Announcement

Collapse
No announcement yet.

The UVD/UVD2 thread.

Collapse
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by Qaridarium View Post
    if you are happy you can make a p2p Cluster of linux users and bruteforce the 256bit aes key...
    "The resources required for a brute force attack scale exponentially with increasing key size, not linearly. As a result, doubling the key size for an algorithm does not simply double the required number of operations, but rather squares them." http://en.wikipedia.org/wiki/Brute_force_attack
    (Someone check if I make a mistake in the following)
    Let's say the amount of transistors in year 0 is k. The amount in year 2 by Moore's law would be 2^1 * k. Original effort required to break the encryption is p. Effort after increasing p by one is p^2. If we assumed the password was crackable with transistors k, then p = k. Let's say i is the amount of two year cycles it takes to catch up after a single increment in password size. 2^i * k = k^2 (divided by k) => 2^i = k <=> i = log2(k). You quickly notice from that that you can't indefinitely keep up with the password length increases. Like if it took a two-billion-transistor CPU to break a password with length n, it would take 61.794705708 years by my calculations before you could create a computer that could break a password with length n+1.
    Last edited by nanonyme; 04-22-2009, 10:35 AM.

    Comment


    • Originally posted by nanonyme View Post
      The extensions I know of that are related to this are generic ones that you get with fglrx and will get later (as in, soon enough) get with open drivers and have nothing at all to do with nVidia except that nVidia has implemented them in a quite stable fashion.
      its a hartware problem! up to R580 there is no way!

      first R600 can handle this!

      and yes Future Catalyst versions and wine version will bring support for this shit.

      Comment


      • Originally posted by nanonyme View Post
        "The resources required for a brute force attack scale exponentially with increasing key size, not linearly. As a result, doubling the key size for an algorithm does not simply double the required number of operations, but rather squares them." http://en.wikipedia.org/wiki/Brute_force_attack
        (Someone check if I make a mistake in the following)
        Let's say the amount of transistors in year 0 is k. The amount in year 2 by Moore's law would be 2^1 * k. Original effort required to break the encryption is p. Effort after increasing p by one is p^2. If we assumed the password was crackable with transistors k, then p = k. Let's say i is the amount of two year cycles it takes to catch up after a single increment in password size. 2^i * k = k^2 (divided by k) => 2^i = k <=> i = log2(k). You quickly notice from that that you can't indefinitely keep up with the password length increases. Like if it took a two-billion-transistor CPU to break a password with length n, it would take 61.794705708 years by my calculations before you could create a computer that could break a password with length n+1.
        the Point is the error of this copy protection is not the AES key ,,,

        and yes 256bit AES key is not to much to brude force them..

        not for 1 pc or 1 server but in a P2P network all over the world Millon of Servers/PCs can Crack this shit..

        AES is not an opensource s Krypto algiritmus someone can find a bug so its posible to fragment a big KEy into a little one.

        in german wikipedia:
        "Kurz vor der Bekanntgabe des AES-Wettbewerbs stellten verschiedene Autoren eine einfache algebraische Darstellung von AES als Kettenbruch vor. Dies könnte für erfolgreiche Angriffe genutzt werden. Hierzu gibt es einen Videovortrag von Niels Ferguson auf der HAL 2001 [3].
        2002 wurde von Courtois and Pieprzyk ein theoretischer Angriff namens XSL gegen Serpent und Rijndael vorgestellt (siehe Serpent).
        Im Mai 2005 veröffentlichte Daniel Bernstein einen Artikel (PDF-Version) über eine unerwartet einfache Timing-Attacke (eine Art der Seitenkanalattacke) auf den Advanced Encryption Standard."

        http://de.wikipedia.org/wiki/Advance...n_und_Angriffe

        in simple worts... big key Goes to a Little and brudeforce will be easy!

        Comment


        • NIST sums it up nicely, even at 128 bit

          People 7.00E+09
          Computers per person 10.00
          Computers 1.00E+09
          Combos per second per computer 7.00E+19
          Total combos per second 7.00E+19
          Seconds per year 3.15E+07
          Total combos per year 2.22E+12
          128-bit key combos (*50%) 1.70E+38
          Years to crack 7.66E+25

          Comment


          • Originally posted by Qaridarium View Post
            PCGH in germany shows that ATI render shadows korrekt @ crysis and nvidia save shader power and do not render all shadows!

            And so? I just provided you with proof that that type of thing is not limited to nvidia but is rampant among all cards. Hell don't even mix intel and s3 renders into it, they are even worse.

            Comment


            • Originally posted by Kano View Post
              Did you see any UVD2 (libamdxvba1) programming examples? I never saw those, but I use VDPAU since serveral month. Currently my CPUs are so fast that I don't need VDPAU anymore on the systems which would support it (Intel Q9300@3GHz or E8400), but before 1080p was really much better with it. AMD waits so long that they could sell cpus instead of gpus as well. After 1-2 years nobody will need h264 accelleration when the minium speed is about 3 ghz for a 50$ cpu.
              That is what I have been saying. They are taking so long to release this that by the time they do it will be a moot point.

              Comment


              • Not exactly moot, the poor devs who can't afford / make up a reason to buy a new 50$ cpu (which most likely would also include a completely new set, expensive ddr3 ram and all) would then get the chance for great playback on their -current- computers.

                Which is a nice goal in itself.

                Comment


                • Originally posted by deanjo View Post
                  NIST sums it up nicely, even at 128 bit

                  People 7.00E+09
                  Computers per person 10.00
                  Computers 1.00E+09
                  Combos per second per computer 7.00E+19
                  Total combos per second 7.00E+19
                  Seconds per year 3.15E+07
                  Total combos per year 2.22E+12
                  128-bit key combos (*50%) 1.70E+38
                  Years to crack 7.66E+25
                  the IX the pro part of the (C't) has a artikle of password cracking WPA2! with the VGA's!
                  100 000 Valid checks per second! per (PC+vga's) unit!

                  (Combos per second per computer) = 100 000 !

                  with only a quatcore in 2 monds amd bring the 6 core! and this carts only 55nm! in 40nm VGAs you will have 150 000 Combos per second per computer.

                  Comment


                  • Originally posted by Qaridarium View Post
                    (Combos per second per computer) = 100 000 !.
                    For your information 100 000 combos per second is insignificant. We're talking about easily over 91292051633079798989750131910067116342455228306074 83146366674788070551428931526296681935903540008509 26342401 total amount.

                    Comment


                    • Also another note: the thing about WPA2 is that passwords are weak. If the passwords weren't that weak, WPA2 wouldn't be breakable in a timely manner. WPA2 in itself is not weak, humans are. You're not breaking the AES key there, you are just breaking a password set by a silly human. Key-based authentication would be closer to unbreakable.
                      Last edited by nanonyme; 04-22-2009, 06:54 PM.

                      Comment


                      • Originally posted by nanonyme View Post
                        For your information 100 000 combos per second is insignificant. We're talking about easily over 91292051633079798989750131910067116342455228306074 83146366674788070551428931526296681935903540008509 26342401 total amount.

                        That's right, and to top it off that Cuda cracker is only effective against extremely weak passwords.

                        Another great breakdown of real numbers.

                        Brute forcing a 256 bits cryptographic code without any known flaws (like WPA2 AES) means that they should test at worst 2^256 keys and average 2^255 (~ 10^76) keys.
                        Let’s say that they are using a new nVidia GPU with 1000 stream processors running at 10 GHz
                        That is 10^13 instructions per second
                        Now with their magical software they can check 1 key per instruction.
                        So, they can check 10^13 keys per second
                        Even with 1 billion of those GPU (10^9) they can "only" calculate 10^22 key per second (.
                        At that rate, they need 10^51 seconds which is 10^43 years.
                        As a reference, the estimate age of the universe is 10^14 years.

                        Comment


                        • Originally posted by deanjo View Post
                          That's right, and to top it off that Cuda cracker is only effective against extremely weak passwords.

                          Another great breakdown of real numbers.
                          my point is diverend...

                          NOW we have 100 000 valid tests per second..

                          in 2 monds we have 150 000 valid tests per second becourse of 6 core CPU's

                          and 40nm GPUs...

                          2010 we have 32nm 12 core CPUs and 32 nm GPUs

                          then we have 300 000 valid tests per second per PC unit....

                          so if we make a P2P cluster and 1000 Linux users help with 2 PC units...

                          we have 1000*2*300000= 600000000 valid tests per second:..

                          so we now do this 1 year:

                          600000000*365*24*60*60=1,89216^16 valid tests!

                          "Brute forcing a 256 bits cryptographic code without any known flaws "

                          but in AES there are known flaws!
                          Last edited by Qaridarium; 04-23-2009, 09:31 AM.

                          Comment


                          • Originally posted by Qaridarium View Post
                            600000000*365*24*60*60=1,89216^16 valid tests!
                            Oh, cool. So it would take only 48247532784267608970568097787748983353656788171230 145158795634555590179630324741547659478604 years to crack an alphanumeric password with spaces with length 60 by bruteforce methods by your calculations with the grid you visioned. You, sir, are a genious.

                            Comment


                            • It's a good thing most companies require strong passwords these days, so users write their passwords down on yellow sticky notes or leave them in clear-text files on the PC.
                              Last edited by bridgman; 04-23-2009, 10:34 AM.

                              Comment


                              • Originally posted by Qaridarium View Post
                                but in AES there are known flaws!
                                I read the section about cracking AES you read. The first one if I understood just related to factorization which is hard unless someone starts building powerful quantum computers. While my German lacks, I'd say for relative certainty that the German Wiki page says that the second approach is purely theoretical.
                                I might quote this thing from English Wikipedia page on timing attack which the Wiki page you talked of mentioned as the latest approach on the problem http://en.wikipedia.org/wiki/Timing_attack
                                "Timing attacks are easier to mount if the adversary knows the internals of the hardware implementation, and even more so, the crypto system in use. Since cryptographic security should never depend on the obscurity of either (see security through obscurity, specifically both Shannon's Maxim and Kerchoff's Law), resistance to timing attacks should not either. If nothing else, an exemplar can be purchased and reverse engineered. Timing attacks and other side-channel attacks may also be useful in identifying, or possibly reverse-engineering, a cryptographic algorithm used by some device."
                                Meaning someone would need to purposefully reverse-engineer the AMD security implementation to get absolute certainty that the cracking would actually work indefinitely.
                                This seems to be one of the reasons for why AMD is so cautious about giving out full hardware specifications of their cards.

                                Comment

                                Working...
                                X