OmniSci.jl: Bringing the open-source, GPU-accelerated relational database to Julia

OmniSci.jl: Bringing the open-source, GPU-accelerated relational database to Julia
By Randy Zwitch, Senior Developer Advocate at OmniSci

JuliaCon 2019 - Baltimore | July 24, 2019

Transcript

1. OmniSci.jl: Bringing the open-source,
   GPU-accelerated relational database to Julia
   JuliaCon 2019 - Baltimore
   

   View Slide

2. Randy Zwitch
   Senior Director of Developer Advocacy
   @randyzwitch
   randy.zwitch@omnisci.com
   /in/randyzwitch/
   /randyzwitch
   

   View Slide

3. Data Grows Faster Than CPU Processing
   

   View Slide

4. GPU Processing Keeps Moore’s Law Alive
   

   View Slide

5. OmniSci and Julia: Separated at Birth?
   https://cmes.fas.harvard.edu/news/todd-mostak-am-12-wins-2011%E2%80%9312-am-thesis-prize
   https://cmes.fas.harvard.edu/news/cmes-am-graduate-explores-power-twitter-big-data
   Although not publicly available until June 2014, OmniSci
   (then MapD, “Massive Parallel Database”) was born out of
   academic research by (founder) Todd Mostak on the Arab
   Spring in 2011-2012
   MapD milestones:
   2011-2012: Mostak at Harvard Center for Middle Eastern Studies
   2013: Mostak at MIT CSAIL under Michael Stonebreaker and Sam Madden
   2013: MapD founded
   2014: Project moved to GitHub
   2014-2015: LLVM JIT query compilation added
   May 2017: MapD Core open-sourced under Apache 2.0 license
   

   View Slide

6. Tweetmap Most Popular OmniSci Demo
   https://www.omnisci.com/demos/tweetmap/
   

   View Slide

7. View Slide

8. OmniSci.jl: Standing on the Shoulders of Giants!
   https://omnisci.github.io/OmniSci.jl/latest/
   

   View Slide

9. Future Opportunities for Collaboration
   

   View Slide

10. Let’s Slay the Two-Language Problem!
    

    View Slide

11. Questions & Free Stuff
    

    View Slide

12. Randy Zwitch
    Senior Director of Developer Advocacy
    @randyzwitch
    randy.zwitch@omnisci.com
    /in/randyzwitch/
    /randyzwitch
    

    View Slide