this is forked from paq8pxd. i include the source, maybe the result is 15.9xx.xxx bytes for enwik8 using -s14 option.
this is forked from paq8pxd. i include the source, maybe the result is 15.9xx.xxx bytes for enwik8 using -s14 option.
Comparing to v47_6 the difference is this:
Code:1c1 < /* paq8pxd file compressor/archiver. Release by Kaido Orav, Mar. 4, 2018 --- > /* paq8lab file compressor/archiver. forked from paq8pxd 552c552 < #define PROGNAME "paq8pxd47_6" // Please change this if you change the program. --- > #define PROGNAME "paq8lab 1.0" // Please change this if you change the program. 5741a5742,5750 > > cm.set(hashxxk(507,lastLetter,x.col,x.frstchar,x.spaces)); > cm.set(hashxxk(508,x.frstchar,x.spaces,x.col,wordGap)); > cm.set(hashxxk(509,text0,x.col,x.spaces)); > > > cm.set(hashxxk(510,wrdhsh,lastUpper<x.wordlen,lastLetter,(lastUpper<x.col))); > cm.set(hashxxk(511,wrdhsh,firstLetter,lastLetter,(lastUpper<x.col)));// >
Mike (21st November 2019)
@bwt - Could you compile Win64 binary?
And second question: why you didn't use paq8pxd v48_bwt4 as a base?
For paq8pxd v48_bwt4 I've got such scores:
16'429'216 - enwik8 -s6 by Paq8pxd_v48_bwt4 (your score)
16'183'320 - enwik8 -s8 by Paq8pxd_v48_bwt4
15'977'376 - enwik8 -s14 by Paq8pxd_v48_bwt4 - the result was below 16'000'000 bytes.
16'091'113 - enwik8.drt -s14 by Paq8pxd_v48_bwt4
I've tried ywo times. Unfortunatelly both attempts ends without success. This version uses more than 32GB of memory and (besides much more timeconsuming) crashes at about 80% of the file - I don't know if it was in the same moment...
My last scores are:
16'525'916 - enwik8 -s6 by Paq8pxd_v48_bwt1 - your score
16'248'408 - enwik8 -s8 by Paq8pxd_v48_bwt1
16'004'759 - enwik8 -s14 by Paq8pxd_v48_bwt1
16'088'990 - enwik8.drt -s14 by Paq8pxd_v48_bwt1
136'732'561 - enwik9 -s6 by Paq8pxd_v48_bwt1 - your score
126'183'029 - enwik9_1423 -s14 by Paq8pxd_v48_bwt1 - the paq series record, submitted to LTCB as 4'th score at now! With -s15 option program crashes...
16'472'556 - enwik8 -s6 by Paq8pxd_v48_bwt2
16'228'952 - enwik8 -s8 by Paq8pxd_v48_bwt2
16'001'495 - enwik8 -s14 by Paq8pxd_v48_bwt2
16'088'506 - enwik8.drt -s14 by Paq8pxd_v48_bwt2
126'157'295 - enwik9_1423 -s14 by Paq8pxd_v48_bwt2 - estimate, program didn't finish even on -s14
16'429'216 - enwik8 -s6 by Paq8pxd_v48_bwt4
16'183'320 - enwik8 -s8 by Paq8pxd_v48_bwt4
15'977'376 - enwik8 -s14 by Paq8pxd_v48_bwt4
16'091'113 - enwik8.drt -s14 by Paq8pxd_v48_bwt4
125'967'139 - enwik9_1423 -s14 by Paq8pxd_v48_bwt4 - very goooood result but as above it's estimate, program didn't finish even on -s14
It compiles with parts other than .cpp taken from this: https://github.com/kaitz/paq8pxd/releases/tag/v47
Darek (23rd November 2019)
No, there were binaries published on paq8pxd forum.
Like mentioned above v47_6 - https://encode.su/threads/1464-Paq8p...ll=1#post57094
Darek (23rd November 2019)
this is based on paq8pxd48_bwt1
Might be too new... try getting gcc7 on mingw-w64.org
I used this to compile it
bwt (25th November 2019)
1. Get this: https://vorboss.dl.sourceforge.net/p...-rt_v5-rev1.7z
2. Install to C:\mingw720x
3. run g.bat in archive above
i use devc++ 5.11
I've tried to compress with -s14 and -s15 options and there are "out of memory" error. This version even not tried to use swap file...
-s14 option uses, as it wrote, 42GB of memory and (strange) -s15 try to use 39GB of memory.
Instead of this I've testes enwik8 with -s6, -s8 and -s13. Comparison to latest version is like follows:
16'553'942 - enwik8 -s6 by Paq8pxd_v47_5
16'267'313 - enwik8 -s8 by Paq8pxd_v47_5
16'027'128 - enwik8 -s15 by Paq8pxd_v47_5
16'099'328 - enwik8.drt -s15 by Paq8pxd_v47_5
126'387'497 - enwik9_1423 -s15 by Paq8pxd_v47_5 - estimated, program finish to compress but didn't add the header..
16'550'555 - enwik8 -s6 by Paq8pxd_v47_6
16'264'217 - enwik8 -s8 by Paq8pxd_v47_6
16'525'916 - enwik8 -s6 by Paq8pxd_v48_bwt1
16'248'408 - enwik8 -s8 by Paq8pxd_v48_bwt1
16'004'759 - enwik8 -s14 by Paq8pxd_v48_bwt1
16'088'990 - enwik8.drt -s14 by Paq8pxd_v48_bwt1
136'732'561 - enwik9 -s6 by Paq8pxd_v48_bwt1
126'183'029 - enwik9_1423 -s14 by Paq8pxd_v48_bwt1
16'472'556 - enwik8 -s6 by Paq8pxd_v48_bwt2
16'228'952 - enwik8 -s8 by Paq8pxd_v48_bwt2
16'001'495 - enwik8 -s14 by Paq8pxd_v48_bwt2
16'088'506 - enwik8.drt -s14 by Paq8pxd_v48_bwt2
126'157'295 - enwik9_1423 -s14 by Paq8pxd_v48_bwt2 - estimated due to crashes
16'429'216 - enwik8 -s6 by Paq8pxd_v48_bwt4
16'183'320 - enwik8 -s8 by Paq8pxd_v48_bwt4
15'977'376 - enwik8 -s14 by Paq8pxd_v48_bwt4
16'091'113 - enwik8.drt -s14 by Paq8pxd_v48_bwt4
125'967'139 - enwik9_1423 -s14 by Paq8pxd_v48_bwt4 - estimated due to issues
16'526'213 - enwik8 -s6 by Paq8lab_2
16'249'207 - enwik8 -s8 by Paq8lab_2
16'024'588 - enwik8 -s13 by Paq8lab_2
That looks that paq8lab_2 version is very similar to bwt1 (slightly worse), then my estimate for -s14/15 is as below:
16'005'546 - enwik8 -s14 by Paq8lab_2 - estimate - unfortunatelly slightly higher than 15'9xx'xxx
16'108'923 - enwik8.drt -s14 by Paq8lab_2 - estimate
126'339'363 - enwik9_1423 -s14 by Paq8lab_2 - estimate
bwt (29th November 2019)
There's a commented out line above that:Also it'd compile if you didn't remove -DNDEBUG:Code://U64 scompsize=0;
Code:#ifndef NDEBUG printf("Stream(%d) block from %0lu to %0lu bytes\n",i,datasegmentlen, out->curpos()-scompsize); scompsize= out->curpos(); #endif
Change this:to this:Code://U64 scompsize=0;Code:U64 scompsize=0;
paq8pxdv48_bwt2
@shelwien could you compile this source code please ? maybe could you remove zlib n gif function so i can compile it my self please ? thank you