Squeeze Chart

[joerg]

@Stephan Busch:

2017 Mar 01 ZSTD 1.1.2 - slightly better compression than previous version

why you are using the older ZSTD 1.1.2 ?

and not the newer version 1.1.3 (february 2017) from

https://github.com/facebook/zstd/releases
https://github.com/facebook/zstd/rel....1.3-win32.zip
https://github.com/facebook/zstd/rel....1.3-win64.zip

best regards

[Stephan Busch]

I wanted to make an update today.
ZSTD 1.1.3 is on my to-do-list (among others)

[Stephan Busch]

Well folks, here's another update:

2017 Jun 30 David Allen Scott passed away on January, 26th 2016 - our condolences go to his family, friends and students
2017 Jun 30 added new testset chart: EXE
2017 Jun 30 added new testsets: more camera raw formats in 'PDF, JPG, MP3, PNG, Installer.. chart
2017 Jun 30 PackRAW 0.0.1, PackRAW 0.0.2, PackRAW 0.1 ('camera raw arw' chart, 'PDF, JPG, MP3, PNG, Installer.. chart) - outstanding work, Márcio
2017 Jun 30 PowerArchiver 17.00.80 - reaches #22 of the chart, that's a giant climb up compared to previous version - much better compression
2017 Jun 30 JamPack 0.70 - interesting idea
2017 Jun 30 ZSTD 1.1.4
2017 Jun 30 ZSTDMT 1.1.4
2017 Jun 30 Lizard 1.0
2017 Jun 30 LibDeflate 0.7
2017 Jun 30 LZG 1.0.8
2017 Jun 30 Snappy 1.1.4
2017 Jun 30 EMMA 0.1.23 ('camera raw' chart, 'PDF, JPG, MP3, PNG, Installer.. chart)
2017 Jun 30 UPX 3.93 ('exe' chart)

I also added Márcio to the Hall of Fame because of his PackRAW creation.

Am I the only one here that is happy PackRAW was created?
I really appreciate his work which was done in countless hours
and I am feeling honored to test and use it - especially in times,
where Sachin Garg (Rawzor author), Dave Coffin (DCRAW author, Klaus Post (RawSpeed author)
do not answer any of my mails (I assume they keep ignoring them) for whatever reason.

[Stephan Busch]

Does anybody still have BruteCM v0.1d written by Vcore in 2008?
https://encode.su/threads/1031-bruteCM-compressor/page2

I couldn't compile those.. what about you?

Slimox https://github.com/dolphinking/slimox
Zmolly https://github.com/richox/zmolly

[Shelwien]

Slimox more or less worked after some patches (at least for a single file)
Zmolly compiles, but crashes during encoding, dunno why - I tried IntelC and cygwin/gcc (mingw/gcc doesn't have std::thread).

[Stephan Busch]

hey guys,
its time for another update.
Here you go:

2017 Aug 25 PackRAW 0.2 ('PDF, JPG, MP3, PNG, Installer.. chart) - outstanding work, Márcio
2017 Aug 25 PAQ8pxd_v28 ('TEXT (Bibles)' chart) - many new world records on text compression - well done, Kaido
2017 Aug 25 PAQ8pxd_v25 ('exe', 'image', 'TEXT (Bibles)', 'PDF, JPG, MP3, PNG, Installer.. chart) - some new world records - well done, Márcio, Jan & Kaido
2017 Aug 25 PAQ8pxd_v23 ('slower than practical' chart) - some new world records, reaches rank 2 of the chart - well done, Márcio, Jan & Kaido
2017 Aug 25 PAQ8px_v99 ('PDF, JPG, MP3, PNG, Installer.. chart) - some new word records
2017 Aug 25 PAQ8px_v95 ('image' chart) - some new world records
2017 Aug 25 PAQ8px_v92 ('image' chart)
2017 Aug 25 PAQ8px_v87 ('slower than practical' chart) - some new world records, reaches rank 1 of the chart - well done, Márcio & Jan
2017 Aug 25 PAQ8px_v79 ('slower than practical' chart)
2017 Aug 25 PAQ8px_v77 ('slower than practical' chart)
2017 Aug 25 EMMA 0.1.24 ('slower than practical' chart) - some new world records, reaches rank 6 of the chart - well done, Márcio
2017 Aug 25 FP8_v5 ('slower than practical' chart) - nice improvements
2017 Aug 25 CMV 0.1.1 ('slower than practical' chart) - interesting CM, Mauro
2017 Aug 25 Quark 0.96r ('slower than practical' chart)
2017 Aug 25 PowerArchiver 17.00.90 - reaches #12 of the chart
2017 Aug 25 CM4 Extended
2017 Aug 25 CM4 - pretty fast CM - well done, Nauful
2017 Aug 25 CM5
2017 Aug 25 ZSTD 1.3.0
2017 Aug 25 Crush Codec
2017 Aug 25 Slimox
2017 Aug 25 PX
2017 Aug 25 Zmolly 0.1
2017 Aug 25 RH4
2017 Aug 25 PPM o4
2017 Aug 25 ULZ 0.06
2017 Aug 25 RCM
2017 Aug 25 HeatShrink 0.4.1
2017 Aug 25 FPC
2017 Aug 25 Chameleon
2017 Aug 25 Nakamichi - all versions I tried failed at end of audio testset (in tar)
2017 Aug 25 ABC 2.4 ('TEXT (Bibles)' chart)
2017 Aug 25 GLZA 0.9 ('TEXT (Bibles)' chart) - much faster than previous version - well done, Kennon

[Stephan Busch]

Hello guys,

here's another update:

2017 Oct 18 Razor Archiver 1.00/1.01 - reaches rank 11 of the chart - very well done and welcome back, Christian
2017 Oct 18 ZSTD 1.3.2
2017 Oct 18 PAQ8px_v113 ('image' chart) - some new world records - well done, Márcio
2017 Oct 18 PAQ8pxd_v34 ('image' chart) - some new world records - well done, Márcio & Kaido
2017 Oct 18 PAQ8pxd_v32 ('slower than practical' chart) - some new world records, reaches rank 1 of the chart - well done, Márcio, Jan & Kaido
2017 Oct 18 PAQ8px_v112 ('slower than practical' chart) - some new world records, reaches rank 2 of the chart - well done, Márcio & Jan
2017 Oct 18 PAQ8px_v110 ('slower than practical' chart)
2017 Oct 18 EMMA 0.1.25 ('slower than practical' chart) - some new world records, reaches rank 9 of the chart - well done, Márcio
2017 Oct 18 PowerArchiver 17.01.04
2017 Oct 18 PowerArchiver 17.00.92
2017 Oct 18 CHC 1.0

At present, I am testing CMV 0.2.0

[Christian]

Hi Stephan,

thank you very much.

Your main test set is ~5G and you measure 648s decompression time for rz. This is less than 10M/s. What happened? rz should be much faster.

[Stephan Busch]

Hi Christian,

this is the time I measured on my system.
Most time seems to be needed to decompress sources and xml testsets.
I'll double-check it and report back.

[Mauro Vezzosi]

I saw that in the column "algorithms used" of CMV there is "CM + filter": CMV doesn't have filters and doesn't detect any file type, if you want to include/exclude a specific model (e.g. for exe) you must switch it on/off directly or indirectly.

[Stephan Busch]

@ Christian

the timings are slower than expected because of a drive issue.
I will update Razor timings with the next chart update.

[Stephan Busch]

Here's the last update until xmas season.
Wish you a merry christmas.


2017 Dec 20 PAQ8px_v120 ('exe', 'image' & 'slower than practical' chart) - some new world records - well done, Márcio
2017 Dec 20 PAQ8px_v119 ('slower than practical' chart) - some new world records - well done, Márcio
2017 Dec 20 PAQ8px_v116 ('PDF, JPG, MP3, PNG, Installer.. chart) - some new word records - well done, Márcio

[Stephan Busch]

Hey guys,

here's another update.

2018 Mar 14 added new testset chart: Video - this introduces a lossless video compression benchmark
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala & Dr. Zoltán Szabadka (created PIK image format)
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala & Dr. Zoltán Szabadka (created Brotli)
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala & Dr. Zoltán Szabadka & Robert Obryk (created Guetzli)
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala & Lode Vandevenne (created Zopfli)
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala & Dr. Rastislav Lenhardt (created Gipfeli)
2018 Mar 14 Hall of Fame: added Jyrki Alakuijala (created WEBP)
2018 Mar 14 Hall of Fame: added Balázs Oroszi (created lossless video codec MagicYUV)
2018 Mar 14 Hall of Fame: added Takeshi Umezawa (created lossless open-source video codec based on HuffYUV (UT Video Codec))
2018 Mar 14 Hall of Fame: added Miroslav Linek (created lossless open-source video codec (MLC))
2018 Mar 14 Hall of Fame: added Maxim Smirnov (created lossless video codec (YUVsoft Lossless Video Codec)))
2018 Mar 14 Hall of Fame: added Ben Greenwood (created lossless open-source video codec (Lagarith))
2018 Mar 14 Hall of Fame: added Dr. Dmitry S. Vatolin & Dmitry Popov (created lossless video codec (MSU))
2018 Mar 14 Hall of Fame: added Michael Niedermayer (created lossless open-source video codec (FFV1))
2018 Mar 14 Hall of Fame: added Ben Rudiak-Gould (invented lossless open-source video codec (HuffYUV))
2018 Mar 14 PAQ8px_v140 ('TEXT (Bibles)' chart) - some new world records - well done, Márcio
2018 Mar 14 PAQ8px_v137 ('slower than practical' chart) - some new world records - well done, Márcio
2018 Mar 14 PAQ8pxd_v47 ('TEXT (Bibles)' chart) - some new world records - well done, Márcio & Kaido
2018 Mar 14 PAQ8pxd_v46 ('slower than practical' chart) - some new world records - well done, Márcio & Kaido
2018 Mar 14 OGGRE 0.1.1 ('PDF, JPG, MP3, PNG, Installer.. chart) - new word record on lossless compression of .OGG - well done, ProFrager
2018 Mar 14 WEBP 0.6.1 ('image' chart) - a bit better compression than previous version
2018 Mar 14 BCM 1.30 - faster & better compression than previous version - well done, Ilya
2018 Mar 14 Radyx 0.92 - about the same performance as previous version
2018 Mar 14 LOLZ 21a7 - better than LZMA compression but much slower
2018 Mar 14 SQUID 0.0.2 - better compression than previous version - well done, Ilya

[hexagone]

Any chance you could run kanzi (level 5) against your test suite ? https://encode.su/threads/2083-Java-...ighlight=kanzi
I am curious to see where it would land...

[Jyrki Alakuijala]

Thanks for the update!

Any chance you could try large-window brotli for the squeeze chart? It is now easier to try than about 20 months ago when it was introduced. The large-window mode is no longer a separate branch in github, but has been added to the main repository.

Enwik9 gets an 11 % density improvement and there is practically no impact on the decode speed. A lot more memory is needed at decoding because brotli decoder is with streaming I/O, but not more than LZMA and ZStandard are using in streaming decoding. I believe that large window brotli will be sitting firmly (with quite a huge density margin) on the pareto-optimal front of density-decoding speed, and it is not ideal that it is not part of a comparison of current compression algorithms.

[Stephan Busch]

yes I'll test large window brotli, Jyrki.
Thanks for the info.

And I'll also take a look on kanzi.
Thanks for pointing me to that, hexagone.

[Stephan Busch]

Any chance you could run kanzi (level 5) against your test suite ? https://encode.su/threads/2083-Java-...ighlight=kanzi
I am curious to see where it would land...

there is no recent executable of Kanzi

[hexagone]

The one in the link is from January (fairly recent and latest release).
I can send you a more recent one but the code is in flux until the next release, so I am not sure it would be meaningful.

[Stephan Busch]

As the last post seems to have vanished, I'll repost:

The old laptop used for SqueezeChart benchmarks keeps freezing which results in many restarts.
A new laptop is planned, and when it is chosen, many many timings have to be measured again,
so please be patient.

The website hosted at http://www.squeezechart.com also needs some refactoring - even after cleaning it up, I am not happy with it.
Jan Ondrus wrote it for me many years ago and I liked it, but I cannot figure out how to move logo, change fonts, font colors, menu color, menu font...

What do you think?
What's good there and what's not? What is missing?

Please feel free to post your suggestions or sample templates here or send them to me by email.

[SolidComp]

As the last post seems to have vanished, I'll repost:

The old laptop used for SqueezeChart benchmarks keeps freezing which results in many restarts.
A new laptop is planned, and when it is chosen, many many timings have to be measured again,
so please be patient.

The website hosted at http://www.squeezechart.com also needs some refactoring - even after cleaning it up, I am not happy with it.
Jan Ondrus wrote it for me many years ago and I liked it, but I cannot figure out how to move logo, change fonts, font colors, menu color, menu font...

What do you think?
What's good there and what's not? What is missing?

Please feel free to post your suggestions or sample templates here or send them to me by email.

Your fonts are at the top of this file: http://www.squeezechart.com/style.css

Your menu and body fonts are both Arial. A better font stack would be Calibri, Roboto, Helvetica, sans-serif;

Calibri is nice on Windows and there's never any point in having Arial in the stack – it's the default backup sans-serif already, in most browsers at least on desktop. Roboto will give you a better look on Android, though it might default to that anyway (or Droid Sans). If you want fancier fonts, you might check out Google Fonts or Adobe Typekit.

Further down the css file you'll see the header and headerbody selectors, which control your logo placement.

[joerg]

@Stephan Busch:

your website looks good

- but why there is not a sub-website, which is showing the benchmark results as html?

- i would propose you to use only one font within your website
- i would propose you "Segoe UI"
- this font seems to be free usable ...

- the german and the english versions are not identical
- several links do not work

- i think, it would be better
not showing big pictures or animation on the startpage
and if, then only one animation on the startpage

- i think the main menu should be designed identical
in "english" and in "german" - for example:

"Start" "Hall of Fame" "LINKS" "support" "email"


--- --- --- --- ---
- "The Hall of Fame":
-
-- name wiki-link
-- foto
-- * 1592 April 22 - where
-- † 1635 October 23 - where
-- what he has done
-
--- --- --- --- --- --- ---
KEEP IT SHORT & SIMPLE
--- --- --- --- --- --- ---

best regards

of course it is your website ...
just my two cents

[hexagone]

Some links on the site are messed up.