This is a cache of https://sourceforge.net/p/sevenzip/discussion/45797/thread/b95432c7ac/. It is a snapshot of the page as it appeared on 2025-09-03T12:15:56.584+0200.
7-Zip / Discussion / Open Discussion: 7-Zip 24.09
Menu ▾ ▴

7-Zip 24.09

Open Discussion
Igor Pavlov
2024-11-30
2025-06-22
1 2 3 > >> (Page 1 of 3)

  • Igor Pavlov

    Igor Pavlov - 2024-11-30

    7-Zip 24.09 was released.

    Download

    7-Zip for 64-bit Windows x64:
    https://7-zip.org/a/7z2409-x64.exe

    7-Zip for 32-bit Windows x86:
    https://7-zip.org/a/7z2409.exe

    7-Zip for 64-bit Windows ARM64:
    https://7-zip.org/a/7z2409-arm64.exe

    7-Zip Extra: standalone console version, 7z DLL, Plugin for Far Manager:
    https://7-zip.org/a/7z2409-extra.7z

    7-Zip (console version) for 64-bit linux x86-64 (AMD64):
    https://7-zip.org/a/7z2409-linux-x64.tar.xz

    7-Zip (console version) for 32-bit linux x86:
    https://7-zip.org/a/7z2409-linux-x86.tar.xz

    7-Zip (console version) for 64-bit linux ARM64:
    https://7-zip.org/a/7z2409-linux-arm64.tar.xz

    7-Zip (console version) for 32-bit linux ARM:
    https://7-zip.org/a/7z2409-linux-arm.tar.xz

    7-Zip (console version) for macOS (ARM64 and x86-64):
    https://7-zip.org/a/7z2409-mac.tar.xz

    What's new in 7-Zip 24.09:

    • The default dictionary size values for LZMA/LZMA2 compression methods were increased:
             dictionary size   compression level
  v24.08  v24.09  v24.09   
          32-bit  64-bit    
    8 MB   16 MB   16 MB   -mx4
   16 MB   32 MB   32 MB   -mx5 : Normal
   32 MB   64 MB   64 MB   -mx6
   32 MB   64 MB  128 MB   -mx7 : Maximum
   64 MB   64 MB  256 MB   -mx8
   64 MB   64 MB  256 MB   -mx9 : Ultra

    The default dictionary size values for 32-bit versions of LZMA/LZMA2 don't exceed 64 MB.
    - 7-Zip now can calculate the following hash checksums: SHA-512, SHA-384, SHA3-256 and MD5.
    - APM and HFS support was improved.
    - If an archive update operation uses a temporary archive folder and the archive is moved to the destination folder, 7-Zip shows the progress of moving the archive file, as this operation can take a long time if the archive is large.
    - The bug was fixed: 7-Zip File Manager didn't propagate Zone.Identifier stream for extracted files from nested archives (if there is open archive inside another open archive).
    - Some bugs were fixed.

     
    🎉
    1
    👍
    11

    Last edit: Igor Pavlov 2024-11-30

    • lelik007

      lelik007 - 2024-11-30

      Igor, thank you for the new version. I tested every new method + Blake2sp, XXH64 like this:
      7z a -thash file.blake2sp folder
      and checked like this:
      7z t -thash file.blake2sp folder
      Everything is fine, but why does 7-zip use in this case linux path separator / for file.hash instead of the Windows one \ ? For compatibility? For the modern hashers it's nothing but is there any switch to change the separators in this case?

      And. yes,
      SHA-384, SHA-512 are + 50% speed to SHA-256,
      SHA3-256 is + 30% speed to SHA-256, without Intel SHA Extensions.
      And I think guys will thank you for MD5 later.

       

      • Igor Pavlov

        Igor Pavlov - 2024-11-30

        These hash text files formats are from linux. So we try to keep byte to byte compatibility with linux tools.
        To check that 7-zip works correctly, we must use any another program to check calculated hashes.
        To check created files in linux:

        sha384sum -c file.sha384
sha512sum -c file.sha512
sha3sum -c file.sha3-256
md5sum -c file.md5
         
        👍
        1

        Last edit: Igor Pavlov 2024-11-30

        • lelik007

          lelik007 - 2024-11-30

          TC for example doesn't mind linux separators, but I don't have linux even in WSL form to check the native commands.

           

          • Igor Pavlov

            Igor Pavlov - 2024-11-30

            Maybe someone should report TC developers about that problem.

             

            • lelik007

              lelik007 - 2024-12-01

              I meant that TC does OK, It can check the sums that were created with both commands:
              7z a file.sha384 folder (linux path separators)
              or
              7z h -ba -scrcsha384 -slfhn -xtd Folder > File.sha384 (Windows path separators)
              TC has a visual option to switch the separators for a sum creation, so It handles both fine.

               

              Last edit: lelik007 2024-12-01

    • wwp

      wwp - 2024-11-30

      Thanks!
      Does 24.09 fix CVE-2024-11477?

       

      • Igor Pavlov

        Igor Pavlov - 2024-11-30

        CVE-2024-11477 was fixed in 7-Zip v24.07 (2024-06-19)
        https://www.7-zip.org/history.txt

         
        👍
        2

        • wwp

          wwp - 2024-11-30

          Great that it was fixed so quickly, thanks @ipavlov!

           

    • lelik007

      lelik007 - 2024-12-01

      @ipavlov Igor, this command:
      7z a file.sfv folder
      is creating 7z file instead of .sfv
      7z a -thash file.sfv folder works as it should be.
      In the first case auto-detection for the extension doesn't work as I think.
      The other checksum extensions work correctly - 7z detects them.

       

      Last edit: lelik007 2024-12-01

    • lelik007

      lelik007 - 2024-12-01

      @ipavlov 7z i shows:
      0 256 231 SHA3-256
      there should be hash size in bytes:
      0 32 231 SHA3-256 as for:
      0 32 A SHA256

       

      • Igor Pavlov

        Igor Pavlov - 2024-12-01

        Yes, I'll fix it.
        Actually these numbers are for i command only. So real hash size is correct in another code.
        Thanks!

         
        👍
        1

    • arhfk

      arhfk - 2025-01-06

      every version of 7-zip is like a great milestone but version 24.09 is absolutely perfect.

       

  • Igor Pavlov

    Igor Pavlov - 2024-11-30

    SHA-512 TEST

    New version 7-Zip 24.09 supports hardware SHA-512 instructions.
    If you have Windows with Qualcomm X Elite / X Plus processor or latest Intel Core Ultra Lunar Lake / Arrow Lake (285K/265K/245K), please run the following test from command line and post the results here to forum thread:

    "C:\Program Files\7-Zip\7z" b -mmt1 -mm=hash -mtic=32 > 1.txt

    Windows for ARM64 doesn't provide a simple way to detect SHA-512 instructions extension. So 7-Zip parses registry key value

    HKEY_LOCAL_MACHINE\HARDWARE\DESCRIPTION\System\CentralProcessor\0
"CP 4030"

    And we need to check that it works correctly.
    If you have older arm64 processor, you also can run this test and post results to check that it correctly detects SHA-512 extension absence.

     

    Last edit: Igor Pavlov 2024-11-30

    • SteezCram

      SteezCram - 2025-01-22

      Here are the results on my laptop as an attachment. It seems to detect the SHA512 extension correctly.

       
      1.txt

      • Igor Pavlov

        Igor Pavlov - 2025-01-22

        Thanks!
        SHA512 looks ok.
        registry value was parsed: cp4030:1221111110212120:SHA256:SHA512

        Also some interesting about Snapdragon(R) X Elite pareformance, that CRC32:64 is only 9 GB/s - it means 3-cycles latency for crc32 instruction, like in Apple M1, and modern Cortex is 3 times faster.

         

      • Igor Pavlov

        Igor Pavlov - 2025-01-22

        Also I see that some integer benchmark workload results (like MD5,SHA1) are slow on X Elite. Apple M1 is faster.
        But maybe it's because of slow MSVC compiler.
        If you have WSL in that computer, please compare benchmarks for both Windows and linux version (via WSL) of 7-Zip:

        7z b -mmt1 > win.txt 
wsl ./7zz b -mmt1 > lin.txt
         

        Last edit: Igor Pavlov 2025-01-22

        • SteezCram

          SteezCram - 2025-02-12

          Sorry for the long reply. Here are the results.
          TL:DR: the linux results are higher but not that much (around 5-10% higher). I don't know if this expected.

          MSVC
7-Zip 24.09 (arm64) : Copyright (c) 1999-2024 Igor Pavlov : 2024-11-29

 mt1
Compiler:  MSC 1942.194234435
Windows 10.0 26100
ARM64 0 1.201 threads:12 128TB f:3804EB8C1004
Snapdragon(R) X Elite - X1E78100 - Qualcomm(R) Oryon(TM) CPU
(ARMv8 (64-bit) Family 8 Model 1 Revision 201) 2976 MHz
cp4030:1221111110212120:SHA256:SHA512

1T CPU Freq (MHz):  3402  3402  3404  3400  3400  3375  3395

RAM size:   32317 MB,  # CPU hardware threads:  12
RAM usage:    437 MB,  # Benchmark threads:      1

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       8286   100   8094   8061  |      97911   100   8393   8360
23:       6712    99   6902   6839  |      97262   100   8444   8419
24:       5879    99   6359   6322  |      94934    98   8542   8334
25:       5280    99   6074   6029  |      91658   100   8131   8158
----------------------------------  | ------------------------------
Avr:      6539    99   6857   6813  |      95441    99   8378   8318
Tot:              99   7617   7565

-----

linux (WSL)
7-Zip (z) 24.09 (arm64) : Copyright (c) 1999-2024 Igor Pavlov : 2024-11-28
 64-bit arm_v:8-A locale=C.UTF-8 Threads:12 OPEN_MAX:1024, ASM

 mt1
Compiler:  ver:9.2.1 20191025 GCC 9.2.1 : UNALIGNED
linux : 5.15.167.4-microsoft-standard-WSL2 : #1 SMP Tue Nov 5 00:19:58 UTC 2024 : aarch64
PageSize:4KB THP:always hwcap:3EBFFFFB:CRC32:SHA1:SHA2:SHA3:SHA512:AES:ASIMD hwcap2:316181
arm64

1T CPU Freq (MHz):  3650  3674  3675  3671  3673  3668   586

RAM size:   15846 MB,  # CPU hardware threads:  12
RAM usage:    437 MB,  # Benchmark threads:      1

                       Compressing  |                  Decompressing
Dict     Speed Usage    R/U Rating  |      Speed Usage    R/U Rating
         KiB/s     %   MIPS   MIPS  |      KiB/s     %   MIPS   MIPS

22:       7998   109   7141   7780  |     105189   109   8236   8981
23:       6957   108   6572   7089  |     104368   109   8286   9034
24:       6312   109   6249   6788  |     103720   109   8360   9106
25:       5695   108   6007   6503  |     101039   109   8255   8993
----------------------------------  | ------------------------------
Avr:      6740   108   6492   7040  |     103579   109   8284   9028
Tot:             109   7388   8034
           

          Last edit: Igor Pavlov 2025-02-12

          • Igor Pavlov

            Igor Pavlov - 2025-02-12

            please run another commands:

            7z b -mmt1 -mm=* -bt > win.txt 
wsl ./7zz b -mmt1 -mm=* -bt > lin.txt

            Also if your system supports x64 emulation (prism), unpack 7-Zip x64 installer to some folder, and run x64 version:

            7z b -mmt1 -mm=* -bt > win_x64.txt

            TL:DR: the linux results are higher but not that much (around 5-10% higher). I don't know if this expected.

            There is some error with timers in wsl:
            cpu usage column must be 100%.
            1T CPU Freq (MHz) freq must be 3400 instead of 3650.

             

            Last edit: Igor Pavlov 2025-02-12

            • SteezCram

              SteezCram - 2025-02-13

              Here are the results. I have put them in a single txt file. One thing I have noticied is that when I run the test, I do not have one CPU core max out at 100%. Instead I have two CPU cores at around 50% each.
              I do not know why I have this behavior with 7zip since other programs seems fine. Maybe it can explain why the results are a bit low?

               
              results.txt

              • Igor Pavlov

                Igor Pavlov - 2025-02-13

                Thanks for tests!

                System can switch working cores for single thread programs.
                There is switch -stm1 in 7-zip that can set affinity to first core. But I don't think that it will change results.

                About results. ARM64 code from MSVS compiler is slow for most HASH calculation workloads of 7-Zip. GCC (WSL) is 20-60% faster.
                Even X64 (MSVS) emulated code is faster than ARM64 (MSVS) for HASH code. Probably I'll look the code later to find the reason of bad performance of MSVS-ARM64.

                 

                Last edit: Igor Pavlov 2025-02-13

  • VictorVG

    VictorVG - 2024-11-30

    Small typo - all Windows installel's (SFX and MSI) included old redaction readme.txt for v24.09 :) - find insorg at forum.ru-board.com.

     

  • Melchior

    Melchior - 2024-11-30

    thx Igor Pavlov for a 7zip update! :D :D

     

  • Sam Tansy

    Sam Tansy - 2024-11-30

    Where are those compression levels/dictionaries defined, if I'd like to change them?

     
1 2 3 > >> (Page 1 of 3)

Log in to post a comment.