Slashdot Mirror
Sorting Algorithm Breaks Giga-Sort Barrier, With GPUs
An anonymous reader writes "Researchers at the University of Virginia have recently open sourced an algorithm capable of sorting at a rate of one billion (integer) keys per second using a GPU. Although GPUs are often assumed to be poorly suited for algorithms like sorting, their results are several times faster than the best known CPU-based sorting implementations."
GPUs have always been better at sorting your money from your wallet.
I've also found this annoying when reading papers. Perhaps I just spend too much time learning how to use gnuplot so that my graphs look nice.
It's just a milestone.
Hang on, since when do you measure sorting performance using a distance indicator? And an imperial one at that!
No, this is not a serious comment.
It means the Queen of England herself endorses the calculations?
I think the bubble sort would be the wrong way to go.
—Barack Obama
Hold on, so I can play Jezz Ball, Chips Challenge and all my favourite old school games with a greater level of speed.
Thank god, because my online Porn collection isn't going to sort itself.
Actually, if you look at shockwave dynamics during the moment an object crosses from subsonic to supersonic velocity, it can very easily be considered much more of a barrier than 1gkeys/sec can.
Actually in this case, your analogy should use ludicrous speed.
What one fool can do, another can. (Ancient Simian Proverb)
I never would have suspected the GTX480 would have been good at this sorta thing.
Be relentless!
O( n / log(P) ) not so much.
That algorithm does particularly poorly on just one processor. In fact, if it ran successfully the universe would implode.
You know, if they up it to just a bit over 4 billion unique 32-bit keys, say around 4,294,967,296 or so, I think I could sort them rather efficiently, as long as they weren't attached to any payload. ;-)
Program Intellivision!