>write naive sort implementation
>outperform glibc's qsort
Write naive sort implementation
Other urls found in this thread:
cs.rpi.edu
twitter.com
Yeah, that naive sort is one hell of an algorithm.
>wrote radix
>3 times faster than std sort on randomly generated uint's
feels good man
same
>work-specific tool works better than Swiss army knife
No shit.
chansort
>for x, wait x, then print x
>not writing a custom, inlined, recursive template metaprogramming mergesort instead of using the slow function pointer based C-sort.
How many elements?
I could outperform you with bubblesort if all the elements fit in a cache line.
>bubble sort
cs.rpi.edu
The C++ master race dabs of C-faggots once again.
Knowing Jow Forums your "faster than libc/std" sorting algorithms are either bugged or are only faster in a very specific context. Basically your idiots.
chansort beats all other sorts when you have 1940s level of clockspeed and IPC, but an infinite number of cores
>makes my randomized algorithms on graphs 3x times faster
u wot m8
kys
sorting is too high iq this is why i'm not a programmer
fucking qsort is the bane of all C fags who think their language is superior to c++... you pass a goddamn function pointer to it, even if you are trying to sort an array of primitives. This overhead literally slows down the algorithm by 2.5 times.
1 see keyword I don't like
2 post brainlet wojak
3 goto 1
Compare a linear sorting algorithm and a semi logarithmic
> 3 TiMeS FaStEr oN uint
that'll sort'em alwight, huhuhuhuh
both from small numbers (several thousands where run time is no longer sub 1 ms) to massive (gigabytes of data)
it would not be that disasterous if there was not the optimization barrier of dynamic linking and the function got inlined
radix can work on signed and floats as well, just saying
What's the proper way to do it?
This is really interesting
O(nlog(n)) on everything
O(m+n) on numerical
Just saying.