| Thu, 25 Nov 2010 13:42:40 +0000 |
mas01mc |
Adding mkc_lsh_update branch, trunk candidate with improved LSH: merged trunk 1095 and branch multiprobe_lsh
multiprobeLSH
|
| Thu, 25 Nov 2010 13:42:40 +0000 |
mas01mc |
Adding mkc_lsh_update branch, trunk candidate with improved LSH: merged trunk 1095 and branch multiprobe_lsh
mkc_lsh_update
|
| Thu, 25 Nov 2010 13:42:40 +0000 |
mas01mc |
Adding mkc_lsh_update branch, trunk candidate with improved LSH: merged trunk 1095 and branch multiprobe_lsh
fewerQueryDatumReads
|
| Tue, 13 Oct 2009 20:17:06 +0000 |
mas01cr |
hack lshlib into happiness with mingw
|
| Tue, 11 Aug 2009 21:42:13 +0000 |
mas01cr |
Header file rearrangement
|
| Sun, 22 Feb 2009 03:44:25 +0000 |
mas01mc |
Added a test program and lshlib functionality to inspect on-disk and in-core hashtable representations.
multiprobeLSH
|
| Sat, 21 Feb 2009 21:56:57 +0000 |
mas01mc |
Added public lsh hashtable inspection methods.
multiprobeLSH
|
| Wed, 28 Jan 2009 17:18:58 +0000 |
mas01mc |
_LSH_DEBUG_ option now reports correct point counts per hashtable read.
multiprobeLSH
|
| Wed, 28 Jan 2009 16:02:17 +0000 |
mas01mc |
Fixed a bug in LSH hashtable writing to disk that doesn't always sort the t2 entries into strict weak ordering. Now it does. Lots of debugging informational code inserted.
multiprobeLSH
|
| Wed, 28 Jan 2009 05:18:14 +0000 |
mas01mc |
insert some statistics gathering for hash tables, add _LSH_DEBUG_ macro to output LSH statistics on INDEX and QUERY
multiprobeLSH
|
| Mon, 26 Jan 2009 05:42:15 +0000 |
mas01mc |
lshlib MultiProbe glue added. Compiles and links against audioDB. Fails LSH tests at the moment.
multiprobeLSH
|
| Sat, 24 Jan 2009 06:16:00 +0000 |
mas01mc |
Removed redundant copy in LSH hashfunction computation when #!defined USE_U_FUNCTIONS. Moved LSH macro definitions into lshlib.h
|
| Fri, 23 Jan 2009 18:45:44 +0000 |
mas01mc |
Implemented correct LSH table via compile-time switch -DLSH_DUMP_CORE_TABLES. Dumps on LSH load.
memory-leaks
|
| Fri, 09 Jan 2009 18:05:32 +0000 |
mas01mc |
Fixed memory leaks, added WS --no_unit_norming, and removed capping of LSH_N_POINT_BITS to 15 bits, instead allow any number of bits to encode points, remaining bits encode tracks
|
| Tue, 30 Dec 2008 17:16:02 +0000 |
mas01mc |
feature DIM upper limit changed. Needed for working with very large data sets
|
| Tue, 07 Oct 2008 21:03:26 +0000 |
mas01mc |
unionized punning of the sbucket.snext field into {subucket* ptr, unsigned numBuckets}snext; so that list-head code is 64-bit friendly.
|
| Wed, 10 Sep 2008 18:55:16 +0000 |
mas01mc |
Added LSH_CORE_ARRAY structure for hashtables instead of linked lists. Maintained Backwards Compatibiliity with indexes build for linked list format. Added tests for indexing and merging. Tested backwards compatibility OK.\n\n The purpose of the LSH_CORE_ARRAY data structure is greater space efficiency and L1/2 cache usage. Essential for multiple indexes with multiple hashtables in RAM
|
| Fri, 05 Sep 2008 14:16:21 +0000 |
mas01mc |
converted read/write into fread/fwrite for LSH hashtable serialize and unserialize. INDEXING is now faster.
|
| Thu, 21 Aug 2008 21:28:33 +0000 |
mas01mc |
Merge of branches/large_adb -r 514:524 onto the trunk. No conflicts. Added LARGE_ADB support. Turn on with --ntracks 20001 or greater. Use --adb_feature_root to locate feature files at QUERY time. A bug fix in LSH indexing that was incorrectly thresholding large numbers of shingles.
|
| Tue, 19 Aug 2008 20:27:15 +0000 |
mas01mc |
Added indexing support for O2_FLAG_LARGE_ADB. Tested on indexed query by features. No indexed query-by-key yet. No --lsh_exact yet.
large_adb
|
| Tue, 19 Aug 2008 14:27:21 +0000 |
mas01mc |
Adding large_adb support (up to 1M tracks)
large_adb
|
| Thu, 07 Aug 2008 17:05:22 +0000 |
mas01mc |
Fixed persistent index bug where indexName identifier was an un protected string pointer rather than a copy of the string. Hence the index name was getting deleted and the test for persistent index was failing.
|
| Thu, 07 Aug 2008 01:53:38 +0000 |
mas01mc |
Added server-side loading of persistent index (LSH hashtables) via --load_index -d dbName -R radius -l sequenceLength. Queries using these parameters will lookup the memory-resident hashtable instead of loading one from disk.
|
| Tue, 05 Aug 2008 22:40:38 +0000 |
mas01mc |
changed FORMAT2 index serialization so that token bits don't scribble over point index for >=32768 tracks
|
| Fri, 01 Aug 2008 15:04:31 +0000 |
mas01mc |
fixed file size allocation for FORMAT2 files. Made LSH index size() in bytes an unsigned long long. Changed the name of lsh_inCore flag to lsh_on_disk (to reverse the sense of the 'flag').
|
| Wed, 30 Jul 2008 15:22:22 +0000 |
mas01mc |
Refactored LSH interface to separate hashfunctions and parameters from insertion/retrieval/serialization
|
| Tue, 29 Jul 2008 22:01:17 +0000 |
mas01mc |
Completed merge of lshlib back to current version of the trunk.
|
