Timeline for Why is ParallelTable slower than Table?
Current License: CC BY-SA 3.0
13 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Jan 2, 2021 at 11:53 | answer | added | Igor Ceapa | timeline score: -2 | |
| May 22, 2018 at 9:18 | comment | added | Albert Retey | you should actually do the comparison of Timing and AbsoluteTiming for an example where there is some speed gain by parallelizing. The two will only give the approximately same results in cases where the overhead dominates and there is no speed gain (or more precisely when the time spent in the main kernel with handling overhead is comparable to the total computation time)... | |
| Oct 19, 2016 at 14:44 | comment | added | Dr. Wolfgang Hintze | @Quantum_Oli Thank you for your suggestion, I confirm it. See my EDIT. | |
| Oct 19, 2016 at 14:43 | history | edited | Dr. Wolfgang Hintze | CC BY-SA 3.0 | added 853 characters in body |
| Oct 13, 2016 at 9:21 | comment | added | Quantum_Oli | Try using EulerPhi with a table going up to 10^6 elements. On my system it takes 3.7s using Table and 1.1s using ParallelTable. | |
| Oct 12, 2016 at 23:12 | history | tweeted | twitter.com/StackMma/status/786343748091383808 | ||
| Oct 12, 2016 at 18:02 | answer | added | Szabolcs | timeline score: 13 | |
| Oct 12, 2016 at 17:23 | answer | added | Albert Retey | timeline score: 7 | |
| Oct 12, 2016 at 13:00 | comment | added | Dr. Wolfgang Hintze | @Quantum_Oli I appreciate any suggestions for a complicated p to perform this test. | |
| Oct 12, 2016 at 12:57 | history | edited | Dr. Wolfgang Hintze | CC BY-SA 3.0 | AbsoluteTiming for measurement of execution time added |
| Oct 12, 2016 at 10:36 | answer | added | Szabolcs | timeline score: 7 | |
| Oct 12, 2016 at 10:14 | comment | added | Quantum_Oli | Likely the overhead time cost of distributing the task across cores is longer than it takes for one core to just do the task, in this case. If you try a more complicated p which takes more time to execute you should find a situation where ParallelTable wins. | |
| Oct 12, 2016 at 9:30 | history | asked | Dr. Wolfgang Hintze | CC BY-SA 3.0 |