Great find, and thanks for the patch!
Branch predictors do a good job for predictable things, like bounds
checking, as you say. But they are poison for data-dependent branches,
especially in sorts where by the nature of things they will miss half
the time.
Henry Rich
On 2/24/2022 9:19 PM, Elijah Stone wrote:
On Thu, 24 Feb 2022, Elijah Stone wrote:
http://www.jsoftware.com/pipermail/programming/2014-March/035754.html
(Incidentally, the note on branches in that thread is nonsense now and
was nonsense in 2014. Branch predictors are quite good and will make
mincemeat of a bounds check in a loop.)
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
--
This email has been checked for viruses by AVG.
https://www.avg.com
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
