Talk:CalmoSoft primes: Difference between revisions
Content added Content deleted
(→Sequence length: Commented.) |
|||
| Line 41: | Line 41: | ||
:What would make a difference is to speed up the sieving. The C++ entry is using ''primesieve'' which, amongst other tricks, uses all available cores to maximize speed. Unfortunately, Wren is single threaded and so this technique is not available to the CLI version though, in an embedded application, one could write a function to utilize this library from a C/C++ host which Wren could then call. --[[User:PureFox|PureFox]] ([[User talk:PureFox|talk]]) 13:23, 10 April 2023 (UTC) |
:What would make a difference is to speed up the sieving. The C++ entry is using ''primesieve'' which, amongst other tricks, uses all available cores to maximize speed. Unfortunately, Wren is single threaded and so this technique is not available to the CLI version though, in an embedded application, one could write a function to utilize this library from a C/C++ host which Wren could then call. --[[User:PureFox|PureFox]] ([[User talk:PureFox|talk]]) 13:23, 10 April 2023 (UTC) |
||
::Of course the real key to performance is ''not'' summing all 12.5e14 consecutive subsequences and testing every single one of those for primality, which would take quite some time no matter how good your compiler or how many cores you had. --[[User:Petelomax|Petelomax]] ([[User talk:Petelomax|talk]]) 01:55, 12 April 2023 (UTC) |
|||