There is a paper “Parallel Search on Video cards”, saying that the P-Ary Search could outperform the binary-search based algorithm, I have followed the paper’s idea and re-implement the P-Ary search. But i do not get that good results. Could any one help me improve my code or give me advice.
My code is in the attachment. Please email me or reply this post. And my email: email@example.com By the way: I am preparing the CIKM10 Conference, if any one have interest, we could co-write the paper about this topic. And I have paper "Efficient Lists Intersection by CPU-GPU Cooperative Computing" accepted by the Workshop LSPP of IPDPS10 conference. Thank you in advanced.
paryListIntersection.tar.gz (2.16 KB)