Description of changes:
Added Numerical (float32) C with Python binding mmap value store (2D matrix).
Accelerates sub-matrix retrieval given rows and cols indices, comparing to for-loop np.array.item(i, j).
For value store of 66M x 100, time cost of retrieving (Python binding) random sub-matrix 500x50, comparing to numpy for-loop is 0.09ms (4 threads) vs. 5ms
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.
Issue #, if available: N/A
Description of changes: Added Numerical (float32) C with Python binding mmap value store (2D matrix).
Accelerates sub-matrix retrieval given rows and cols indices, comparing to for-loop
np.array.item(i, j). For value store of 66M x 100, time cost of retrieving (Python binding) random sub-matrix 500x50, comparing to numpy for-loop is 0.09ms (4 threads) vs. 5msBy submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.