pZLib v3 (Final release)

[FitGirl]

Razor, thanks for cls, great update! One question: is it locked to 50p or 75p of available cores? And can it be adjusted like you did it in config for cls-precomp?

[Simorq]

Mad Max

pZLibv2:c32m+srep+Lzma2 4.36 GB

pZLibv3:c256m+srep+Lzma2 6.4 GB

pZLibv3:x:c32m+srep+Lzma2 4.26 GB

[Razor12911]

Originally Posted by Simorq

Mad Max

pZLibv2:c32m+srep+Lzma2 4.36 GB

pZLibv3:c256m+srep+Lzma2 6.4 GB

pZLibv3:x:c32m+srep+Lzma2 4.26 GB

That's not surprising.

Mad Mad is compressed using level 1, by default pzlib is level 6, or rather reflate, the moment you place raw2hif_dll.dll, you are disabling internal recompression, so this is what you will get.

ZLibv3:c256m+srep+Lzma2 6.4 GB
because level 6 was used, you didn't set reflating level, which is the -r# option.
Secondly, the option -x, enables both internal recompression to work with reflate. which is why 4.26GB is the best size because unlike reflate, internal recompression works like precomp, trial and error, it will try all levels and pick one that is best.

@Everyone
pzlib does work without raw2hif_dll.dll, only place the library if you know what it does, or how it works or how it is used else you'll just end up with negative ratios.

[Razor12911]

CLS updated

-Added confirguration
-There a few noobs out there, so hif2raw_dll.dll is embedded in cls, it's loaded automatically, no need to place near it near the CLS, at least this could rule out a few errors.

[FitGirl]

Tested new updated CLS - it works good, but the bug from pzlib.exe still exists - it's impossible to cancel unpacking. Before it was possible to kill pzlib.exe process, now the whole setup must be shut down.

[FitGirl]

A friend of mine did more tests of cls vs generic pzlib. Decoding speed.

          |                       t1 |                       t2 |                       t4 |                        t7 |
----------*--------------------------*--------------------------*--------------------------*---------------------------*
pzlib_x86 |  192 sec. / 2613 kB/sec. |  118 sec. / 4249 kB/sec. | 78.3 sec. / 6419 kB/sec. | 61.7 sec.  / 8144 kB/sec. |
pzlib_x64 |  141 sec. / 3556 kB/sec. | 87.5 sec. / 5745 kB/sec. | 57.9 sec. / 8680 kB/sec. | 44.1 sec. / 11386 kB/sec. |
cls-pzlib |  198 sec. / 2537 kB/sec. |  119 sec. / 4223 kB/sec. | 78.3 sec. / 6422 kB/sec. | 61.5 sec.  / 8169 kB/sec. |

enc opts: pzlib:m2

win7_sp1_x64
i7-4700MQ / 8 GB RAM
RAM-Disk

[Razor12911]

CLS updated

-Cancel bug fixed
-Added multi threading user control example

[Razor12911]

Originally Posted by FitGirl

A friend of mine did more tests of cls vs generic pzlib. Decoding speed.

          |                       t1 |                       t2 |                       t4 |                        t7 |
----------*--------------------------*--------------------------*--------------------------*---------------------------*
pzlib_x86 |  192 sec. / 2613 kB/sec. |  118 sec. / 4249 kB/sec. | 78.3 sec. / 6419 kB/sec. | 61.7 sec.  / 8144 kB/sec. |
pzlib_x64 |  141 sec. / 3556 kB/sec. | 87.5 sec. / 5745 kB/sec. | 57.9 sec. / 8680 kB/sec. | 44.1 sec. / 11386 kB/sec. |
cls-pzlib |  198 sec. / 2537 kB/sec. |  119 sec. / 4223 kB/sec. | 78.3 sec. / 6422 kB/sec. | 61.5 sec.  / 8169 kB/sec. |

enc opts: pzlib:m2

win7_sp1_x64
i7-4700MQ / 8 GB RAM
RAM-Disk

Well it's not surprising, cls is actually x86, it should produce roughly the same speed as pzlib x86.

[danswano]

Originally Posted by Simorq

Mad Max

pZLibv2:c32m+srep+Lzma2 4.36 GB

pZLibv3:c256m+srep+Lzma2 6.4 GB

pZLibv3:x:c32m+srep+Lzma2 4.26 GB

What are your srep/lzma2 parameters you got 4.36GB with?
Please provide .bat and arc.ini parameters.
Thanks.

[Simorq]

Originally Posted by danswano

What are your srep/lzma2 parameters you got 4.36GB with?
Please provide .bat and arc.ini parameters.
Thanks.

pzlib+srep+lzma2 4.26 GB
Arc.ini

[External compressor:pzlib3]
header = 0
packcmd   = pZLib e -m2 -x -c32m -t25p - -o - <stdin> <stdout>
unpackcmd = pZLib d -t100p - -o - <stdin> <stdout>

[External compressor:pzlib2]
header = 0
packcmd   = pZLib e -m2 -c32m -t25p - -o - <stdin> <stdout>
unpackcmd = pZLib d -t100p - -o - <stdin> <stdout>


[External compressor:srep]
header    = 0
;packcmd   = srep {options} -a1 -m5f -l256 -t8 $$arcdatafile$$.tmp $$arcpackedfile$$.tmp
packcmd   = srep {options} -m5f $$arcdatafile$$.tmp $$arcpackedfile$$.tmp
unpackcmd = srep -d $$arcpackedfile$$.tmp $$arcdatafile$$.tmp

[External compressor:lzma2, Lzma2, LZMA2, XZ, xz]
header = 0
packcmd   = 7z a -txz -an -mcrc=0 -m1=lzma2:d25:fb=273:mf=bt4:mc=1000000:lc=4:lp=0 -mmt=on -mx9 -si -so <stdin> <stdout>
unpackcmd = 7z x -txz -mmt=on -an -y -si -so <stdin> <stdout>

Test pzlib2

[FitGirl]

Razor, thank you for the latest update - now the installer closes properly!
You're a gift to this community!

[pakrat2k2]

Originally Posted by KaktoR

Someone on the previous pages telling not to use <stdin> - <stdout>. Then you should use all settings without crc error. But honestly, i don't know how to use this.

change arc.ini to use the commands listed ( parts without crc error )

[LuzhinE]

With t25p and t75p works well until now free of CRC, before there was that problem to cancel and did not, do not comment because Razor12911 is busy in classes, but now if it works well.

But waiting for an improvement so that the CPU can be used to maximum when compressing t8.

Thank you very much

[Razor12911]

Version 3.2 now available

Changes

-Tried to fix crc errors, no idea if they are fixed or not.
-Removed several options because there are just to many for some people to understand what they do.
-Tried to fix decoding termination, in case you want to cancel decoding process, it must now work without killing process, not sure either.
-method setting removed, from now it just searches for deflate streams which is m2 by default.
-by default reflate and zlib are blended.
-special stream option "s" is removed, found a way to process them.

[Simorq]

Deflate Streams Better.

Test (Act Of Aggression Reboot Edition Deflate Streams)
pZLib version 2 and 3 Not Support
pZLib2017 Support