User contributions
Jump to navigation
Jump to search
- 04:39, 25 October 2007 diff hist +568 m Talk:Hex →Efficiency Tester vs. Code Racer: 3 or 10 or 30 ms per run, that is the question
- 04:34, 25 October 2007 diff hist +54 Code Racer →Code: report llGetRegionTimeDilation with each burst of results, alongside llGetTimestamp
- 04:10, 25 October 2007 diff hist +942 Talk:Hex →Solution: ack kind words, hope, red mist
- 18:48, 24 October 2007 diff hist −8,743 m Talk:Code Racer archive the talk that developed the code to run a number of versions of code against one another, not just two
- 18:47, 24 October 2007 diff hist +2,902 Code Racer run a number of versions of code against one another, not just two -- as developed at Talk:Code Racer
- 18:33, 24 October 2007 diff hist +8 Talk:Code Racer (1) yes report ms not s (2) begin with a 2:1:0 ratio between examples rather than 1000:100:0
- 18:31, 24 October 2007 diff hist +469 m Talk:Hex →Solution: ack the Talk:Hex#Third Talk
- 18:21, 24 October 2007 diff hist +630 Talk:Hex →Solution: copy an older alternative in from the rest of the talk
- 18:18, 24 October 2007 diff hist +11 m Talk:Code Racer reshuffle reported numbers to more clearly contrast one run time with the next
- 18:08, 24 October 2007 diff hist +194 m Talk:Code Racer conclude -- give much different times as the first example, subtract out the measured overhead, restore the feature of just 1 run of the often inaccurate leading burst of 10
- 17:58, 24 October 2007 diff hist +93 m Talk:Hex →Solution: clarify
- 17:53, 24 October 2007 diff hist +205 Talk:Hex →Solution: no thank you
- 17:51, 24 October 2007 diff hist +142 Talk:Code Racer credit the runner that ran the race, even while runners rotated -- rotate the runners more often -- etc.
- 17:06, 24 October 2007 diff hist +115 m Talk:Code Racer guess again how to persuade llList2ListStrided to return every distinct stride of the list once, etc.
- 15:25, 24 October 2007 diff hist +561 Talk:Hex →Why bother with fast or small or different code exemplars: propose the article tab hosts brief & clear & conventional while the discussion tab hosts clever & small & fast
- 15:12, 24 October 2007 diff hist +666 m Talk:Hex →Why bother with fast or small or different code exemplars: ack ack ack, yes we're synched
- 14:54, 24 October 2007 diff hist −42 m Talk:Code Sizer delete all text to archive the closed threads that by now exist as part of this Code Sizer article, specifically: The Plus Four Instrument, Why The Code Sizer Article Exists
- 14:54, 24 October 2007 diff hist −2,180 m Talk:Code Sizer This page intentionally left mostly blank.
- 14:49, 24 October 2007 diff hist +241 m Code Sizer →How Small Is That Code: paraphrase Strife's explanation that the extra four bytes of code space go to create the table addressed by GFR when you add the first function to a script
- 14:31, 24 October 2007 diff hist +3,798 Talk:Hex →Third Talk: thanks to Strife for explaining how some Code_Sizer instruments report results off by four, etc.
- 13:43, 24 October 2007 diff hist +248 m Talk:Hex →Talk:Hex Archives: archive five closed threads: Edit Conflicts, 18 October Rewrite, Loop Costs Unscientific, Due Diligence, Correct At A Glance
- 13:41, 24 October 2007 diff hist −17,731 m Talk:Hex begin to archive closed threads by deleting them - hold off on summarising them to avoid overwhelming this wiki's differencing engine
- 12:28, 24 October 2007 diff hist +243 N Talk:LlGetAndResetTime we've lost the astonishing point that llGetTime returns results that differ by more than zero yet less than one.
- 12:26, 24 October 2007 diff hist +226 N Talk:LlGetTime we've lost the point that llGetTime returns results that differ by more than zero yet less than one.
- 12:23, 24 October 2007 diff hist +1 m User:Strife Onizuka/Talkpage Archive 01 missed a bracket in my last edit, how, who can say
- 12:22, 24 October 2007 diff hist +482 m User:Strife Onizuka/Talkpage Archive 01 thanks Strife for coining the term "script time"
- 20:23, 22 October 2007 diff hist −2 m Talk:Code Racer likely not finished, yet still improved, by removing spurious semicolons from the ends of "for" source lines
- 19:54, 22 October 2007 diff hist +4,347 m Talk:Code Racer towards running indefinitely many versions of code -- this version compiles, but doesn't yet run correctly
- 15:05, 22 October 2007 diff hist +303 Talk:Hex →LSO, Defined: the LL client doesn't reliable cache a .lso file per script compiled
- 15:02, 22 October 2007 diff hist +2,495 m Talk:Code Racer start towards running indefinitely many versions of code
- 14:56, 22 October 2007 diff hist +6 m LlListSort spelling: to affected from effected, to descending from desending, to ascending from assending, also mention slowness of bubble sort more prominently
- 12:29, 22 October 2007 diff hist +586 m Talk:Hex →LSO, Defined: guess we sometimes mean LSL byte code when we sloppily say LSO
- 12:14, 22 October 2007 diff hist +440 Talk:Code Racer let's race indefinitely many versions, not just two
- 17:12, 20 October 2007 diff hist +504 m Talk:Hex Efficiency Tester and Code Racer maybe begin to disagree since changing to llGetTime from llGetTimestamp
- 11:05, 20 October 2007 diff hist +109 Hex report the Efficiency Tester llGetTime results that show cleverness to be fast is too negligible to measure reliably, not yet the Code Racer results that show a measurable good
- 10:58, 20 October 2007 diff hist +726 Talk:Hex →llGetFreeMemory and llGetTime results: add the arithmetic of { middling = ((fastest + slowest) / 2) } with a +- imprecision of { max(middling - fastest, slowest - middling)) }.
- 10:49, 20 October 2007 diff hist +2,195 Talk:Hex here's the raw data for a new round of summary results, measured with llGetTime and llGetFreeMemory rather than with llGetTimestamp and llGetFreeMemory -- ~~~~
- 09:30, 20 October 2007 diff hist +108 m Hex add note: To see the raw data and arithmetic that stand behind the summary results briefly presented here, visit the discussion tab.
- 09:24, 20 October 2007 diff hist −40 m Hex change to recommend llGetTime for measuring elapsed run time, rather than llGetTimestamp -- recklessly change now in advance of any volunteer completing the work of submitting the new results
- 09:23, 20 October 2007 diff hist −330 m Hex clarify by balancing the tree that is the Table Of Contents
- 09:11, 20 October 2007 diff hist +10 m Efficiency Tester quote byte code size as commented, not the raw llGetFreeMemory result
- 09:09, 20 October 2007 diff hist +178 Efficiency Tester report ms of llGetTime rather than s, more like the legacy of ms of llGetTimestamp
- 08:58, 20 October 2007 diff hist +450 m Talk:Hex add missing -- ~~~~ signatures
- 08:54, 20 October 2007 diff hist +2,924 Talk:Hex →Different & Clever & Fast: a preview - a run of just 1000, using llGetTime rather than llGetTimestamp, one or two runs for each implementation
- 08:34, 20 October 2007 diff hist +302 Talk:Hex →Different & Clever & Fast: 1000 runs of Efficiency Tester says different & clever & fast actually runs slower than different & concise & small
- 08:27, 20 October 2007 diff hist +5 m Talk:Hex →Different & Clever & Fast: link the code example directly back to this talk page, not distractingly to the nearby article
- 07:44, 20 October 2007 diff hist +531 m Talk:Hex →Different & Clever & Fast: show the code we described
- 06:53, 20 October 2007 diff hist +164 Talk:Efficiency Tester Change complete! The run time cost of the harness may have dropped to below 0.5 milliseconds. -- ~~~~
- 06:53, 20 October 2007 diff hist +8 m Efficiency Tester add units of seconds to the report of time elapsed, also post new sample results to synch with the new code
- 06:48, 20 October 2007 diff hist +380 Efficiency Tester add sample results, clarify to "elapsed" from "passed"