Faster normalXAngle --> faster sin,cos,tan

Fragment of a discussion from User talk:Rednaxela/FastTrig
← Thread:User talk:Rednaxela/FastTrig/Faster normalXAngle -- faster sin,cos,tan/reply (2)

Jump to navigation Jump to search

Table lookups are slow I suspect due to array index out-of-bounds checking.

It also happens in kNN search. I was able to double the speed of my bot by replacing a few arrays with objects with hardcoded number of attributes. Only works with constant sized arrays, like data point coordinates.

MN‎

You do not have permission to edit this page, for the following reasons:

The action you have requested is limited to users in the group: Users.
You must confirm your email address before editing pages. Please set and validate your email address through your user preferences.

You can view and copy the source of this page.

No, I'm fairly sure it's caching, because profiling shows LUT twice as slow as polynomial, but a micro-benchmark shows LUT ~30% faster.

Return to Thread:User talk:Rednaxela/FastTrig/Faster normalXAngle -- faster sin,cos,tan/reply (5).

How did you do the micro-benchmark?

MN‎

I generate a list of random numbers in an array and then time how long it takes for each of the functions to process the entire array in a loop. The result of each function I add to a running total, and I print the total at the end to make sure it doesn't get optimised away, and also to see how accurate they are. I could also calculate the overhead if I just add the results of the array, but I haven't tried that yet, and I think it might get optimised into SSE ops anyway.

Retrieved from "http://robowiki.net/wiki/Thread:User_talk:Rednaxela/FastTrig/Faster_normalXAngle_--_faster_sin,cos,tan/reply_(2)#Faster_normalXAngle_--.3E_faster_sin.2Ccos.2Ctan_3188"