[MAHOUT-1781] Dense matrix view multiplication is 4x slower than non-view one

• Type: Improvement
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 0.11.0
• Fix Version/s: 0.11.1
• Epic Link: Samsara-framework
• Sprint: Nov/Dec-2015
• Assignee: Dmitriy Lyubimov
• Reporter: Dmitriy Lyubimov
• Created: 25/Oct/15 20:27
• Updated: 11/Apr/16 08:42
• Resolved: 27/Oct/15 07:26

Description

if mxA and mxB are two in-core DenseMatrix matrices, then

mxA(::,: %% mxB(::,: takes 4x the time of mxA %% mxB.

possibly an issue of dot products on VectorViews vs. DenseVectors.

dot product over DenseVectors seems to not to go through aggregate() cost-optimized framework.