Could you spare a clue stick. What's an
out-of-process component and how it is
different from any other component? Is it
just that you need a wrapper for it?
There is this fascinating language called
unlambda. It's minimalistic, Turing complete
and compilable. If you are not convinced by
the above poster and want an example of why
Turing machines are hideous, try coding a
calculator in unlambda.
I dream of a day when we use lossless
compression exclusively because there is
no need to conserve space. No more MPEG
compression artifacts, no more decoding
delays and excessive CPU/graphics card
loads - every frame is stored ready to
load to screen, every song is stored with
maximum details in the most sound-preserving
format. This in turn would rapidly force
better monitors and speakers to go
mainstream.
When I was choosing a grad school I actually
spoke to Likharev about his advances. This was
three years ago. At that time, they were finalizing design of their D/A and projecting it to be at 900 GHz. The only integration problems
they had issues with was the fact that at that
speed, parts of your chip are in separate light
cones (you either know what I am talking about or
you don't). IIRC, they claimed that once
lithography reaches 0.1 um, making chips would be
as cheap as semiconductor chips.
I wasn't a cryo person at the time so I asked
how he envisioned commercial products with such
low temperatures. His vision was a closed-cycle
LHe fridge, kinda like cryotech on steroids.
I know my lab would consider buying a 10 K
RF-capable D/A. I think they have a market.
He is too expensive to just let it sit
around. They use closed-cycle refirgerators.
Lotsa compressor noise but little heat loss.
A properly engineered system would not run out
of LHe for years and years.
1. Why?
2. Reducible to typing one word on command line.
3. When timing is critical, people hand code in
assembly.
4. Your "I/O box" should be powerful enough to
rarely call main box. Latency?
5. Write your apps for "I/O box".
6. see 4.
Before transistor there were tubes. Some people
still swear by them. This transistor revolution
you speak of, is really all about scalability/power
consumption. There was nothing more revolutionary
to Bardeen's baby.
In our lab, we had a need for hard real time
digital experiment control. But as the author
of the paper correctly states, I/O system is
a source of much latency, so you have to use
a separate DSP to do control directly to avoid
bus timing issues. So why would anyone need a
non-embedded RTOS?
So you're saying it'd be easy to
code/. in LaTeX. Even image next
to other image is hard because it'll
depend on your margins. Forget image
within image, a newbie will have a hard
time having an image fit in a double-
column format without screwing up alignment.
Everyone, including writers of LaTeX
manuals, agrees that LaTeX is not a page
layout tool, it is best for cases where
your document fits a prebuilt template,
e.g. for scientific atricles.
Remember that HTML has been designed to render
well on any size screen. LaTeX has a page size
spec as part of your document. To make a TeX
renderer for arbitrary size display would be
hard.
Last but not least, LaTeX has no bindings for
an interactive scripting language, so Java and
ECMAscript would be tough.
Read the article. You'll need a little
"ring" to use it in a regular CD-Rom.
I wonder why you'd get yourself this "ring"
given that this would only allow you
access to junk mail.
That's not the only problem. LaTeX is not
convinient for precise layout. It was meant
for scientific articles where structure of the
document is laid out in advance. But try
getting two images to be rendered next to
each other with pixel precision and you run into
a wall as this is contrary to how LaTeX -> dvi -> ps
scheme works.
Russia's Putin made the offer, BTW.
The President of Zimbabwe (I think it's
Zimbabwe though I may be wrong) said that
if they pulled off an election like that,
UN would have sent troops to install an
arbitrarily preferred government.
Re:Damned if you do, Damned if you don't
on
Golden Rice
·
· Score: 1
Well, to do this right you need to understand
what your engineered product might do. As it
stands we do not know all the genes in all
living things, we have no model for our Earth
eco-system, we have don't even know how genes
interact with each other to steer organism
development. Do you know of a system that
looks at your genes and predicts your facial
features? Can you claim that all genes that cause
diseases are understood? Even just in humans?
How do you "engineer" without knowing safety
margin? As it stands, GE reminds me of old
definition of hacking: modifying source code
without understanding how it works as a whole
(e.g. Linux kernel development for most
developers). With that, you are bound to have
bugs. I just don't want the system to ever crash,
that's all.
I'd personally ban all commercial GE applications
for a few hundred years until we understand things
better.
I wonder if this is entirely true. I have no
knowledge of chip design and layout issues
but it seems to me that sheer interest in
advanced features wouldn't push Intel to
P4-like processors. It is twice the size of
P3, which means less chips per wafer and less
yeild.
I would guess that stuff like SIMD can be kept
proprietary, so they push on that at the
expense of generic x86.
It also seems as though Intel squeezes about a
factor of 5 increase between initial
introduction speed and how far the core is
eventually pushed. Is there a reason? And if so,
should we expect P4 core to be pushed to 5Ghz
before topping out.
They ought to either count them for Buchanan
or throw them out (the mangled votes that is).
You can't have people come to some office and
redefine their vote after election. A mistake
in marking the ballot can only invalidate it.
If they count those votes for Gore, expect
long court challenges, all the way to Supreme
Court.
I am more troubled by Jesse Jackson's allegation
that some people (esp. immigrants) were
pressured to vote for Bush. I don't think it'll
be proven, but if it is, could it invalidate
Bush's candidacy altogether?
Lastly, could anyone from Florida comment on the
two contradictions I have heard:
1. People who were still in line to vote at 7 pm
were allowed to vote after 7 pm. Some said it is
state law, others claim Jeb Bush ordered it.
Which is true?
2. How many votes were Buchananized? I've seen
figures ranging from 2,000 to 10,000.
I have said this before and I'll say it again:
all arms should be allowed including nukes.
The reason is that the population must have
more military power than the state so it could
overthrow government if the government
displeases it.
If you don't want to grant people the right to
have nukes then you must destroy or give away
all nukes that federal gov't has in its
possession.
I surely don't have the right to shoot at you,
but I have every right to plan your assasination
down to last detail. I cannot be guilty until
I commit a crime. Intellectual pursuit is not
a crime, regardless of its topic, purpose or
subject.
Power savings are dubious at this point.
To get reliable operation from single
electron devices you need ambient energy
(temperature) to be low enough to not
distort signals too much. So most likely
practical devices will need liquid helium
scale temparatures. Researchers in this area
routinely envision PCs with something like
cryotech stuff only much fancier.
Supercooling will consume a lot of power.
So there may be a net gain in power consumption,
but that is not obvious right now.
The story makes it sound like you heard
the speech. If so, could you clarify whether
Michael Tiemann was speaking for himself,
for Cygnus or for entire RedHat? What was
the context?
AFAIK, most of the drag is due to space near
Earth not being perfect vacuum. They call it
outer space, but our atmosphere still has
some presence. Small junk will see less drag
and will persist longer, although it also
depends on its shape and mass.
The dynamics is quite complicated, so if you are
far enough away, then you can run away from
Earth due to gravitational effects (IIRC that's
what's happening to Moon, sloowly).
LinLogFS claims to be a real LFS (if not yet
completed).
I think this entire discussion needs to start
by people reading:
http://www.penguin.cz/~mhi/fs/Filesystems-HOWTO/ Filesystems-HOWTO.html
Slashdot Mirror
Could you spare a clue stick. What's an
out-of-process component and how it is
different from any other component? Is it
just that you need a wrapper for it?
I got the impression Turing machine WAS their
innovation. The sad thing is, I am sure he would
have no trouble patenting it.
There is this fascinating language called
unlambda. It's minimalistic, Turing complete
and compilable. If you are not convinced by
the above poster and want an example of why
Turing machines are hideous, try coding a
calculator in unlambda.
I dream of a day when we use lossless
compression exclusively because there is
no need to conserve space. No more MPEG
compression artifacts, no more decoding
delays and excessive CPU/graphics card
loads - every frame is stored ready to
load to screen, every song is stored with
maximum details in the most sound-preserving
format. This in turn would rapidly force
better monitors and speakers to go
mainstream.
When I was choosing a grad school I actually
spoke to Likharev about his advances. This was
three years ago. At that time, they were finalizing design of their D/A and projecting it to be at 900 GHz. The only integration problems
they had issues with was the fact that at that
speed, parts of your chip are in separate light
cones (you either know what I am talking about or
you don't). IIRC, they claimed that once
lithography reaches 0.1 um, making chips would be
as cheap as semiconductor chips.
I wasn't a cryo person at the time so I asked
how he envisioned commercial products with such
low temperatures. His vision was a closed-cycle
LHe fridge, kinda like cryotech on steroids.
I know my lab would consider buying a 10 K
RF-capable D/A. I think they have a market.
He is too expensive to just let it sit
around. They use closed-cycle refirgerators.
Lotsa compressor noise but little heat loss.
A properly engineered system would not run out
of LHe for years and years.
1. Why?
2. Reducible to typing one word on command line.
3. When timing is critical, people hand code in
assembly.
4. Your "I/O box" should be powerful enough to
rarely call main box. Latency?
5. Write your apps for "I/O box".
6. see 4.
Before transistor there were tubes. Some people
still swear by them. This transistor revolution
you speak of, is really all about scalability/power
consumption. There was nothing more revolutionary
to Bardeen's baby.
In our lab, we had a need for hard real time
digital experiment control. But as the author
of the paper correctly states, I/O system is
a source of much latency, so you have to use
a separate DSP to do control directly to avoid
bus timing issues. So why would anyone need a
non-embedded RTOS?
So you're saying it'd be easy to /. in LaTeX. Even image next
code
to other image is hard because it'll
depend on your margins. Forget image
within image, a newbie will have a hard
time having an image fit in a double-
column format without screwing up alignment.
Everyone, including writers of LaTeX
manuals, agrees that LaTeX is not a page
layout tool, it is best for cases where
your document fits a prebuilt template,
e.g. for scientific atricles.
Remember that HTML has been designed to render
well on any size screen. LaTeX has a page size
spec as part of your document. To make a TeX
renderer for arbitrary size display would be
hard.
Last but not least, LaTeX has no bindings for
an interactive scripting language, so Java and
ECMAscript would be tough.
Read the article. You'll need a little
"ring" to use it in a regular CD-Rom.
I wonder why you'd get yourself this "ring"
given that this would only allow you
access to junk mail.
That's not the only problem. LaTeX is not
convinient for precise layout. It was meant
for scientific articles where structure of the
document is laid out in advance. But try
getting two images to be rendered next to
each other with pixel precision and you run into
a wall as this is contrary to how LaTeX -> dvi -> ps
scheme works.
Russia's Putin made the offer, BTW.
The President of Zimbabwe (I think it's
Zimbabwe though I may be wrong) said that
if they pulled off an election like that,
UN would have sent troops to install an
arbitrarily preferred government.
Well, to do this right you need to understand
what your engineered product might do. As it
stands we do not know all the genes in all
living things, we have no model for our Earth
eco-system, we have don't even know how genes
interact with each other to steer organism
development. Do you know of a system that
looks at your genes and predicts your facial
features? Can you claim that all genes that cause
diseases are understood? Even just in humans?
How do you "engineer" without knowing safety
margin? As it stands, GE reminds me of old
definition of hacking: modifying source code
without understanding how it works as a whole
(e.g. Linux kernel development for most
developers). With that, you are bound to have
bugs. I just don't want the system to ever crash,
that's all.
I'd personally ban all commercial GE applications
for a few hundred years until we understand things
better.
I wonder if this is entirely true. I have no
knowledge of chip design and layout issues
but it seems to me that sheer interest in
advanced features wouldn't push Intel to
P4-like processors. It is twice the size of
P3, which means less chips per wafer and less
yeild.
I would guess that stuff like SIMD can be kept
proprietary, so they push on that at the
expense of generic x86.
It also seems as though Intel squeezes about a
factor of 5 increase between initial
introduction speed and how far the core is
eventually pushed. Is there a reason? And if so,
should we expect P4 core to be pushed to 5Ghz
before topping out.
Einstein patented a fairly obvious
fridge design. Don't know if he was
such a good examiner but he abused
the system like evryone else.
Me and my friends long advocated a cocaine
snort-off.
They ought to either count them for Buchanan
or throw them out (the mangled votes that is).
You can't have people come to some office and
redefine their vote after election. A mistake
in marking the ballot can only invalidate it.
If they count those votes for Gore, expect
long court challenges, all the way to Supreme
Court.
I am more troubled by Jesse Jackson's allegation
that some people (esp. immigrants) were
pressured to vote for Bush. I don't think it'll
be proven, but if it is, could it invalidate
Bush's candidacy altogether?
Lastly, could anyone from Florida comment on the
two contradictions I have heard:
1. People who were still in line to vote at 7 pm
were allowed to vote after 7 pm. Some said it is
state law, others claim Jeb Bush ordered it.
Which is true?
2. How many votes were Buchananized? I've seen
figures ranging from 2,000 to 10,000.
I have said this before and I'll say it again:
all arms should be allowed including nukes.
The reason is that the population must have
more military power than the state so it could
overthrow government if the government
displeases it.
If you don't want to grant people the right to
have nukes then you must destroy or give away
all nukes that federal gov't has in its
possession.
I surely don't have the right to shoot at you,
but I have every right to plan your assasination
down to last detail. I cannot be guilty until
I commit a crime. Intellectual pursuit is not
a crime, regardless of its topic, purpose or
subject.
If you generate information and have no
storage for it then it is not information.
In other words, you can't have more info than
you can store.
Power savings are dubious at this point.
To get reliable operation from single
electron devices you need ambient energy
(temperature) to be low enough to not
distort signals too much. So most likely
practical devices will need liquid helium
scale temparatures. Researchers in this area
routinely envision PCs with something like
cryotech stuff only much fancier.
Supercooling will consume a lot of power.
So there may be a net gain in power consumption,
but that is not obvious right now.
The story makes it sound like you heard
the speech. If so, could you clarify whether
Michael Tiemann was speaking for himself,
for Cygnus or for entire RedHat? What was
the context?
Yesterday's comments suggest this bug is
fixed. If it still doesn't work for you,
I'd consider voting for it.
AFAIK, most of the drag is due to space near
Earth not being perfect vacuum. They call it
outer space, but our atmosphere still has
some presence. Small junk will see less drag
and will persist longer, although it also
depends on its shape and mass.
The dynamics is quite complicated, so if you are
far enough away, then you can run away from
Earth due to gravitational effects (IIRC that's
what's happening to Moon, sloowly).
LinLogFS claims to be a real LFS (if not yet/ Filesystems-HOWTO.html
completed).
I think this entire discussion needs to start
by people reading:
http://www.penguin.cz/~mhi/fs/Filesystems-HOWTO