Hi Gilles, Would it be sufficient to include .txt files for the tables and the .ods file provided by Matt?
Best regards, rodde ma 11.7.2022 klo 10.33 Gilles Sadowski <gillese...@gmail.com> kirjoitti: > Hi. > > Le lun. 11 juil. 2022 à 07:23, Rodion Efremov <codero...@gmail.com> a > écrit : > > > > Hello Matt and community, > > > > I have created an ASF JIRA issue back in the days: > > > https://issues.apache.org/jira/projects/COLLECTIONS/issues/COLLECTIONS-797?filter=allopenissues > > I suggest that the benchmark tables and figures be copied over there. > [Preferably, in separate files in formats that can be readily > displayed in a browser.] > > Gilles > > > > > Best regards, > > rodde > > > > On Mon, Jul 11, 2022 at 6:23 AM Matt Juntunen <matt.a.juntu...@gmail.com > > > > wrote: > > > > > Hello rodde, > > > > > > Thanks for your patience while I looked at this. I've made a PR [1] on > > > your benchmark project with an updated benchmark class. (I used the > > > completely uninspired class name of IndexedLinkedListPerformance2 :-) > > > The results back up what you've been saying about the performance of > > > this list implementation. I've attached a spreadsheet summarizing the > > > data for a number of different operations along with some images of > > > some of the most interesting comparisons. I've compared results for > > > java.util.ArrayList, java.util.LinkedList, > > > org.apache.commons.collections4.list.TreeList, and > > > com.github.coderodde.util.IndexedLinkedList (the list in question > > > here) using JDK 18 on list sizes of 10, 100, 1000, and 10000. > > > > > > Below are some notes on the attached images. > > > - get-random.png - Displays timings for element access at random > > > indices. As expected, ArrayList is by far the best. TreeList and > > > IndexedLinkedList are relatively close to each other but > > > IndexedLinkedList is consistently faster. LinkedList was too terrible > > > to even include on the graph. > > > - iterate.png - Displays timings for list traversal using the list's > > > iterator. This was unexpectedly bad for TreeList, which performed far > > > worse than the others. The performance of IndexedLinkedList was on par > > > with the JDK lists overall. > > > - iterate-and-modify.png - Displays timings for iterating through the > > > list while randomly adding and removing elements via the iterator. > > > IndexedLinkedList did extraordinarily well here, with performance very > > > close to LinkedList. Surprisingly, TreeList did worse than all of the > > > others, including ArrayList. > > > > > > Overall, I think this list implementation would be a good option to > > > include in commons-collections. Does anyone have any objections to > > > opening a Jira ticket to pursue this? > > > > > > Regards, > > > Matt J > > > > > > [1] https://github.com/coderodde/IndexedLinkedListBenchmark/pull/3 > > > > > > > [...] > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > >
