
14:00-16:20

Debugging Cuda: optimizing memory usage and going float->double.

10:00-10:59

Reading Longstaff & Schwartz (LSM). Looking for LSM implementation articles (found a promising one: Choudhury, King, Kumar & Sabharwal, IPDPS 2008). Configured XMonad to have thesis time logs as named scratchpads.
11:00-11:59

Logging. IPDPS2008 seems to have a number of useful references for motivation and introduction. Reading "Experimental Evaluatoin in Computer Science".
12:00-12:59

Reading "An empirical comparison of seven programming languages".
14:00-14:59

Reading "Writing for Computer Science".
15:00-15:59

Reading "Writing for Computer Science".

10:00-10:59

Preparing benchmarks, doubling years from 1 until 64 years. Nikola requires extending the benchmark routine, as it runs more than one benchmark.
11:00-11:59

Still extending test routine. Benchmarks are now running from 1..64 years, doubling each time. Next up after lunch: Filling out evaluation schemas.
13:00-13:20

Procrastination.
13:21-13:59

Looking at evaluation schema. Checking on benchmarks on napoleon. Revising evaluation schema to more clearly distinguish between the haskell DSLs and the external comparison languages (CUDA, R).
14:00-15:00

debugging logging.

Survey "VectorMARK" in progress