Beowulf is a specific project/software for doing clusters. In reality, it is not that popular. There are lots of different "whole clustering solutions", and beowulf is one of those. Even more common in the HPC world is probably homegrown solutions, based on common components.
How about another quote, this time from a video game (Alpha Centauri):
As the Americans learned so painfully in Earth's final century,
free flow of information is the only safeguard against tyranny.
The once-chained people whose leaders at last lose their grip on
information flow will soon burst with freedom and vitality, but
the free nation gradually constricting its grip on public discourse
has begun its rapid slide into despotism. Beware of he who would
deny you access to information, for in his heart he dreams himself
your master.
The top500 list only includes existing supercomputers, not future ones. You have to run the benchmark, not guess how fast it will go.
Now, for a more "realistic" benchmark than hplinpack, this has been tried and talked about for quite some time. It is a hard problem actually, because different supercomputers are designed for different usage. HPL is a useful upper bound for realistic calculations over the whole computer, but it is far from the whole truth.
The BlueGenes out there have don real work, in doing the signal processing of a distributed radio telescope (the one in the netherlands) and protein folding/molecular dynamics (the US one).
And while efficiency might be important, remember that if you can get a machine twice as big by going down to 90% of the efficiency for the same price, the smart move is usually, but not always, to buy the larger machine.
/Mattias Wadenstein - sysadmin at #388 on the list
Not really, but with rsync or jigdo you can reuse most of the data.
Otherwise, the "patch" would be to manually add the security.debian.org line in sources.list after installation. Just like it says in the errata in the grandparent to this comment.
Sorry about that, one mirror that ran out of space at the middle of syncing. Thanks for the report, I've fixed it now.
Please tell me if there are any other problems, this was the first time I heard about these 404s. Btw, saying which links will help even more, in this case I'm guessing at the powerpc isos?
Assuming you stay within the hand-tuned codes that are available. You won't see anything near that performance from compiled code, and that's with a good compiler. I've done some tests myself, and the G5 performes about the same clock-for-clock as the Opteron. And these days, the Opteron clocks a bit higher...
Yes, I got that detail on removal/modification wrong. And the rest of the info fits with my memory. Now I know what to look for when I try to find it for reading it again.:)
I don't think so, since I remember this to be placed in a collection of Asimov's robot stories and featuring Susan Calvin. Besides, I've never heard of Roger Macbride Allen and I usually remember authors. The rest of it seems fairly foreign too, so I don't think it comes from there.
In the Asimov robot world, this was an early failure that Calvin took care of and then was remembered as evidence for the "don't mess with the three laws"-rule.
Oh, but they did break the three laws of robotics in the most obvious way in one of his stories, a number of robots were manufactured without the first law (never harm a human being...).
The plot of the story was that one of these robots made it into a large population of robots with all three laws and Susan Calvin had to sort it out (while of course saying that these lesser roboticists were morons for creating robots without the first law).
Unfortunately I do not remember the title of the short story, nor the collection in which it appeared, it was over 10 years I read it. Perhaps someone else can help me with that? This would probably be a fairly early collection of short stories by Asimov.
To help the memory, the reason they were created was to serve as help on a science station (in space, I think), where humans were submitted to low levels of radiation which killed the positronic brains instantly.
The scientists accepted the low risk of harm for the sake of observing whatever it was, but the the three laws of robotics didn't allow the robots to idly stand by and let even that low level of harm happen. And when they rushed in, they were instantly killed by radiation.
We still have those rights (to record a radio show or copy a friend's CD for personal use), they weren't taken away by this new law. We have had those for a long time now, at least since 1960.
Yes, this is true. It is legal to make a copy for personal use. This has been commented upon by the Minister of Justice as a problem for getting at filesharing, where only the one sharing might be commiting a crime, the downloader is perfectly safe.
The big difference in this new law is that for making this personal copy the thing you are copying from has to be a legal copy. Essentially taking care of a loophole in the law.
Allowing copying for private use is also the reasoing behind the CD-fee, even if that is highly annoying for me that only use it for software (debian boot/install-discs).
For all of those claiming "the internet is now illegal", there is both provisions for temporary copies (as in the ones your browser are making) and a discussion about "good faith" in the paper. They are expecting to be able to go after filesharers and allow ISPs to disconnect users under the "we'll cut your access if you break any laws" sections of the AUP if they detect this stuff.
All in all, I was fearing a worse law after reading the press release, even the law against anti-circumvention tools have provisions to allow DVD-players, even if DVD-copiers might be disallowed. They even make an example of region coding not being an effective technical meassure.
Sincgle-use card numbers with all that you describe are already here. My bank here in Sweden offers this for their bank cards, and if your normal bank card includes a Visa function, your one-use number also is a Visa card number.
All the functions you say, first vendor, N transactions, N months. And also a charge limit, so that you can't lose too much money from a bad company either. I'm actually not afraid to give out a credit card number to companies I've never heard of anymore.
The bank? Föreningssparbanken in Sweden./Mattias Wadenstein
So Fed-Ex is shipping in less than 10 seconds these days? A 1 Gbit link is just a matter of some money. Any big ISP should be able to sell you that and actually deliver as long as both endpoints are on the same, bigish, ISP.
If we were talking a factor 1000 more than 1GB, the Fed-Ex solution might have some validity. But then you have to take into account high-bandwidth tape drives in parallell and so on. My guess that you would be very hardpressed to find a solution that is faster than the internet for getting data from one file system in one city to another file system in another city./Mattias Wadenstein
Well, 250 ms ping times seems resonable. That is what I see from here in Sweden to alaska.edu on an i2 link. I would guess that they get at most 50 ms lower ping times.
And if my quick calculations are right (the distance and c is known, the difference between c and the glass medium can found out (I didn't, I just guessed, these were quick calculations)), about half of that is the propagation time for the light.
That is the big problem with long distances, there is a hard limit on how low the latency can get.
Sure, the other servers would keep on truckin'. But how would your resolving nameserver know which server handles the ".com" or ".se" domains?
Even if the nameservers for.com still worked, you wouldn't know where to find them. That is where the root nameservers come in. Those are the only ones that all nameservers know the IP of and can ask where all the other ones are.
If you want to know more, you can take a look at the Linux DNS Howto, I found it very informative even for non-Linux situations.
Well, so far I have seen 3 different stories. And a 4th one in a comment (that looks like it fits along with the other three comments).
All the versions I've seen personally in chronological order:
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it. And when our qualified personel arrived, we discovered that she wasn't actuually as qualified as we had hoped. Then she quit. Hemos or I will update this story as soon as we know what the hell happened
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it. Hemos or I will update this story as soon as we know what the hell happened.
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it, thus triggering the end of several nearby worlds as well. Props to Yazz, KurtG and Scott from Cisco for managing to help get us back online. We'll post more when we know it.
The one from another comment:
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it, thus triggering the end of several nearby worlds as well. And when our qualified personel arrived, we discovered that she wasn't actuually as qualified as we had hoped. Then she quit, thus terminating 3 local star systems. Hemos or I will update this story as soon as we know what the hell happened. But apparently creds go to Kurt Grey and Cisco tech support. Hopefully we'll have more info soon.
And for those that missed that part, the story was originally:
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it. And when our qualified personel arrived, we discovered that she wasn't actuually as qualified as we had hoped. Then she quit. Hemos or I will update this story as soon as we know what the hell happened
Hope this clears up what that poster was trying to say for all those late-comers.
Slashdot Mirror
Now, for a more "realistic" benchmark than hplinpack, this has been tried and talked about for quite some time. It is a hard problem actually, because different supercomputers are designed for different usage. HPL is a useful upper bound for realistic calculations over the whole computer, but it is far from the whole truth.
The BlueGenes out there have don real work, in doing the signal processing of a distributed radio telescope (the one in the netherlands) and protein folding/molecular dynamics (the US one).
And while efficiency might be important, remember that if you can get a machine twice as big by going down to 90% of the efficiency for the same price, the smart move is usually, but not always, to buy the larger machine.
Otherwise, the "patch" would be to manually add the security.debian.org line in sources.list after installation. Just like it says in the errata in the grandparent to this comment.
Please tell me if there are any other problems, this was the first time I heard about these 404s. Btw, saying which links will help even more, in this case I'm guessing at the powerpc isos?
As it is now, I have to apologize to the mirrors I asked for, since they aren't getting any load either.
So, bring it on!
And if we run out, we will do http-redirects to our mirrors around the world, so don't be afraid to get your Sarge now!
Well, we fixed it now. Enjoy! /maswan - admin of ftp.gnome.org
Here is a mirror of the english trailer:
http://ftp.acc.umu.se/mirror/temp/lnluksm.avi
The current server seems a bit slow to respond, so..
Assuming you stay within the hand-tuned codes that are available. You won't see anything near that performance from compiled code, and that's with a good compiler. I've done some tests myself, and the G5 performes about the same clock-for-clock as the Opteron. And these days, the Opteron clocks a bit higher...
Thank you!
:)
Yes, I got that detail on removal/modification wrong. And the rest of the info fits with my memory. Now I know what to look for when I try to find it for reading it again.
I don't think so, since I remember this to be placed in a collection of Asimov's robot stories and featuring Susan Calvin. Besides, I've never heard of Roger Macbride Allen and I usually remember authors. The rest of it seems fairly foreign too, so I don't think it comes from there.
In the Asimov robot world, this was an early failure that Calvin took care of and then was remembered as evidence for the "don't mess with the three laws"-rule.
Oh, but they did break the three laws of robotics in the most obvious way in one of his stories, a number of robots were manufactured without the first law (never harm a human being...).
The plot of the story was that one of these robots made it into a large population of robots with all three laws and Susan Calvin had to sort it out (while of course saying that these lesser roboticists were morons for creating robots without the first law).
Unfortunately I do not remember the title of the short story, nor the collection in which it appeared, it was over 10 years I read it. Perhaps someone else can help me with that? This would probably be a fairly early collection of short stories by Asimov.
To help the memory, the reason they were created was to serve as help on a science station (in space, I think), where humans were submitted to low levels of radiation which killed the positronic brains instantly.
The scientists accepted the low risk of harm for the sake of observing whatever it was, but the the three laws of robotics didn't allow the robots to idly stand by and let even that low level of harm happen. And when they rushed in, they were instantly killed by radiation.
We still have those rights (to record a radio show or copy a friend's CD for personal use), they weren't taken away by this new law. We have had those for a long time now, at least since 1960.
Yes, this is true. It is legal to make a copy for personal use. This has been commented upon by the Minister of Justice as a problem for getting at filesharing, where only the one sharing might be commiting a crime, the downloader is perfectly safe.
The big difference in this new law is that for making this personal copy the thing you are copying from has to be a legal copy. Essentially taking care of a loophole in the law.
Allowing copying for private use is also the reasoing behind the CD-fee, even if that is highly annoying for me that only use it for software (debian boot/install-discs).
For all of those claiming "the internet is now illegal", there is both provisions for temporary copies (as in the ones your browser are making) and a discussion about "good faith" in the paper. They are expecting to be able to go after filesharers and allow ISPs to disconnect users under the "we'll cut your access if you break any laws" sections of the AUP if they detect this stuff.
All in all, I was fearing a worse law after reading the press release, even the law against anti-circumvention tools have provisions to allow DVD-players, even if DVD-copiers might be disallowed. They even make an example of region coding not being an effective technical meassure.
Sincgle-use card numbers with all that you describe are already here. My bank here in Sweden offers this for their bank cards, and if your normal bank card includes a Visa function, your one-use number also is a Visa card number.
/Mattias Wadenstein
All the functions you say, first vendor, N transactions, N months. And also a charge limit, so that you can't lose too much money from a bad company either. I'm actually not afraid to give out a credit card number to companies I've never heard of anymore.
The bank? Föreningssparbanken in Sweden.
So Fed-Ex is shipping in less than 10 seconds these days? A 1 Gbit link is just a matter of some money. Any big ISP should be able to sell you that and actually deliver as long as both endpoints are on the same, bigish, ISP.
/Mattias Wadenstein
If we were talking a factor 1000 more than 1GB, the Fed-Ex solution might have some validity. But then you have to take into account high-bandwidth tape drives in parallell and so on. My guess that you would be very hardpressed to find a solution that is faster than the internet for getting data from one file system in one city to another file system in another city.
And if my quick calculations are right (the distance and c is known, the difference between c and the glass medium can found out (I didn't, I just guessed, these were quick calculations)), about half of that is the propagation time for the light.
That is the big problem with long distances, there is a hard limit on how low the latency can get.
This is exactly this problem that was adressed with that patch referenced in the story.
Sure, the other servers would keep on truckin'. But how would your resolving nameserver know which server handles the ".com" or ".se" domains?
.com still worked, you wouldn't know where to find them. That is where the root nameservers come in. Those are the only ones that all nameservers know the IP of and can ask where all the other ones are.
Even if the nameservers for
If you want to know more, you can take a look at the Linux DNS Howto, I found it very informative even for non-Linux situations.
/Mattias Wadenstein
Well... I made a better guess of where to find it on ftp.gnome.org and I found it. We also added some symlinks to make the published url work.
s es /gnome-2.0-lib-alpha1/
What should have been posted:
http://download.gnome.org
Or if people want it from ftp.gnome.org (which really is just another mirror):
http://ftp.gnome.org/pub/GNOME/pre-gnome2/relea
Or ftp:// for those that still use that. http is a much nicer protocol from our point of view.
/Mattias Wadenstein - Admin of ftp.acc.umu.se aka ftp.gnome.org
All the versions I've seen personally in chronological order:
The one from another comment:
And for those that missed that part, the story was originally:
I'm still not exactly clued in as to why we're back online, but hey, we are. Sometime saturday morning our Cisco router melted down. Ordinarily this would only be the end of the world, but none of our qualified personel were available to fix it. And when our qualified personel arrived, we discovered that she wasn't actuually as qualified as we had hoped. Then she quit. Hemos or I will update this story as soon as we know what the hell happened
Hope this clears up what that poster was trying to say for all those late-comers.
/Mattias Wadenstein