Shelwein's REFLATE and &7zip

[Gupta]

Hey Guys!
Here is a "WORKING PUBLIC" Reflate from SHELWEIN and
a 7zip that support ppmd_sh, plzma, the most important for us REFLATE

REMARKS

+ REFLATE
+ A working recompressor of any raw deflate streams
+ Faster
+ Better
+ No Temps

Command line uses

For Re-compression
reflate [c#] File .out .hif
c is for encode and d is for decode
# level u want to use (there is no level detection yet)
for recursion use something like that "c9876" , upto 7
.out is the file containing re-compressed
.hif is the recompressor's diff - information required for lossless
restoration of original deflate streams

For Restoration
reflate [d] .out .hif File
*everything is self-explanatory

Reflate-7zip is also good and fast
You can see batch file for help

CLS-Reflate for Shelwien's release (RAZOR12911)
FOR INNO Implementation Check post #3

[~MAK~]

ahh, I hope someone will find it useful, me and shelwien are still working to make it better :P

EDIT: But I will say, that I do tests and tell where he needs to change/improve the code, I didn't coded it, so I give Full credit to Shelwien only

[Gupta]

Here is reflate with INNO
1. Extract the archive anywhere
2. Go To Addons folder and run start.exe
3. Follow the instructions
4. Compress *.hif and *.out files with Freearc
5. Change Archives.ini as according to ur bins
and compile the script

[Gupta]

testfile Character.vpp_pc [595,921,053 bytes]

reflate output
712,401,303 bytes // -m0=reflate:x5
707,211,609 bytes // -m0=reflate:x9
595,921,058 bytes // zlib_jo e
595,921,087 bytes // precomp045_64 -intense
595,921,103 bytes // pzlib64 e -t4 -t75p

Compressed to
578,359,654 bytes // -m0=plzma4:a1:mt2

542,402,038 bytes // -m0=reflate:x5 -m1=plzma4:a1:mt2 -m2=lzma:d19 -mb0s0:1 -mb0s1:2
542,484,178 bytes // -m0=reflate*4:x5 -m1=plzma4:a1:mt2 -m2=lzma:d19 -m3=lzma:d19 -m4=lzma:d19 -m5=lzma:d19
542,484,241 bytes // -m0=reflate*7:x5 -m1=plzma4:a1:mt2 -m2=lzma:d19 -m3=lzma:d19 -m4=lzma:d19 -m5=lzma:d19 -m6=lzma:d19 -m7=lzma:d19 -m8=lzma:d19
470,730,352 bytes // -m0=reflate:x5 + srep64 -m5f -a8 reflate.7z - | 7z.exe -m0=lzma2:d19

537,347,805 bytes // -m0=reflate:x9 -m1=plzma4:a1:mt2 -m2=lzma:d19
537,444,599 bytes // -m0=reflate*4:x9 -m1=plzma4:a1:mt2 -m2=lzma:d19 -m3=lzma:d19 -m4=lzma:d19 -m5=lzma:d19
537,444,662 bytes // -m0=reflate*7:x9 -m1=plzma4:a1:mt2 -m2=lzma:d19 -m3=lzma:d19 -m4=lzma:d19 -m5=lzma:d19 -m6=lzma:d19 -m7=lzma:d19 -m8=lzma:d19
466,050,868 bytes // -m0=reflate:x9 + srep64 -m5f -a8 reflate.7z - | 7z.exe -m0=lzma2:d19

NOTE Razor's reflate can't restore file correctly so i don't mention testing it

[ChronoCross]

HI i have a 2 problem here with 7zdll_v7 when i use reflate filter the 7z.exe crash!
and the rawfilt_v1j says i need an INTEL prosessor :S

[Gupta]

Originally Posted by ChronoCross

HI i have a 2 problem here with 7zdll_v7 when i use reflate filter the 7z.exe crash!
and the rawfilt_v1j says i need an INTEL prosessor :S

May be this is helpfull
https://github.com/Shelwien/ic64patch/releases

[FitGirl]

Originally Posted by PrinceGupta2000

May be this is helpfull
https://github.com/Shelwien/ic64patch/releases

Well, it may help on newer CPUs, but for AMD Phenoms I/II or older Core 2 Duo/Quad it still doesn't work after patching, IDK why. Maybe some modern SSE instructions are used.

I've made a test with my AMD/Intetl users, 30+ people. Only Phenoms/C2D/Q failed.

[ChronoCross]

7z v8 slpit ok plzma ok lzma ok but reflate filter reflateN_dll.dll crash 7z app!. And rawfilt v1k crash with the same thing. So the problem is the dll and i don't know why. I was patch the dll correctly but the problem persist so is not a problem with my hardware the problem is the compiler i guess.

[Gupta]

Originally Posted by ChronoCross

7z v8 slpit ok plzma ok lzma ok but reflate filter reflateN_dll.dll crash 7z app!. And rawfilt v1k crash with the same thing. So the problem is the dll and i don't know why. I was patch the dll correctly but the problem persist so is not a problem with my hardware the problem is the compiler i guess.

Yes Shelwein uses Intel's compiler for reflate and it has anti amd checks . may be because it is faster

Originally Posted by Shelwein

i7 930 @ 2.8ghz, 2008R2
c = reflate c all_icons_by_Icons8_free.zip .out .hif1 .hif2
d = reflate d .out .hif1 .hif2 .unp
filesize = 80,398,442, scan depth = 2

(c) (d)
147.811s 115.690s // IC17x64_SSE4
148.731s 115.270s // IC16x64_SSE4
149.683s 114.021s // VS2015u3x64_SSE4 PGO
154.269s 115.160s // VS2015x64_SSE4 PGO
152.584s 116.236s // VS2015x64_SSE4
155.455s 115.409s // clang400x64_SSE4 PGO
155.580s 119.060s // VS2015u3x64_SSE4
157.015s 116.471s // clang400x64_SSE4
161.851s 129.918s // gcc611x64_SSE4
161.289s 130.027s // gcc620x64_SSE4
171.960s 129.933s // IC11x64_SSE4
172.490s 132.087s // IC11x64_SSE2

211.880s 149.730s // clang400x32_SSE4
215.469s 150.494s // clang400x32_SSE4 PGO
212.661s 152.631s // VS2015x32_SSE4
216.904s 157.982s // VS2015u3x32_SSE4
243.315s 195.376s // gcc611x32_SSE4
235.281s 182.740s // gcc700x32_SSE4
239.509s 191.896s // IC17x32_SSE4

265.919s 203.644s // IC17x32_ia32
299.038s 229.743s // IC11x32_ia32

[ChronoCross]

Ouch! Very bad for me.
So intel only right?
That's racist! LOL.

[Razor12911]

Originally Posted by ChronoCross

Ouch! Very bad for me.
So intel only right?
That's racist! LOL.

You can just head over to encode and request MSC compiled program instead ICC.