Download presentation
Presentation is loading. Please wait.
1
D4M for Julia Alex Chen
2
Dynamic Distributed Dimensional Data Model (D4M)
A tool for big data analysis designed by Dr Jeremy Kepner (Enabling Composable Analytic Computation on Triple Store data) Computation ex. Logical * ./ + D4M A Associative Array DB table What and Why D4M on Julia?
3
What is Implemented? ⬤ Assoc() ⬤ CSV Loading/Writing
⬤ Condense ⬤ GetIndex ⬤ Matrix goodies: find, value =, ./ , *, +, sum, transpose … ⬤ Regex Indexing ⬤ Union Assoc ⬤ A few examples from D4M Matlab Implementations What has been done
4
What is learned? Sorted Algorithms are useful (Case: Matrix Multiply)
Identifying and reasoning performance on a developing language is hard. (Case: Addition) Making Module in Julia is easy. Lesson Learned What has been learned
5
Matrix Multiply sorted_intersect() Implemented
6
Addition Fix
7
Difference between sad and fixed?
Sad Version … ABA = spzeros(n,n) ABA[Arow,Acol] = At.A ABA[Brow,Bcol] += Bt.A ... Fixed Version … ABA = spzeros(n,n) ABA[Arow,Acol] += At.A ABA[Brow,Bcol] += Bt.A ...
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.