Timeline for Speed and the MapThread function
Current License: CC BY-SA 3.0
12 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jun 17, 2015 at 4:47 | comment | added | m_goldberg |
@Pickett. Yes. "MapCompileLength" -> 100. Not only that, but Map[# + RandomInteger[] &, t] produces a packed array and runs much quicker than MapThread[# + RandomInteger[] &, {t}] which does not. In this case, the Map is being compiled.
|
|
| Jun 17, 2015 at 0:22 | history | tweeted | twitter.com/#!/StackMma/status/610965735062507521 | ||
| Jun 16, 2015 at 23:33 | comment | added | C. E.♦ | @m_goldberg Strange. And your "MapCompileLength" is set to 100? This is confirmed? | |
| Jun 16, 2015 at 23:16 | comment | added | m_goldberg | @Pickett. From the compiled function `f`` I get timings comparable to what he got. The returned array is packed. | |
| Jun 16, 2015 at 23:06 | comment | added | C. E.♦ | @m_goldberg I'm used 10.1 on OS X to write my answer. What happens if you run the code in Oleksandr's answer? | |
| Jun 16, 2015 at 22:44 | comment | added | m_goldberg |
In V10.1 on OS X, I'm getting the same time, ~3.7 sec, with both Map and MapThread. Can anyone confirm this? PackArrayQ returns False for both a and b.
|
|
| Jun 16, 2015 at 22:38 | history | edited | m_goldberg | CC BY-SA 3.0 |
deleted 31 characters in body
|
| Jun 16, 2015 at 22:31 | vote | accept | CommunityBot | ||
| Jun 16, 2015 at 22:28 | answer | added | Oleksandr R. | timeline score: 12 | |
| Jun 16, 2015 at 22:24 | answer | added | C. E.♦ | timeline score: 14 | |
| Jun 16, 2015 at 22:18 | comment | added | ciao |
MapThread unpacks the array... and does more work under the covers (what would you expect re: latter?), and I'd venture compilation (auto) behaves differently.
|
|
| Jun 16, 2015 at 21:59 | history | asked | user30193 | CC BY-SA 3.0 |