WP34S NQueens Benchmark Based on Patrice Torchet's BITMAP Solution

10182014, 01:36 AM
Post: #10




RE: WP34S NQueens Benchmark Based on Patrice Torchet's BITMAP Solution
Accurate benchmarking seems to be a very hard task. I choosed the nqueens problem with a simple algorithm, to be able
to compare the basic programming elements of the languages of as many calculators as possible, even if the programming capabilities are limited. Of course there are much better algorithms to solve this puzzle, but these are not suited for many calculators. I think that in most cases the translation of the original BASIC code is accurate enough, because the speed difference between one array access or two is very small, but as Patrice mentioned, I also see a disadvantage for some RPN versions, because there is only one register for indirect addressing unlike the WP34S, that makes the inner loop more complicated than neccessary. So a RPN friendly translation of the code would be more fair. If I remember correctly Patrice already tried this with about 30% speed advantage compared to the actual RPN versions. IMHO using this version would be really fair, because it's RPN friendly but also close to the original algorithm. Too bad that we haven't used this version from the beginning. I like the BITMAP solution a lot, but how can I use it in the benchmark list without putting many other calculators at a disadvantage? Please take into consideration that the BITMAP version would be also much faster on some other non HP calculators and the comparability of the results will suffer, due to the use of different algorithms. The BITMAP solution has earned an own article. Calculator Benchmark 

« Next Oldest  Next Newest »

User(s) browsing this thread: 1 Guest(s)