[MXNET-125] Sparse Operators

• Type: Improvement
• Status: Done
• Priority: Major
• Resolution: Unresolved
• Sprint: Apache Sprint 3, Apache Sprint 4, Apache Sprint 5, Apache Sprint 6, Apache Sprint 7
• Assignee: Hao Jin
• Reporter: Chris Olivier
• Created: 22/Mar/18 03:05
• Updated: 06/Jul/18 16:59

Sub-Tasks

• broadcast_mul/div between csr and 1D dense vector: Done
• Fix race condition in sparse batch loader: Done
• dot(dense, csr) = dense and dot(dense, csr.T) = dense on GPU: Done
• elemwise_add/sub between dense and csr matrices: Done
• elemwise_add/sub between dense and rsp matrices: Done
• broadcast_add/sub between csr and 1D dense vector: Done
• elemwise_mul between dense and csr on CPU & GPU: Done
• elemwise_add/sub between rsp and rsp on GPU: Done
• broadcast_mul/div between csr and 1D dense vector on GPU: Done
• concat for csr: Done
• add_n(dense, csr, dense) and add_n([dense, csr, rsp]*, dense, [dense, csr, rsp]*) on CPU & GPU: Done
• unary for csr: Done