I guess what you see is a bad thing, is actually a good thing in my book. Do you want your leader's
vote to be for sale to the most powerful lobby, or would you rather it be for sale to the public
opinion of the majority?
But this is a false dichotomy! I want a person that honestly believes in what I agree is right, and I want the democratic majority to agree enough to elect that person. Naive? perhaps....
The Real Beauty of Irix is in its capbilities on the big multiprocessor Onyx systems.
So is Linux: this is the tail of 'cat/proc/cpuinfo' on our Altix:
processor : 31 vendor : GenuineIntel arch : IA-64
32 processors, 32 GB of memory. It is nice to be able to allocate an 8GB array.
SGI certainly has the street cred for well-rounded systems. Fast memory, fast disk IO; check this output from 'hdparm -tT/dev/sda':
Timing buffer-cache reads: 9900 MB in 2.04 seconds = 4850.53 MB/sec
Timing buffered disk reads: 268 MB in 3.01 seconds = 89.10 MB/sec
Another neat feature: the 'BIOS' console is available on the ethernet. The console is actually a small embedded ARM computer, running Linux, of course, which has access to the hardware of the big computer: it can power up/down the nodes remotely, read temperature/voltage data, run hardware diagnostics etc.
shows that for last 450,000 years Earth was mostly in the Ice Age, interrupted by 10,000 year long warm periods spaced 100,000 years apart. We are about 15,000 years into the last warm period on record.
Because of strong periodicity, the current best explanation of this cycle is by astronomical phenomena (Earth orbit/axis wobble).
This does not contradict global warming---it just shows that the climate is a very delicate balance between strong opposing phenomena; the point being that we should be real careful how we influence it.
We may be on a cusp here, in a 'golden age' of WiFi. Up to now, as more and more businesses and individuals deployed Access Points, we were seeing increasing coverage, but as the article says, at about 20 reachable APs in a given area, they start to interfere destructively and just steal each other's bandwidth.
From various anecdotal info, it may be that WiFi coverage in some urban area is close to saturating: my officemate reports 19 visible APs in his appartment in downtown DC.
If most APs were open, sort of equilibrium might establish itself---people would realize that adding their own APs actually decreases their bandwidth, and so the Tragedy of the Commons scenario would be avoided. However, with APs locked down, everyone MUST install their own, and thus everyone is drawn into the ToC fight.
I put together this web-scraping chart comparing price/performance of available x86-compatible CPU families on http://dclug.tux.org/cpu.png. The daily-run script collects web-advertised prices, and displays them as a series of lines showing price-vs.-nominal clockspeed data within the CPU sub-families.
Note the logarithmic scale of the Y (price in US$) axis---in linear scale it's easier to see the knee in the curve, where additional speed increments begin to cost disproportionately more, but the linear graph was much less readable.
The method used is a hacked-together heuristic, easily fooled by vendors' ever inventive approaches to reporting the parameters of the CPUs they are selling. Still, it's a good visual aid showing how various CPU families are positioned with respect to each other.
150kW using photovoltics requires about 1000 sq metres of space in the
middle of the desert at high noon. You'll need about 4000-5000 sq metres
Solar energy flux is around 1.5 kW/m2, so desert/high noon would require 100 m2.
A generally accepted estimate of seasonal/diurnal average of not quite
20% efficency would require 750 square meters:
http://linas.org/theory/solar-electric.html
Old Boniface he took his cheer,
Then he bored a hole through a solid sphere,
Clear through the center, straight and strong,
So that the hole was just six inches long.
Now tell me, when the end was gained,
What volume in the sphere remained?
Sounds like I haven't told enough,
But I have, and the answer isn't tough!
(the answer isn't hard to calculate, but try proving that you don't really need to know the information that wasn't provided)
He has valid points about the different level of expectations for technology of strategic value. The conclusions, however, are flawed because he misses one of the fundamental principles of the Open Source methodology: conformance with Open Standards! In the real world, we can replace the the refrigerator (Strategic Food Storage Unit) easily. It'll cost in capital and labor, but it's not a tragedy.
Of course it is a good practice to watch out for vendor lock-in with any technology; one avoids it by selecting suppliers that don't play the lock-in game, and by paying attention to implementation details, refusing or at least controlling the use of unique, non-standard features. Open Source is just consistently making it easier to avoid the lock-in.
Why does Open Source have less lock-in? The source availability means that truly useful features propagate to all competing systems; also, the user community expects the providers to knock it off. It is therefore easier to deploy systems that are not critically dependent on any particular component. MySQL not good enough? fine, I will run my SQL queries against Postgress. Can people paint themselves in a corner? Sure, but it is not because they had no choice.
For what it is worth, I just calculated how this would work out, using 'units'
You have: (2$/gal) 70 mph / 35 mpg 100 hp
You want: cent/kWh
* 5.3640884
Explanation: 100 hp car engine can run a car at 70 mph, at 35 mpg. With gas at 2$/gal, this means that we are producing energy at the cost of ~5 cents per kiloWattHour, which is slightly less than my power company charges me. Since the engine power is not really 100hp at 70 mph, it probably costs 10-15 cents/kWh.
Not true---copyright contracts sometime contain clauses valid 'in all universe'. I remember one sci-fi writer modifying one such clause to say 'except in Andromeda'.
I always maintained that there is an analogy between the software technology and scientific knowledge. Just like science is the basis for our civilization, software underlies the expanding digital sphere of our lives. The development model of both science and sofware can vary between proprietary and public, and the society has to make a policy choice about supporting the right mix.
Even though scientific and technological knowledge started as proprietary, we as society made a historical choice, dating back to the age of Enlightenment, to develop knowledge in a collegial, public fashion. This model, of course, works rather well, and no one seriously argues that it should be rolled back to some kind of proprietary science development.
Similarly, I argue that software, whose importance tracks the growing influence of computing on our lives, must be developed in a public model; the Free Software is currently the closest approach, which eventually will be augmented by some sort of peer-reviewed public commitment, just as is the case for scientific research.
The analogy of software and science is not perfect; but I argue that, firstly, the negative effects of closed software are almost identical to negative effects of closed knowledge: it forces duplicate work, creates artificial monopolies, and slows down progress. Secondly, because software IS the infrastructure of the digital age, there is the issue of public interest, and the development model must accomodate that.
In this context, the strategy of SCO in their Linux lawsuit is especially retrograde. Their position, as laid out in their recently issued letters seems to counter the very idea of a public stake in technical knowledge. It occurred to me to modify their argument, substituting human knowledge' for 'software'. Here's what we'd get:
As you may know, the development process for public scientific
knowledge has differed substantially from the development process
for other enterprise scientific research. Commercial research is
built by carefully selected and screened teams of scientists
working to build proprietary scientific results. The process is
designed to monitor the security and ownership of intellectual
property rights associated with the knowledge.
By contrast, much of human scientific knowledge has been built
from contributions by numerous unrelated and unknown scientists,
each contributing a small scientific discovery. There is no
mechanism inherent in the public science development process to
assure that intellectual property rights, confidentiality or
security are protected. The public science process does not
prevent inclusion of knowledge that has been stolen outright, or
developed by improper use of proprietary methods and concepts.
Put this way, their argument is nonsensical, and would find no support in anyone even a tiny bit familiar with the scientific process, which arguably forms the basis of our civilization.
This has actually happened. Alexander Flemming refused to patent penicillin [...] The result was that nobody worked on making penicillin available in commercial quantities, because there was no money to be made doing so. This only changed when the military funded development in the early days of WW II. So Flemming's life-saving discovery went unused for more than a decade, because he wasn't greedy.
Can you provide any corroboration for that? As recalled by E.B.Chain, who first isolated the ctive penicillin, and subsequently was a co-recipient, with Fleming, of the Nobel Prize, it simply wasn't clear to anyone in 1928 that penicillin would have drug qualities. Chain claims that his motivation for isolating it came from his biochemical research, not as a drug. He doesn't mention military funding either.
Well, Altera had the Excalibur system out for several years: it's their large FPGA with VHDL synthesis tools, including several parametrized CPU cores: MIPS and ARM (requiring license) and their own NIOS CPU. The cores can be customized with different peripherals, and, in the case of NIOS, can be even modified to implement a 6-bit or a 32-bit CPU. The CPUs use up a fraction of tens of thousands of gates---the rest is available for custom digital logic designed using standard FPGA techniques.
There is a companion Linux implementation, running on the CPU. The basic Excalibur kit (software and eval board) is around $2k, and the Linux for NIOS kit costs another 2k, IIRC.
Furthermore, Altera and XILINX were planning (or maybe already have) FPGAS with hardwired CPU (ARM or MIPS) sitting in the corner. This tends to leave more gates for the programmable part, and ive better speed.
Would putting radioactive waste in old uranium mines satisfy the eco-nazis?
In the same vein,
last Nature has an article about natural toxins
(pesticides and other cancerogens) exceeding by
far the levels of artificial ones in food. It's ironic that natural food producers advertise the use of naturally occurring crop protection substances, as if it was healthier to eat a natural toxin.
Slashdot Mirror
I guess what you see is a bad thing, is actually a good thing in my book. Do you want your leader's
vote to be for sale to the most powerful lobby, or would you rather it be for sale to the public
opinion of the majority?
But this is a false dichotomy! I want a person that honestly believes in what I agree is right, and
I want the democratic majority to agree enough to elect that person. Naive? perhaps....
The Real Beauty of Irix is in its capbilities on the big multiprocessor Onyx systems.
/proc/cpuinfo' on our Altix:
/dev/sda':
So is Linux: this is the tail of 'cat
processor : 31
vendor : GenuineIntel
arch : IA-64
32 processors, 32 GB of memory. It is nice to be able to allocate an 8GB array.
SGI certainly has the street cred for well-rounded systems. Fast memory, fast
disk IO; check this output from 'hdparm -tT
Timing buffer-cache reads: 9900 MB in 2.04 seconds = 4850.53 MB/sec
Timing buffered disk reads: 268 MB in 3.01 seconds = 89.10 MB/sec
Another neat feature: the 'BIOS' console is available on the ethernet.
The console is actually a small embedded ARM computer, running Linux, of course,
which has access to the hardware of the big computer: it can power up/down the
nodes remotely, read temperature/voltage data, run hardware diagnostics etc.
The fascinating average temperature data from Vostok Antarctic ice data:
/ tempplot5.gif
http://cdiac.ornl.gov/trends/temp/vostok/graphics
shows that for last 450,000 years Earth was mostly in the Ice Age, interrupted
by 10,000 year long warm periods spaced 100,000 years apart. We are about 15,000
years into the last warm period on record.
Because of strong periodicity, the current best explanation of this cycle is
by astronomical phenomena (Earth orbit/axis wobble).
This does not contradict global warming---it just shows that the climate
is a very delicate balance between strong opposing phenomena; the point being
that we should be real careful how we influence it.
The Amazing Google has an online translator from Arabic (it's different than Farsi, of course): http://www.google.com/translate_t. It was pretty amazing to direct itc ial_results_release_20050801_v3.html
to Al-Jazeera arabic site and see the headlines in English. Google claims to have
submitted it to NIST translation tests, and beating competition to a punch:
http://www.nist.gov/speech/tests/mt/mt05eval_offi
We may be on a cusp here, in a 'golden age' of WiFi. Up to now, as more and more businesses and individuals deployed Access Points, we were seeing increasing coverage, but as the article says, at about 20 reachable APs in a given area, they start to interfere destructively and just steal each other's bandwidth.
From various anecdotal info, it may be that WiFi coverage in some urban area is close to saturating: my officemate reports 19 visible APs in his appartment in downtown DC.
If most APs were open, sort of equilibrium might establish itself---people would realize that adding their own APs actually decreases their bandwidth, and so the Tragedy of the Commons scenario would be avoided. However, with APs locked down, everyone MUST install their own, and thus everyone is drawn into the ToC fight.
I put together this web-scraping chart comparing price/performance of available x86-compatible CPU families on http://dclug.tux.org/cpu.png. The daily-run script collects web-advertised prices, and displays them as a series of lines showing price-vs.-nominal clockspeed data within the CPU sub-families.
Note the logarithmic scale of the Y (price in US$) axis---in linear scale it's easier to see the knee in the curve, where additional speed increments begin to cost disproportionately more, but the linear graph was much less readable.
The method used is a hacked-together heuristic, easily fooled by vendors' ever inventive approaches to reporting the parameters of the CPUs they are selling. Still, it's a good visual aid showing how various CPU families are positioned
with respect to each other.
Old Boniface he took his cheer,
Then he bored a hole through a solid sphere,
Clear through the center, straight and strong,
So that the hole was just six inches long.
Now tell me, when the end was gained,
What volume in the sphere remained?
Sounds like I haven't told enough,
But I have, and the answer isn't tough!
(the answer isn't hard to calculate, but try proving that you don't really need to know the information that wasn't provided)
He has valid points about the different level of expectations for technology of strategic value. The conclusions, however, are flawed because he misses one of the fundamental principles of the Open Source methodology: conformance with Open Standards! In the real world, we can replace the the refrigerator (Strategic Food Storage Unit) easily. It'll cost in capital and labor, but it's not a tragedy.
Of course it is a good practice to watch out for vendor lock-in with any technology; one avoids it by selecting suppliers that don't play the lock-in game, and by paying attention to implementation details, refusing or at least controlling the use of unique, non-standard features. Open Source is just consistently making it easier to avoid the lock-in.
Why does Open Source have less lock-in? The source availability means that truly useful features propagate to all competing systems; also, the user community expects the providers to knock it off. It is therefore easier to deploy systems that are not critically dependent on any particular component. MySQL not good enough? fine, I will run my SQL queries against Postgress. Can people paint themselves in a corner? Sure, but it is not because they had no choice.
OK, time to check out the wonderful 'units' program, which can convert between an amazing array of, well, units. It says:
i.e. a dozen batteries would have energy storage equal to a third of a kilogram of a TNT explosive.For what it is worth, I just calculated how this would work out, using 'units'
You have: (2$/gal) 70 mph / 35 mpg 100 hp
You want: cent/kWh
* 5.3640884
Explanation: 100 hp car engine can run a car at 70 mph, at 35 mpg. With gas at 2$/gal, this means that we are producing energy at the cost of ~5 cents per kiloWattHour, which is slightly less than my power company charges me. Since the engine power is not really 100hp at 70 mph, it probably costs 10-15 cents/kWh.
In space, copyright laws don't apply
Not true---copyright contracts sometime contain
clauses valid 'in all universe'. I remember one
sci-fi writer modifying one such clause to say
'except in Andromeda'.
I always maintained that there is an analogy between the software technology and scientific knowledge. Just like science is the basis for our civilization, software underlies the expanding digital sphere of our lives. The development model of both science and sofware can vary between proprietary and public, and the society has to make a policy choice about supporting the right mix.
Even though scientific and technological knowledge started as proprietary, we as society made a historical choice, dating back to the age of Enlightenment, to develop knowledge in a collegial, public fashion. This model, of course, works rather well, and no one seriously argues that it should be rolled back to some kind of proprietary science development.
Similarly, I argue that software, whose importance tracks the growing influence of computing on our lives, must be developed in a public model; the Free Software is currently the closest approach, which eventually will be augmented by some sort of peer-reviewed public commitment, just as is the case for scientific research.
The analogy of software and science is not perfect; but I argue that, firstly, the negative effects of closed software are almost identical to negative effects of closed knowledge: it forces duplicate work, creates artificial monopolies, and slows down progress. Secondly, because software IS the infrastructure of the digital age, there is the issue of public interest, and the development model must accomodate that.
In this context, the strategy of SCO in their Linux lawsuit is especially retrograde. Their position, as laid out in their recently issued letters seems to counter the very idea of a public stake in technical knowledge. It occurred to me to modify their argument, substituting human knowledge' for 'software'. Here's what we'd get:
As you may know, the development process for public scientific
knowledge has differed substantially from the development process
for other enterprise scientific research. Commercial research is
built by carefully selected and screened teams of scientists
working to build proprietary scientific results. The process is
designed to monitor the security and ownership of intellectual
property rights associated with the knowledge.
By contrast, much of human scientific knowledge has been built
from contributions by numerous unrelated and unknown scientists,
each contributing a small scientific discovery. There is no
mechanism inherent in the public science development process to
assure that intellectual property rights, confidentiality or
security are protected. The public science process does not
prevent inclusion of knowledge that has been stolen outright, or
developed by improper use of proprietary methods and concepts.
Put this way, their argument is nonsensical, and would find no support in anyone even a tiny bit familiar with the scientific process, which arguably forms the basis of our civilization.
This has actually happened. Alexander Flemming refused to patent penicillin [...] The result was that nobody worked on making penicillin available in commercial quantities, because there was no money to be made doing so. This only changed when the military funded development in the early days of WW II. So Flemming's life-saving discovery went unused for more than a decade, because he wasn't greedy.
Can you provide any corroboration for that? As recalled by E.B.Chain, who first isolated the ctive penicillin, and subsequently was a co-recipient, with Fleming, of the Nobel Prize, it simply wasn't clear to anyone in 1928 that penicillin would have drug qualities. Chain claims that his motivation for isolating it came from his biochemical research, not as a drug. He doesn't mention military funding either.
So, maybe, greed is not good?
Well, Altera had the Excalibur system out for several years: it's their large FPGA with VHDL synthesis tools, including several parametrized CPU cores: MIPS and ARM (requiring license) and their own NIOS CPU. The cores can be customized with different peripherals, and, in the case of NIOS, can be even modified to implement a 6-bit or a 32-bit CPU. The CPUs use up a fraction of tens of thousands of gates---the rest is available for custom digital logic designed using standard FPGA techniques.
There is a companion Linux implementation, running on the CPU. The basic Excalibur kit (software and eval board) is around $2k, and the Linux for NIOS kit costs another 2k, IIRC.
Furthermore, Altera and XILINX were planning (or maybe already have) FPGAS with hardwired CPU (ARM or MIPS) sitting in the corner. This tends to leave more gates for the programmable part, and ive better speed.
In other words, nothing new under the sun here.
Would putting radioactive waste in old uranium mines satisfy the eco-nazis?
In the same vein,
last Nature has an article about natural toxins
(pesticides and other cancerogens) exceeding by
far the levels of artificial ones in food. It's ironic that natural food producers advertise the use of naturally occurring crop protection substances, as if it was healthier to eat a natural toxin.
A dose is a dose is a dose.