Beginner question about CUDA

Accelerated Computing CUDA CUDA Programming and Performance
1

I’m learning more about gpgpu recently and found cuda amazing, but still have some doubts about how and when to use it.

I’ve made a toy project to create a full text search from scratch in c, so far, so good, works well and have very good precision. Now i want to handle a very large set of data and my system starts to get slow, there is some things that i can do to improve, but, don’t gonna solve the whole scenario.

GPGPU can be applied to full text search? Didn’t found anything solid about this.

2

What kind of text search? Two random online examples:

[url]http://aser.ornl.gov/cui/personal/document/epham09.pdf[/url]
Zhang, Yongpeng, et al. “GPU-accelerated text mining.” Workshop on exploiting parallelism using GPUs and other hardware-assisted methods. 2009.

[url]http://www.gpucomputing.net/sites/default/files/papers/1221/pattmatch-report.pdf[/url]
Onsjo, Mikael, Yoshinori Aono, and O. Watanabe. “Online approximate string matching with CUDA.” 2009-12-18]. http: //pds13. egloos. com/pds/2009 07/26/57/pattmatch-report. pdf (2009).

3

thanks njuffa, gonna read this papers.

My engine is a full text search but i don’t use tf-idf, it’s a simpler and more focused algorithm to work with structured data, something like Algolia (Ranking Algorithm Unveiled: How Algolia Makes Search Better | Algolia Blog).

4

If you know what you are looking for, you can identify relevant literature with ease by using Google Scholar (scholar.google.com). Getting access to those papers may be a different story, as not everything is available online for free, and some good papers may be behind a paywall.