Test results ultimate compression
Tim Bernard
in about 18 seconds or so ?
Currently my system is a 1200 Duron CPU with 500 Megs of ram running xp so
if i had a faster cpu I could easily slash the time down to less then 9
seconsd
if the code were to be written to be supported for parallel processing then
with 32 processors one could compress 3200 bytes or 3.2 k in 9 seconds down
to 224 byte chunks
or with 64 processors one could compress 6.4k in 9 seconds down to 448 byte
chunks
--
"What worries me is not the violence of the few, but the indifference of
the many"
Kevin Easton
> Is it considered successful to be able compress any 100 byes down to 7 bytes
> in about 18 seconds or so ?
It is known to be impossible to have a reversible algorithm that maps from
the set of all 100 byte sequences to the set of all 7 byte sequences.
A moment's reflection on the comparison between the number of states that
can be represented by 100 bytes and by 7 bytes, and the reason why will become
clear.
- Kevin.
Tim Bernard
the approach and view it differently to make it happen.
It does work.
"Kevin Easton" <kevin@-nospam-pcug.org.au> wrote in message
news:newscache$4ueodh$m7p$1...@tomato.pcug.org.au...
Kevin Easton
[top-posting fixed]
> news:newscache$4ueodh$m7p$1...@tomato.pcug.org.au...
>> Tim Bernard <notmy...@server.com> wrote:
>> > Is it considered successful to be able compress any 100 byes down to 7
> bytes
>> > in about 18 seconds or so ?
>>
>> It is known to be impossible to have a reversible algorithm that maps from
>> the set of all 100 byte sequences to the set of all 7 byte sequences.
>> A moment's reflection on the comparison between the number of states that
>> can be represented by 100 bytes and by 7 bytes, and the reason why will
> become
>> clear.
>>
> One must understand the limitations of math and its strengths and change
> the approach and view it differently to make it happen.
> It does work.
Clearly a moment's reflection was too much to ask. Oh well, your
deliberate misspelling of "algorithm" shows you as the troll you are.
Goodbye.
- Kevin.
Matt Mahoney
> Is it considered successful to be able compress any 100 byes down to 7 bytes
> in about 18 seconds or so ?
It depends on what you are compressing. gzip will compress the 100,000 byte
file aaa.txt (consisting of 'a' repeated 100,000 times) to 135 bytes in 0.1
seconds on my 750 MHz PC.
-- Matt Mahoney
Dale King
news:CyRoa.119474$jVh....@news01.bloor.is.net.cable.rogers.com...
> One must understand the limitations of math and its strengths and change
> the approach and view it differently to make it happen.
We understand math and its limitations. But it has no limitations in this
regard. It is an absolute fact that one cannot compress all sequences of
size n-bits to something smaller than n bits.
> It does work.
No, it doesn't.
> "Kevin Easton" <kevin@-nospam-pcug.org.au> wrote in message
> news:newscache$4ueodh$m7p$1...@tomato.pcug.org.au...
> > Tim Bernard <notmy...@server.com> wrote:
> > > Is it considered successful to be able compress any 100 byes down to 7
> bytes
> > > in about 18 seconds or so ?
No, it must also decompress to the original. It is trivial to write a lossy
compressor that can compress all 100 byte sequences to 7 bytes. As Kevin
said it is impossible to be able to decompress them all to the original.
> > It is known to be impossible to have a reversible algorithm that maps
from
> > the set of all 100 byte sequences to the set of all 7 byte sequences.
> > A moment's reflection on the comparison between the number of states
that
> > can be represented by 100 bytes and by 7 bytes, and the reason why will
> become
> > clear.
--
Dale King
Petrut
> Is it considered successful to be able compress any 100 bytes down to 7
> bytes in about 18 seconds or so ?
the time is not important...
did you try to compress the uncompressible file of Mark Nelson , the one
million random digits ?
we are all waiting for your answer here, on this newsgroup, in the next 48
hours
bye
petrut
Tim Bernard
40 characters a second.
I dont have the time to do that.
"Petrut" <pet...@wanadoo.fr> wrote in message
news:b81fm8$8fb$1...@news-reader12.wanadoo.fr...
Kevin Easton
> compressing a million random digits ? The fastest my engine works so far is
> 40 characters a second.
> I dont have the time to do that.
You don't have 7 hours?
- Kevin.
Glen Smith
Which is the time it would take.
Send me the link to the file and ill do it.
"Kevin Easton" <kevin@-nospam-pcug.org.au> wrote in message
news:newscache$8g0rdh$jm5$1...@tomato.pcug.org.au...
Glen Smith
down to 7 bytes in 1 second
Can be any 8 bit characters found in ASCII
"Matt Mahoney" <matma...@yahoo.com> wrote in message
news:3EA40E03...@yahoo.com...
Raymond Wan
On Tue, 22 Apr 2003, Glen Smith wrote:
> I can compress any 100 Bytes down to 7 bytes in 30 seconds or any 40 bytes
> down to 7 bytes in 1 second
> Can be any 8 bit characters found in ASCII
Hi Glen, Tim, or whoever,
It's obvious that many in this newsgroup doubt your claims and, I
too, have serious doubts. However, what I don't understand is why you and
those before you even bother posting on comp.compression . I mean, it's
not like you need a blessing from the readers of the newsgroup in order to
perform your experiments. Why don't you just go ahead and do whatever
demonstration you want, without support from readers of this newsgroup, so
that you don't suck everyone here into a long circular argument (i.e.,
"You can't do it"; "Yes I can"; "No you can't", etc.)?
And quite honestly -- it doesn't matter to me even if you *could*
compress 100 bytes down to 7. The fact that you're posting with multiple
names already brings your credibility into question. If I were to support
you financially, who do I make the cheque out to?? :-)
Ray
Petrut
let me tell you something :
this file has exactly 406 kilobytes (around 400.000 characters), I can send
it to you by email
if you want to understand why 406 kilobytes instead of 1.000.000 : the
digits were generated and the result was compressed with zip or bzip2 (I
don't remember exactly) , and nobody (until today) got success in the
compression of this file !
this is a well known file, you should have heard about it...
you say that you can compress 40 characters a second, so to compress 400.000
characters you need 10.000 seconds, that is to say less than 3 hours.
do you think your computer can't run for three hours this night ? I think it
can...
so do you want to compress this file, to uncompress the result, and to
compare the result with the original file, please ?
and tell us the result .
and remember one thing : it's not a shame to say here : guys, I have a bug
in my program !
but we are waiting for some good news, because nobody can prove that a file
is uncompressible, so you have a chance to compress it !
if you have success with the compression of this file, you will try to
compress several files generated by the web site "random.org".
and if you have success again, we will see what you will have to do with
your software
as you can see, I don't laugh, and I don't try to tell you that it's
impossible.
good luck and tell us the results
petrut
Glen Smith
"Petrut" <pet...@wanadoo.fr> wrote in message
news:b83u05$6ms$1...@news-reader12.wanadoo.fr...
Glen Smith
over time.
"Kevin Easton" <kevin@-nospam-pcug.org.au> wrote in message
news:newscache$4ueodh$m7p$1...@tomato.pcug.org.au...
s...@nospam.unt.edu
> It was known before as well that man cannot ever fly. That of course changed
> over time.
No, it was suspected that man could not fly. It was not known.
Here's a quote for you to ponder:
One reason why mathematics enjoys special esteem, above all other
sciences, is that its laws are absolutely certain and
indisputable, while those of all other sciences are to some extent
debatable and in constant danger of being overthrown by newly
discovered facts.
-- Albert Einstein, 1921
--
Steve Tate - srt[At]cs.unt.edu | "A computer lets you make more mistakes faster
Dept. of Computer Sciences | than any invention in human history with the
University of North Texas | possible exceptions of handguns and tequila."
Denton, TX 76201 | -- Mitch Ratliffe, April 1992
Dale King
news:a_gpa.100872$BQi....@news04.bloor.is.net.cable.rogers.com...
> It was known before as well that man cannot ever fly. That of course
changed
> over time.
But there is no comparison here. While it was thought that man would not be
able to fly, it wasn't a logical fallacy. We could at least imagine that it
could be possible for a man to fly. We at least knew that it was possible to
fly, we had seen birds and insects do it.
Or for the other cliché of it being said that we could not journey to the
moon. While this is a difficult feat technically, it is at least logical to
imagine that someone could actually be on the moon.
Your claim would be more likened to flying to the moon while simultaneously
staying on the earth. It is not logical for you to be in two places
separated by millions of miles at once. Nothing will make this possible
unless perhaps you were tall enough to reach the moon while standing on
earth.
Similarly it is not logical to claim that you can create a one-to-one
mapping between n uncompressed files and m compressed files when n > m.
--
Dale King
Kelsey Bjarnason
> It was known before as well that man cannot ever fly. That of course changed
> over time.
No such thing was ever known, and a simple glance at a bird is sufficient
to demonstrate that heavier-than-air items can, in fact, fly quite
handily.
I suspect you're confusing Lord Kelvin's assertions about heavier-than-air
craft with knowledge about flight in general.
Be that as it may, it doesn't help your case one iota; either you can do
what you say you can, or you can't. The entire body of all mathematical
knowledge related to the subject says you can't, and produces very simple,
very clear and very compelling reasons why it should be accepted; on the
other side we have you and countless before you, all asserting that the
math is all wrong, that you can do magic, yet offering _nothing_ to
support your position. This is not a terribly compelling argument.
Send me the decompressor; I'll send you a set of 100 byte files to
compress. If, on decompression, they all match the originals, you'll have
"won". You won't win, though. Need me to sign an NDA? No problem.
Ray
Please, I need that file also, I am looking for a small file for that
characteristic, may u please send it to a__l...@hotmail.com
the __ is two underscores.
Thanks
Regards,
Ray
Bryan W. Taylor
> Is it considered successful to be able compress any 100 bytes down to 7 bytes
> in about 18 seconds or so ?
No it's pretty lousy speed wise. I mean that's **really** slow for
practical uses. You should spend some time optimizing your code.
I've got a program than is fast enough to compress AND decompress 100
bytes down to 7 bytes fast enough to build the complete lookup table.
This took around 36 hours, but now I store it in a MySQL database in a
table called COMPTAB with two columns: input is a char(7) and OUTPUT
is a char(100). The 100 bytes column is indexed, so I just look up the
7 byte compressed value and return it. You should consider using this
method, it really speeds things up. By the way, don't index the table
until AFTER you've populated it, because maintaining the index will
slow it down during the calculations.
Here's the funny thing: guess how many rows the table has?
SELECT count(*) FROM COMPTAB;
Tom St Denis
> It was known before as well that man cannot ever fly. That of course changed
> over time.
Um last time I checked, man can still not fly. Planes can fly, people
can pilot planes, people cannot fly.
Also as others pointed out this is just wrong. People think there is
a god, doesn't make that so.
Tom
Dale King
news:11d78c87.03042...@posting.google.com...
Or another way to think about it is that we could instead of trying to do
this in software, which is too slow, just implement it in hardware with
using ROM. Considering the alternate version that Tim posted of compressing
64 bits to 35 bits to make it more manageable. The compressor device would
have 64 address lines for looking up into the ROM with 35 data bits output.
The decompressor would have 35 address lines and 64 data bits output.
It would of course require an extreme amount of memory, but such a device
could theoretically be constructed (it is up to Tim to provide the data to
fill the ROM). The problem is that with only 35 address lines the
decompressor device cannot produce all 2^64 possible input patterns.