NODE ccb1d251Re: Trapdoors
jim@tadpole.com (Jim Thompson)Thu, 18 Feb 93 12:58:33 PST
> The specific instruction is the so-called "NSA instruction" which John Gilmore
> identified some time ago.
>
> Sun Microsystems was ordered by the NSA to redesign their chips to capture
> keys, which is why the SPARC processor was introduced. SPARC stands for
> "Sun Processor Allowing Remote Capture."
Interestingly enough, V9 SPARC has a population count (number of bits
set in word) instruction that was put in specificly at the request of
No Such Agency.
Jim
NODE 6b2ec4a4Trapdoors
Eric Hughes <hughes@soda.berkeley.edu>Sun, 21 Feb 93 13:58:03 PST
Jim writes:
>Interestingly enough, V9 SPARC has a population count (number of bits
>set in word) instruction that was put in specificly at the request of
>No Such Agency.
Does anybody have a good idea what applications this is useful for?
My first thought is that it's a very quick way to do linear error
detection codes, since this instruction directly computes the Hamming
weight of a code word.
I can also see it being useful to detect correlations between
sequences, such as a trial random stream and a known pseudorandom
number generator. One would XOR the streams together and then count
bits to calculate a correlation frequency.
Other ideas?
Eric
NODE 277d4923Re: Trapdoors
henry strickland <strick@osc.versant.com>Sun, 21 Feb 93 14:28:36 PST
# From cypherpunks-request@toad.com Sun Feb 21 14:14:43 1993
#
# Does anybody have a good idea what applications this is useful for?
The old CDC CYBER machines had population count in its instruction
set. Perhaps some scientific-type programmers would know what they
used it for. The CYBER did not have a lot of instructions -- they
were pretty practical about what they put in. i.e. != VAX
# My first thought is that it's a very quick way to do linear error
# detection codes, since this instruction directly computes the Hamming
# weight of a code word.
That was always my assumption. Anyway, it's not unprecedented.
strick
strick@osc.versant.com
NODE 813f935bRe: Trapdoors
cp@jido.b30.ingr.com (Craig Presson)Mon, 22 Feb 93 12:58:17 PST
In <9302212232.AA12362@versant.com>, henry strickland writes:
|> # From cypherpunks-request@toad.com Sun Feb 21 14:14:43 1993
|> #
|> # Does anybody have a good idea what applications this is useful for?
|>
|> The old CDC CYBER machines had population count in its instruction
|> set. Perhaps some scientific-type programmers would know what they
|> used it for. The CYBER did not have a lot of instructions -- they
|> were pretty practical about what they put in. i.e. != VAX
More precedent: DG Eclipses have a COB (COunt Bits) instruction.
Interestingly enough, there seems to be no such instruction in the
VAX (This from R'ingTFM. I have pitifully little VAX experience.)
^
/
------/---- cp@jido.b30.ingr.com (Craig Presson)
/
/
NODE 68dbd968Re: Trapdoors
peter honeyman <honey@citi.umich.edu>Mon, 22 Feb 93 19:38:12 PST
> Interestingly enough, there seems to be no such instruction in the
> VAX ...
but i believe the vax has ffs (find first set) instruction, which seems
to have similar uses (whatever they may be).
peter
NODE d2f69ba7Re: Trapdoors
Phiber Optik <phiber@eff.org>Mon, 22 Feb 93 21:00:43 PST
>
> > Interestingly enough, there seems to be no such instruction in the
> > VAX ...
>
> but i believe the vax has ffs (find first set) instruction, which seems
> to have similar uses (whatever they may be).
>
> peter
>
But that's not a population count (all bits set count) instruction.
In the Motorola 68020/30/40, there's BFFFO, for Bit Field Find First One,
if anyone was curious about 'find first set' instructions. Might as well
mention that the 386/486 have BSF and BSR, Bit Scan Forward and Reverse,
but now we're just getting into run-of-the-mill bit manipulation.