Re: estimating the value of a computer year

On Sun, 08 Jul 2007 21:27:48 -0500, quasi
<quasi@xxxxxxxx> wrote:

On Sun, 08 Jul 2007 08:10:50 -0700, Barry Schwarz
<schwarzb@xxxxxxxxx>
wrote:

On Sat, 07 Jul 2007 21:24:03 -0500, quasi
<quasi@xxxxxxxx> wrote:

Assume a program requires n sequential steps to
execute to completion.
In other words, assume it's not parallelizable. I
think that already
bars quantum computers, but if not, let's bar them
explicitly. Just
one computer -- a single sequential processor.
Let's assume a step
requires the absolute simplest non-null
instruction -- for example, a
test of a register or something like that.

Questions:

With today's technology (as of July 7, 2007),
using the fastest
possible hardware (cost is no object), what value
of n is feasible for
computation in 1 year?

Let's call the largest such value of n a "computer
year". Thus, a
computer year is the number of steps the world's
fastest computer can
do in 1 year. It's kind of like a light year
except that unlike a
light year, the value of a computer year has
increased over time,
since the advent of computers.

Based on currently accepted standard concepts of
elementary particles,
speed of light, etc, is there a theoretical upper
bound for all time
on the value of a computer year? If so, what would
be a reasonable
estimate for an all-time least upper bound?

quasi

Let's assume you mean a 365 day year. That is
3.1356E7 seconds.

Let's also assume that the fastest any "step" could
be performed is
the amount of time it takes for an electrical or
optical pulse (both
travel at c) to travel the diameter of a hydrogen
atom. The speed of
light is ~2.9979E10 cm/sec. The hydrogen atom is
~1E-8 cm. The time
involved is ~3.3356E-19 sec. At that rate, you
could perform
~9.454E25 steps in a year.

Thanks for the estimates.

So let me ask a followup question.

Suppose we have a "true" random number generator. Of
course, such a
device is a purely hypothetical construct, but let's
allow such an
assumption for now, so as to focus on an ideal case.
More realism can
be added later to deal with the fact that all
software-based RNGs are
inherently only pseudorandom.

The question concerns encryption safety.

Suppose I have a bit-string x of length n bits,
where n>=128.

Let y be a random bit-string, also of length n,
generated by the
random number generator.

Let z = x xor y.

Suppose y is a secret key, known only to 2 people, A
and B.

Assume the value of z is visible to a 3rd party C,
who wants to know
the value of x.

C has no information about the possible values of x,
and does not know
the value of y.

All C knows is that z = x xor y.

With the above assumptions, is z secure against all
brute force
attacks by any physically possible number of
computers, operating at
the theoretical limit of cpu speed, for all time?

Actually, after a moment's thought, I realize now
that the question I
asked above is not meaningful.

Clearly z is uniformly random so could reprsent any
message x.

There is no such thing as an attack on z, since there
would be no way
to recognize the true value of x from a false one.

I'll have to rethink the question.

quasi

what you are looking for is RSA encryption.

it is based on factoring.

you want to know what size of data is neccesary to have a secure RSA key for many many years.

the US army has codes that requires at least 3 generations of computing to be broken. (assuming no exponential speedup in computers in the near future )

RSA is safe if the size of data is big enough.

well at least for brute force attacks.

and most hackers.

but it is not sure if it is secure for a team of mathematicians.

for instance mathematicians having factoring tricks.

:-)

tommy1729
.

Relevant Pages

  • Why Easy To Use Software Is Putting You At Risk
    ... Anyone who has been working with computers for a long time will have noticed ... because DNS does not configure properly or security permissions are relaxed ... Is It Also Secure ... guarantee that no one really knows for sure, not even Microsoft developers. ...
    (Security-Basics)
  • Re: Hlp - DB security mystery
    ... One of the other computers has as well, ... I triple checked the mdw and the Users group has NO ... to secure, including creating a brand new mdw. ... The database in on a network share to which only my group and I have access. ...
    (microsoft.public.access.security)
  • Re: Final Year Project Brainstorming
    ... I think a project that contrasts the security of Ubuntu and Windows ... computers were very old so they were told they would have to purchase new ... money to get new computers at a lower cost, install ubuntu, and secure the ... Figure the cost of IT person for Vista vs ...
    (Ubuntu)
  • Re: Network Hacking
    ... respect the rights of others, or else you risk getting your rights taken ... You don't have to hack into live computers to learn how to secure ...
    (microsoft.public.win2000.security)
  • Re: IPSEC with non-domain Server
    ... Certificates are not the "most secure", rather, they are one of the 2 "more ... > authenticate computers and protect traffic integrity and confidentiality ... > Attacks on IPSec and Other Security Concerns ...
    (microsoft.public.security)