Content deleted Content added
Undid revision 1266772177 by MrOllie (talk)vandalizing revert, the performance is proven by the code given reg. lack of other sources, pls. stop this silly reverting, you also trash edits acc. your own requests, instead bring complaints into discussion on the talk page. |
added 'WIP' to 'performance'. Tags: Reverted Visual edit |
||
Line 16:
== Performance ==
!!! this point is WIP, under investigation and will be improved shortly, but the info is provided 'as of now' because important for programmers. !!!
Performance comparison is not easy, not very accurate and lacks reproducibility on modern IT systems for various reasons. One can roughly say that in a current 64-bit Intel(r) / linux / gcc / libdfp / BID implementation, basic arithmetic operations with decimal64 values are between factor 2 and 15 slower than with binary64 data types, while 'higher' functions like powers ( ~600 ) and trigonometric functions like tangent ( {{gaps|~10|000}} ) suffer more performance penalties. To get an idea about performance on a specific system the code in 'Addendum - code' can be used. Perhaps the [https://gcc.gnu.org/contribute.html GNU gcc project] and the [https://github.com/libdfp/libdfp/issues?q=is%3Aissue 'libdfp' project on github] could like some help to improve.
Line 200 ⟶ 202:
== Addendum - code ==
!!! this point is WIP, under investigation and will be improved shortly, but the info is provided 'as of now' because important for programmers. !!!
try this to see performance of addition and tangent for bin32, bin64, dec32 and dec64 datatypes, compile and run instructions see in header.
| |||