Skip to content

Compression PR 2#1589

Closed
Baunsgaard wants to merge 2 commits into
apache:mainfrom
Baunsgaard:CompressSpeed2
Closed

Compression PR 2#1589
Baunsgaard wants to merge 2 commits into
apache:mainfrom
Baunsgaard:CompressSpeed2

Conversation

@Baunsgaard
Copy link
Copy Markdown
Contributor

@Baunsgaard Baunsgaard commented Apr 19, 2022

(just new branch to not force overwrite other commits for safety)

Continue: #1565

This commit adds optimizations to the encoding combination algorithms
to allow faster sparse-sparse and sparse-dense combine.

[SYSTEMDS-3336] CLA Specialized maps

This commit adds a few specializations to maps.
MapToZero for all zero mappings.
MapToCharPByte for 3 byte mappings (in between char and int).

[SYSTEMDS-???] DDCFOR and SDCFOR and Sparsifying compression

[SYSTEMDS-???] sparsifying compression

general specialization lmm and change cost model for lmm
fix cost estimator on unknown dimensions (set to 16)

fix sparse TSMM in full rows CSR

remove memorizer on Offsets

clear soft reference in case of spark compression

replace shortcut and compressed multiply cost minimum rows processed

more likely to transpose

MM binary no decompression

fix single col table on compressed colgroup

transpose size in memory if compressed is equal to compressed size

if no sparse block is allocated allocate a block

minor updates
@Baunsgaard Baunsgaard deleted the CompressSpeed2 branch August 18, 2022 12:06
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant