René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"DiscussionForumPosting","headline":"Micro-optimization: Use memmove instead of for loop in list.insert()","articleBody":"**Background**\r\n\r\nI noticed a difference in speed when inserting an element to the beginning of the list using list.insert vs list slice assignment. Upon inspecting the code, the usage of `memmove` in the list slice assignment code seems to be the providing a slight speed boost over the for loop used in the list.insert implementation.\r\n\r\nThe following code was used to benchmark the difference:\r\n\r\n```py\r\nimport random\r\nimport timeit\r\n\r\nruns = 100000\r\nsetup = 'import random;lst=[random.randrange(100000) for _ in range(100000)]; x = random.randrange(100000)'\r\n\r\nprint(timeit.timeit(stmt='lst[:0] = [x]', setup=setup, number=runs))\r\nprint(timeit.timeit(stmt='lst.insert(0, x)', setup=setup, number=runs))\r\n```\r\n\r\nOriginal:\r\n```\r\nSlice assignment: 7.394628149999335s\r\nlist.insert:      8.910868348000804s\r\n```\r\n\r\nUsing `memmove` in list.insert():\r\n```\r\nSlice assignment: 7.252985826999065s\r\nlist.insert:      7.215608489997976s\r\n```\r\n (Compiled with optimizations and tested on Arch Linux, kernel 5.18)","author":{"url":"https://github.com/radiantly","@type":"Person","name":"radiantly"},"datePublished":"2022-07-02T18:29:19.000Z","interactionStatistic":{"@type":"InteractionCounter","interactionType":"https://schema.org/CommentAction","userInteractionCount":15},"url":"https://github.com/94508/cpython/issues/94508"}