perfect rainbow table

christiaan_c · **Posted:** 07 Mar 2008, 20:30

Yes, all the perfect tables have the same index number, the orginal tables didn't have same index numbers.
Also another question, is it possible too set the size that the perfect tables needs to be? (if you don't want tables that are 1 Gig)

Thnx,

Christiaan

teamakg · **Joined:** 18 Oct 2007, 20:19 **Posts:** 181

christiaan_c wrote:

Yes, all the perfect tables have the same index number, the orginal tables didn't have same index numbers.
Also another question, is it possible too set the size that the perfect tables needs to be? (if you don't want tables that are 1 Gig)

Thnx,

Christiaan

The original tables all needed the same index number before you perfected them. rtperfect is hard coded for 1GB tables.

PowerBlade · **Posted:** 07 Mar 2008, 21:36

teamakg wrote:

christiaan_c wrote:

Yes, all the perfect tables have the same index number, the orginal tables didn't have same index numbers.
Also another question, is it possible too set the size that the perfect tables needs to be? (if you don't want tables that are 1 Gig)

Thnx,

Christiaan

The original tables all needed the same index number before you perfected them. rtperfect is hard coded for 1GB tables.

I just wanted to confirm what teamakg wrote is true

Sc00bz · **Joined:** 03 Dec 2007, 11:37 **Posts:** 725

Yea it's finally here. This really isn't a patch it's more like a rewrite.

This was tested on my computer with two 2.8 GHz Xeons with HT, 2 GiB PC266 RAM, and one 10k RPM hard drive (2 MiB cache I think) running Ubuntu 7.10. The test files were create from 70,000,000 random chains with a key space of 2^48. They were evenly split into the files and then sorted. The times are how long it took to merge the files together while removing duplicates. This is version 0.0 there might be a 1.0. Version 1.0 will do the sorting and merging and I might add multi threading so one thread will be reading while another thread will be merging.

*** Note these stats don't include IO time. ***
Yours with 220 files took 693 sec
Mine with 220 files took 124 sec
--------------------------------
Mine is 5.59 times faster

Yours with 67 files took 202 sec
Mine with 67 files took 116 sec
--------------------------------
Mine is 1.74 times faster

*** There is an bug that causes it to run slow. This is now fixed, read below ***
The new version contains the source code and two executables one for Windows and one for Linux. I'm leaving this up in case someone wants to look at the original code, but other than that you should use the new version.

pointp · **Joined:** 18 Oct 2007, 19:30 **Posts:** 323

i'll have a look with my QuadCore

Thx you guys (PB & sc00bz)

Edit : i have 2 erros during compilation (VS 2K5 & dirent.h file)
so if someone can/want to have a try

PowerBlade · **Posted:** 11 Mar 2008, 19:38

He made a comment in the code about where you can download dirent.h.
It works fine for me after downloading it and put it in the dir

alone · **Joined:** 22 Oct 2007, 19:07 **Posts:** 66

thx i willl try under win xp!!

teamakg · **Joined:** 18 Oct 2007, 20:19 **Posts:** 181

Would somebody be kind enough to compile this and post an .exe for us non-programming Windows users.

PowerBlade · **Posted:** 11 Mar 2008, 20:19

teamakg wrote:

Would somebody be kind enough to compile this and post an .exe for us non-programming Windows users.

http://www.freerainbowtables.com/download/rtperfect.exe
You need visual studio 2008 redist to run it ( http://www.microsoft.com/downloads/deta ... 4B9F2BC1BF )

pointp · **Joined:** 18 Oct 2007, 19:30 **Posts:** 323

PowerBlade wrote:

He made a comment in the code about where you can download dirent.h.
It works fine for me after downloading it and put it in the dir

i put it in the right dir that's my problem

alone · **Joined:** 22 Oct 2007, 19:07 **Posts:** 66

hi,
just try this last rtperfect.exe from PB&Sc00bz, but time is very very long...??? much more that the old one!
"
Starting with 4 files.
Finished with file 'Nouveau dossier\ntlm_loweralpha-numeric#1-9_0_50000x67108864_distrrtgen_1.rt'
Finished with file 'Nouveau dossier\ntlm_loweralpha-numeric#1-9_0_50000x67108864_distrrtgen_2.rt'
Finished with file 'Nouveau dossier\ntlm_loweralpha-numeric#1-9_0_50000x67108864_distrrtgen_0.rt'
Finished with file 'Nouveau dossier\ntlm_loweralpha-numeric#1-9_0_50000x67108864_distrrtgen_3.rt'
Unique Chains: 249802327
Total time: 4654.45
"
new version (from rtperfecter)>1h15min (4600sec)
old version < 3 min

Do you have any idea?

Sc00bz · **Joined:** 03 Dec 2007, 11:37 **Posts:** 725

This maybe a buffering problem what compiler and OS did you use?

Sc00bz · **Joined:** 03 Dec 2007, 11:37 **Posts:** 725

I found the problem apparently kbhit() is the slowest function ever. This also fixes the fact that in Windows if you press any key it will stop until you hit enter.

Replace:

Code: Select all

   for (int ret = inRt.readChain(&(chain.c)); ret == 0; ret = inRt.readChain(&(chain.c)))
   {
      // Show status
#ifdef _WIN32
      if (kbhit())
      {
         int ch = getchar();
         if (ch == (int)'\n')
         {
            inRt.printStatus();
         }
      }
#else
      int ch = getchar();
      if (ch == (int)'\n')
      {
         inRt.printStatus();
      }
#endif
...

With

Code: Select all

   int loops = 0;
   for (int ret = inRt.readChain(&(chain.c)); ret == 0; ret = inRt.readChain(&(chain.c)))
   {
      // Show status
      loops++;
      if (loops >= 2000000)
      {
#ifdef _WIN32
         // Pressing any key will show the status
         if (kbhit())
         {
            inRt.printStatus();
         }
#else
         // Pressing enter will show the status
         int ch;
         do
         {
            ch = getchar();
         } while (ch != (int)'\n' && ch != EOF);
         if (ch == (int)'\n')
         {
            inRt.printStatus();
         }
#endif
         loops = 0;
      }
...

Edit gahh I should really test code before I post it. Fixed a typo it's good now.

Sc00bz · **Joined:** 03 Dec 2007, 11:37 **Posts:** 725

New version rtperfecter v0.1 now 4 times faster in Linux and like 200 times faster in Windows.

This was tested on my computer with two 2.8 GHz Xeons with HT, 2 GiB PC266 RAM, and one 10k RPM hard drive (2 MiB cache I think) running Ubuntu 7.10. The test files were create from 70,000,000 random chains with a key space of 2^48. They were evenly split into the files and then sorted. The times are how long it took to merge the files together while removing duplicates. This is version 0.0 there might be a 1.0. Version 1.0 will do the sorting and merging and I might add multi threading so one thread will be reading while another thread will be merging.

*** Note these stats don't include IO time. ***
BP's with 220 files took 693 sec
Mine v0.0 with 220 files took 124 sec
Mine v0.1 with 220 files took 33 sec
--------------------------------
Mine v0.0 is 5.59 times faster
Mine v0.1 is 21.00 times faster

BP's with 67 files took 202 sec
Mine v0.0 with 67 files took 116 sec
Mine v0.1 with 67 files took 27 sec
--------------------------------
Mine v0.0 is 1.74 times faster
Mine v0.1 is 7.48 times faster

alone · **Joined:** 22 Oct 2007, 19:07 **Posts:** 66

thx Sc00bz !

You are very reactive!

but as i said i'm not at all a coder. I just use the exe file PB made for us. so i don't kown how he compile your 0.0 version.

I hope PB can compile for us also the newest 0.1 !!!
As soon i can get the exe file for this 0.1 version i try and tell you.

ty

++
Alone

perfect rainbow table

Who is online