The difference? The AMD rack contains many independent machines, the SGI rack is one machine.
A nice quote from SGI's overview: The architecture is designed to run complex models frequently, and because the entire memory space is shared, large models fit into memory with no programming model restrictions.
You wont have 672 GB of shared memory in your AMD rack!
if the complete source is out it will be much easier for someone to put together a full leecher client... and if that becomes very popular the whole network will become untenable.:(
I never thought security through obscurity was a viable philosophy longterm, but it's better than nothing. What now? Have any of the developers addressed this that you know of?
They have. EMule has also introduced a credit system, where you remember who you've downloaded from and how much, and based on that, give the person an elevated queue position should he want to download from you again. This is fairly hacker proof, as there's no way you can cheat yourself to download more than you should.
The actual effect on the network is still beeing evaluated afaik, but around 80% of the ones I upload to, are clients I've gotten something from.
If it takes one CPU-day to do 1ns of folding simulation, then protein folding is difficult to simulate because it occurs over a (relatively) long, not short time.
From what I've gathered reading the papers at their site, the folding process happens in "steps". The changes between states occur quickly, and then the molecules stay in those states for a relativly long time. Thats why it's a problem, because you need very small timesteps to register the state changes properly.
Re:DeCSS is neither for Linux, nor for piracy
on
DeCSS Update
·
· Score: 1
Remember that MPAA/RIAA members do own the copyright on their movies and music, and they are not obligated to distribute or sell them to you at all. Yeah, it would be much better for them to just sit on the movies and watch us cry....
As others have commented, your statement isnt very valid. In the mars issue Scientific American there was an article about swarm intelligence, one variant of AI. They had some examples on how simulating ant behaviour allowed for intelligent routing of trafic in telephone networks. The technique had also been extended for internet routing, and tests showed dramatically improved throughput and reliability compared to the protocol currently in use.
This is just one example of AI that is in no way limited to emulating human intelligence, and that we should very much bother developing AI.
I don't think AI will ever be "perfect" in the way you seem to mean, because I don't belive that AI will every become a single "thing". Rather it will develop into an area where we'll see AI specialiced for various tasks, and outperforming humans.
I would guess that your A- came from your good arguments for your views, not the validity of the views. (not that this is a bad thing)
Slashdot Mirror
It's been pointed out before but:
CPU: 2 AMD Opteron 64-bit capable x86-64 architecture Processors
Memory: 512MB - 16GB ECC, Registered DDR333 SDRAM
Compare that to 128 processors and 256 GB of RAM.
The difference? The AMD rack contains many independent machines, the SGI rack is one machine.
A nice quote from SGI's overview: The architecture is designed to run complex models frequently, and because the entire memory space is shared, large models fit into memory with no programming model restrictions.
You wont have 672 GB of shared memory in your AMD rack!
if the complete source is out it will be much easier for someone to put together a full leecher client... and if that becomes very popular the whole network will become untenable. :(
I never thought security through obscurity was a viable philosophy longterm, but it's better than nothing. What now? Have any of the developers addressed this that you know of?
They have. EMule has also introduced a credit system, where you remember who you've downloaded from and how much, and based on that, give the person an elevated queue position should he want to download from you again. This is fairly hacker proof, as there's no way you can cheat yourself to download more than you should.
The actual effect on the network is still beeing evaluated afaik, but around 80% of the ones I upload to, are clients I've gotten something from.
If it takes one CPU-day to do 1ns of folding simulation, then protein folding is difficult to simulate because it occurs over a (relatively) long, not short time.
From what I've gathered reading the papers at their site, the folding process happens in "steps". The changes between states occur quickly, and then the molecules stay in those states for a relativly long time.
Thats why it's a problem, because you need very small timesteps to register the state changes properly.
Remember that MPAA/RIAA members do own the copyright on their movies and music, and they are not obligated to distribute or sell them to you at all.
Yeah, it would be much better for them to just sit on the movies and watch us cry....
Not to mention the man that put US on the moon.... ...and the list goes on...
As others have commented, your statement isnt very valid.
In the mars issue Scientific American there was an article about swarm intelligence, one variant of AI. They had some examples on how simulating ant behaviour allowed for intelligent routing of trafic in telephone networks. The technique had also been extended for internet routing, and tests showed dramatically improved throughput and reliability compared to the protocol currently in use.
This is just one example of AI that is in no way limited to emulating human intelligence, and that we should very much bother developing AI.
I don't think AI will ever be "perfect" in the way you seem to mean, because I don't belive that AI will every become a single "thing". Rather it will develop into an area where we'll see AI specialiced for various tasks, and outperforming humans.
I would guess that your A- came from your good arguments for your views, not the validity of the views. (not that this is a bad thing)