• TurboPFor: The new synonym for "integer compression"

• 100% C (C++ compatible headers), w/o inline assembly
• Usage as simple as memcpy
• 👍 Java Critical Natives Interface. Access TurboPFor incl. SIMD! from Java as fast as calling from C
• ✨ FULL range 16/32/64 bits integer lists and Floating point
• No other "Integer Compression" compress or decompress faster with better compression
• Direct Access is several times faster than other libraries
• ✨ Integrated (SIMD) differential/Zigzag encoding/decoding for sorted/unsorted integer lists
• Compress better and faster than special binary compressors like blosc

+ **For/PFor/PForDelta** - **Novel** **"TurboPFor"** (Patched Frame-of-Reference,PFor/PForDelta) scheme with **direct access** or bulk decoding. Outstanding compression and speed. More efficient than **ANY** other fast "integer compression" scheme. - Compress 70 times faster and decompress up to 3 times faster than OptPFD - 🆕 **TurboPFor now 30%! more faster**

+ **Bit Packing** - ✨ Fastest and most efficient **"SIMD Bit Packing"** - Scalar **"Bit Packing"** decoding as fast as SIMD-Packing in realistic (No "pure cache") scenarios - Bit Packing with **Direct/Random Access** without decompressing entire blocks - Access any single bit packed entry with **zero decompression** - ✨ **Direct Update** of individual bit packed entries - Reducing **Cache Pollution**

+ **Variable byte** - ✨ Scalar **"Variable Byte"** faster and more efficient than **ANY** other (incl. SIMD MaskedVByte) implementation - 🆕 **now up to 25% more faster**

+ **Simple family** - ✨ **Novel** **"Variable Simple"** (incl. **RLE**) faster and more efficient than simple16, simple-8b or other "simple family" implementation

+ **Elias fano** - ✨ Fastest **"Elias Fano"** implementation w/ or w/o SIMD

+ **Transform** - ✨ Scalar & SIMD Transform: Delta, Zigzag, Transpose/Shuffle, Floating point<->Integer

+ **Inverted Index ...do less, go fast!** - Direct Access to compressed *frequency* and *position* data in inverted index with zero decompression - ✨ **Novel** **"Intersection w/ skip intervals"**, decompress the minimum necessary blocks (~10-15%)!. - **Novel** Implicit skips with zero extra overhead - **Novel** Efficient **Bidirectional** Inverted Index Architecture (forward/backwards traversal) incl. "integer compression". - more than **2000! queries per second** on GOV2 dataset (25 millions documents) on a **SINGLE** core - ✨ Revolutionary Parallel Query Processing on Multicores w/ more than **7000!!! queries/sec** on a quad core PC.
**...forget** ~~Map Reduce, Hadoop, multi-node clusters,~~ ...

CPU: Sandy bridge i7-2600k at 4.2GHz, gcc 5.1, ubuntu 15.04, single thread.

• Realistic and practical "integer compression" benchmark with large integer arrays.
• No PURE cache benchmark

• Generate and test skewed distribution (100.000.000 integers, Block size=128)
  Note: Unlike general purpose compression, a small fixed size (ex. 128 integers) is in general used in "integer compression". Large blocks involved, while processing queries (inverted index, search engines, databases, graphs, in memory computing,...) need to be entirely decoded

   ./icbench -a1.5 -m0 -M255 -n100m
  

CPU: Skylake i7-6700 w/ only 3.7GHz

• CPU: Sandy bridge i7-2600k at 4.2GHz

• gov2.sorted from [DocId data set](#DocId data set) Block size=128 (lz4+blosc+VSimple w/ 64Ki)

   ./icbench -c1 gov2.sorted
  

Ki=1024 Integers. 64Ki = 256k bytes
"lz4+DT 64Ki" = Delta+Transpose from TurboPFor + lz4
"blosc_lz4" tested w/ lz4 compressor+vectorized shuffle

GOV2: 426GB, 25 Millions documents, average doc. size=18k.

• Aol query log: 18.000 queries
  ~1300 queries per second (single core)
  ~5000 queries per second (quad core)
  Ratio = 14.37% Decoded/Total Integers.

• TREC Million Query Track (1MQT):
  ~1100 queries per second (Single core)
  ~4500 queries per second (Quad core CPU)
  Ratio = 11.59% Decoded/Total Integers.

• Benchmarking intersections (Single core, AOL query log)

• Benchmarking Parallel Query Processing (Quad core, AOL query log)

• Search engines are spending 90% of the time in intersections when processing queries.
• Most search engines are using pruning strategies, caching popular queries,... to reduce the time for intersections and query processing.
• As indication, google is processing 40.000 Queries per seconds, using 900.000 multicore servers for searching 8 billions web pages (320 X size of GOV2).
• Recent "integer compression" GOV2 experiments (best paper at ECIR 2014) On Inverted Index Compression for Search Engine Efficiency using 8-core Xeon PC are reporting 1.2 seconds per query (for 1.000 Top-k docids).

make

• test all "integer compression" functions
  

  ./icbench -a1.0 -m0 -M255 -n100m
  

-zipfian distribution alpha = 1.0 (Ex. -a1.0=uniform -a1.5=skewed distribution)
-number of integers = 100.000.000
-integer range from 0 to 255

• individual function test (ex. Copy TurboPack TurboPFor)
  

  ./icbench -a1.5 -m0 -M255 -ecopy/turbopack/turbopfor -n100m
  

• Data file Benchmark (file from [DocId data set](#DocId data set))

  ./icbench -c1 gov2.sorted
  

1 - Download Gov2 (or ClueWeb09) + query files (Ex. "1mq.txt") from [DocId data set](#DocId data set)
8GB RAM required (16GB recommended for benchmarking "clueweb09" files).

2 - Create index file

    ./idxcr gov2.sorted .

create inverted index file "gov2.sorted.i" in the current directory

3 - Test intersections

    ./idxqry gov2.sorted.i 1mq.txt

run queries in file "1mq.txt" over the index of gov2 file

1 - Create partitions

    ./idxseg gov2.sorted . -26m -s8

create 8 (CPU hardware threads) partitions for a total of ~26 millions document ids

2 - Create index file for each partition

  ./idxcr gov2.sorted.s*

create inverted index file for all partitions "gov2.sorted.s00 - gov2.sorted.s07" in the current directory

3 - Intersections:

delete "idxqry.o" file and then type "make para" to compile "idxqry" w. multithreading

  ./idxqry gov2.sorted.s*.i 1mq.txt

run queries in file "1mq.txt" over the index of all gov2 partitions "gov2.sorted.s00.i - gov2.sorted.s07.i".

See benchmark "icbench" program for "integer compression" usage examples. In general encoding/decoding functions are of the form:

char *endptr = encode( unsigned *in, unsigned n, char *out, [unsigned start], [int b])
endptr : set by encode to the next character in "out" after the encoded buffer
in : input integer array
n : number of elements
out : pointer to output buffer
b : number of bits. Only for bit packing functions
start : previous value. Only for integrated delta encoding functions

char *endptr = decode( char *in, unsigned n, unsigned *out, [unsigned start], [int b])
endptr : set by decode to the next character in "in" after the decoded buffer
in : pointer to input buffer
n : number of elements
out : output integer array
b : number of bits. Only for bit unpacking functions
start : previous value. Only for integrated delta decoding functions

header files to use with documentation:

• Linux: GNU GCC (>=4.6)
• clang (>=3.2)
• Windows: MinGW-w64 (no parallel query processing)

• All TurboPFor integer compression functions are thread safe

• FastPFor + Simdcomp: SIMDPack FPF, Vbyte FPF, VarintG8IU, StreamVbyte
• Optimized Pfor-delta compression code: OptPFD/OptP4, Simple16 (limited to 28 bits integers)
• MaskedVByte. See also: Vectorized VByte Decoding
• Index Compression Using 64-Bit Words: Simple-8b (speed optimized version tested)
• libfor
• Compression, SIMD, and Postings Lists QMX integer compression from the "simple family"
• lz4. included w. block size 64K as indication. Tested after preprocessing w. delta+transpose
• blosc. blosc is like transpose/shuffle+lz77. Tested blosc+lz4 and blosclz incl. vectorizeed shuffle.
  
• Document identifier data set
• Integer compression publications:
  • SIMD Compression and the Intersection of Sorted Integers
  • Partitioned Elias-Fano Indexes
  • On Inverted Index Compression for Search Engine Efficiency
  • Google's Group Varint Encoding

Last update: 19 JUN 2016