Couple issues with this as an alternative to the garden-variety x86 cluster connected with InfiniBand:
Slow network interconnect. For problems that are not trivially parallel, network latency is usually a big deal. Ethernet doesn't cut it. Lack of RAM. 'Nuff said. Have to care about Cell and PS3 architecture. The codes ("codes" has a slightly different meaning in the context of supercomputing) have to be modified to take advantage of this very specific architecture. Software always outlives hardware, so in the long run the effort may not be worth it.
That said, it's really cheap. If your application isn't held back too much by these issues then enjoy your insanely cheap cluster!
Wow, you're dense. The judge was clearly being sarcastic in this comment. If you read the rest of the text, it seems awfully like the judge was trying to help Mr. Atkinson by implying that his son should tell the plaintiffs he has no money, having just graduated from school, rather than Mr. Atkinson giving his son money to pay the settlement.
This judge clearly feels that while the RIAA is legally correct, their heavy-handed approach to dealing with people who don't really cause all that much damage, cannot afford a lawyer, and have no clue about the legal system, is a despicable way to attempt to solve their problems.
Windows Server 2008 and some other Microsoft softare is free to university students, from Microsoft, through their "DreamSpark" program. I am downloading it, legally, right now.
Are 64 cores enough for a lot of complex biology and physics simulations? Seriously, if the protein folding experiments are any indication, wouldn't this just be a drop in the bucket of the power needed to crunch those simulations?
I use 32-64 cores at a time for my ~85k atom molecular dynamics simulations, which I run on a Linux cluster. I don't know of any serious MD simulations being done on Windows anyway, and I can't imagine recommending this box for that purpose.
3. IBM has a compelling strategy of reusing their high-end POWER-X processor super-servers, and selling them as supercomputers. The problem with this, is that they are obscenely expensive as supercomputers.
Yup. I was quoted a price on an IBM POWER6 cluster - compared to a Xeon cluster, 10x the price for 1.5x actual work done. I guess there is a niche IBM sells to, but it sure isn't me.
As someone who does science HPC for a living, I am confused. Who actually wants Windows for HPC? What value does it provide that Linux or UNIX doesn't? I've never heard of a single use case where Linux or some UNIX wasn't better by miles.
I have a BlackBerry and happily code for it in Java. I use a Mac with VMware to run Windows so I can run the SDK's tools.
Pros: - The device OS is itself written in Java, so your Java apps are not second class citizens - API is intelligently designed and easy to work with - Good API documentation - Simulator runs the actual OS code - Apps can be loaded using a USB cable - Apps can be debugged as they run on the device itself through that USB cable - SDK downloadable for free in a few clicks - Lots of people have BlackBerries - No retarded NDA!
Cons: - Device OS is not open - SDK is Windows-only - SDK's IDE is garbage (but you can use Eclipse)
It is a question of what is more important to you: 100% hardware support or freedom.
Software "GNU/Freedom" is great but I care less about that issue than I do about getting my work done. I think there is a knock-on effect at work here that indirectly hurts adoption of distros like gNewSense, even among people who agree with its goals.
I need 3D acceleration on my Linux box to do my work, so I use Ubuntu with the nvidia binary blob. Because I like the GNU/FSF ideals, I could use gNewSense on other machines that don't require the binary blob, but that means I have to learn how to administer that distro in addition to Ubuntu. I don't have time for that, especially when I know Ubuntu works, so I just use that.
(Replace "Ubuntu" with "Windows" and "gNewSense" with "Linux" for a parallel argument.)
I understood the first six things on the list. But what the heck is a "Gayaplex"? I'm going to assume that it doesn't mean what its spelling implies...
I'm not sure what 9.4T MRI scanners cost these days, but you really need to get to that resolution to diagnose anything other than the most coarse-grained of stuff.
Correct me if I'm wrong, but you don't use magnets that strong on humans. The most powerful clinical MRI I've seen is 3T. (Granted I haven't paid attention in a few years, but still, 9.4T is quite a lot.)
We are actually fortunate that the current actor is so impotent in his role. Imagine what might happen if some charming, efficacious attorney took his place. The more I consider it the more I think we may be lucky to have Jack playing the part of the alarmist. The alternative might be someone who is actually capable.
Now that it is even more obvious that Thompson is a nutcase, who is going to take his place? Someone competent in that role? That is a scary thought.
But to be fair to them, my CFO asked a little while ago if the power problems we had were a result of her sending an email to Iceland. After all, it must take a lot more power to push the message that far than to push it across the street.
This is the funniest thing I've read all day. I can't blame the CFO for not knowing better; after all it's not her area of competence - presumably that's why she employs you. But still...imagine if she sent *two* emails to Iceland!
1. Don't steal someone else's copyright. Just don't. I know it doesn't sound like much, but the way I see it, an author's book, for whatever reason they choose to write it, is their intellectual heart.
Every textbook is not so much an original work as it is a collection of knowledge in the field. Are you compensating the authors of the countless papers whose hard-won research results you're aggregating and selling? Didn't think so. So don't be so precious about "intellectual heart".
4. The publishing model is growing outdated, but until you can get your professors to choose books that are online only, or to embrace the digital age, we've hit a wall.
This is happening - and the idea of using a new textbook, online or physical, is being rejected outright. Surprisingly few of my grad school classes at major research institutions used current textbooks - instead, the professors assigned papers to read and/or provided their own notes.
The amount of freely-available, high-quality content that can be used to replace textbooks will only grow. The days of expensive textbooks are numbered.
"The Design and Evolution of C++" by Stroustrup is a must-read if you are interested in why C++ is the way it is.
After reading it, I really hated C++. It's the classic example of a project that gets ruined by too many people working on it, all with their own goals, and the book tells you exactly why this happened. C++ now is a hideously complex monstrosity that is popular because it is all things to all people, not because it is a good language.
Anyway, if you disagree with me, have a look at the book. It is a testament to Stroustrup's objectivity that I came to the conclusion I did, and that you may come to the exact opposite conclusion as me after reading it.
Can someone explain why anyone could possibly want Windows on a scientific computing cluster? What does Windows offer that Linux doesn't?
Much of my work involves running molecular dynamics simulations. By HPC standards these are tiny calculations (in my case, usually 32 CPUs at a time). All science HPC software I'm aware of is Unix-oriented, and everything runs on Linux. At my institution we have an OS X cluster and we are in the process of purchasing a Linux cluster. We didn't even consider Windows - given the difficulties we've experienced administering Windows on the desktop, a Windows cluster just seems like an expensive exercise in frustration.
Fair enough. For then and following your own analogy, when you use someone else's connection you should behave as when you are in someone's house. For example, when you visit someone you obviously are going to be with them, and they are going to see what you do. Would it be OK if I had this running, and therefore could listen to your conversations? I wouldn't like it, but I would have no basis to complain about it. Anything sent over the wire or air in plaintext is visible to everyone, so I can't possibly expect privacy through an open access point. You aren't the only one I have to trust not to sniff my plaintext data - I also have to trust everyone who has access to each router at every hop between me and the destination.
Being a nerd I knew this already, but getting the general public to understand this concept is a bit harder...
As it is officially a crime to steal wi-fi (Title 18, Part 1, Chapter 47 of the United States Code, which covers anybody who "intentionally accesses a computer without authorization or exceeds authorized access")." Would this apply to an access point which advertises its SSID and doesn't demand credentials from users? I would argue that it authorizes everyone to use it. To draw an analogy, it isn't just leaving your front door unlocked, it's leaving it unlocked and putting up a sign that says "Please come in!". So I don't see how accessing an open access point is a officially a crime.
[...]you can honestly gain about the same degree of skill as an average Ph. D. program will impart with about three months of research. Nonsense. As a PhD student myself, I believe that the PhD program in question would have to be crap and your advisor an idiot for this to be true. If this is the case for you, you are not in the right place and/or you need a new advisor.
Consider whether you really need to subject yourself to one. But this is a good point. You should only undertake a PhD if you find it intrinsically edifying - all other reasons are secondary.
Why do slashdotters feel the need to brag about how little spam they get? I mean seriously, it's not like you're running your own smtp server out there and programming your own rules. You signed up for a gmail account. You might as well say, "The IT staff that manages my email server has configured the rules so effectively, I rarely get any spam." I think the GP meant to compliment Google on how well they do spam filtering, not show off to a very Internet-savvy crowd how well he or she can sign up for a free account online.
I, too, have a Gmail account and get very little spam. Hooray for Google.
Who cares whether they are manipulative, dishonest, or unaccountable? Let it stay accessible and let the public decide for themselves.
I really dislike Fox News, but even though they are a poor excuse for a news organization, they shouldn't be shut down. We all have the same right to free speech.
Slashdot Mirror
Please allow me to contribute to your efforts.
248
We simply don't understand the human body well enough to know why some things work and why others don't.
Correct, but after this the correctness of your comment ends.
The human psyche plays a significant role that pure science doesn't admit to because it can't be proven in a test scenario.
It's called the "placebo effect" and pure science has shown it many, many times, in many, many test scenarios.
We know the human body gives off energy but people refuse to accept the "auras" are possible or significant for some reason.
Define "energy". Heat? Or something else? Show some kind of evidence backed by data rather than groundless assertions of the significance of auras.
We know every brain has a distinct pattern with a general consistency to that pattern, but we refuse to believe it's anything more than electrical.
What do you propose it is then?
If there's anything to be learned from this, it's that God is a lawyer who is easily kept happy through the use of legal technicalities.
Couple issues with this as an alternative to the garden-variety x86 cluster connected with InfiniBand:
Slow network interconnect. For problems that are not trivially parallel, network latency is usually a big deal. Ethernet doesn't cut it.
Lack of RAM. 'Nuff said.
Have to care about Cell and PS3 architecture. The codes ("codes" has a slightly different meaning in the context of supercomputing) have to be modified to take advantage of this very specific architecture. Software always outlives hardware, so in the long run the effort may not be worth it.
That said, it's really cheap. If your application isn't held back too much by these issues then enjoy your insanely cheap cluster!
Wow, you're dense. The judge was clearly being sarcastic in this comment. If you read the rest of the text, it seems awfully like the judge was trying to help Mr. Atkinson by implying that his son should tell the plaintiffs he has no money, having just graduated from school, rather than Mr. Atkinson giving his son money to pay the settlement.
This judge clearly feels that while the RIAA is legally correct, their heavy-handed approach to dealing with people who don't really cause all that much damage, cannot afford a lawyer, and have no clue about the legal system, is a despicable way to attempt to solve their problems.
Windows Server 2008 and some other Microsoft softare is free to university students, from Microsoft, through their "DreamSpark" program. I am downloading it, legally, right now.
Are 64 cores enough for a lot of complex biology and physics simulations? Seriously, if the protein folding experiments are any indication, wouldn't this just be a drop in the bucket of the power needed to crunch those simulations?
I use 32-64 cores at a time for my ~85k atom molecular dynamics simulations, which I run on a Linux cluster. I don't know of any serious MD simulations being done on Windows anyway, and I can't imagine recommending this box for that purpose.
they should fix the GLUT license.
Use SDL instead. IIRC it does a fair amount of what GLUT does.
3. IBM has a compelling strategy of reusing their high-end POWER-X processor super-servers, and selling them as supercomputers. The problem with this, is that they are obscenely expensive as supercomputers.
Yup. I was quoted a price on an IBM POWER6 cluster - compared to a Xeon cluster, 10x the price for 1.5x actual work done. I guess there is a niche IBM sells to, but it sure isn't me.
As someone who does science HPC for a living, I am confused. Who actually wants Windows for HPC? What value does it provide that Linux or UNIX doesn't? I've never heard of a single use case where Linux or some UNIX wasn't better by miles.
I have a BlackBerry and happily code for it in Java. I use a Mac with VMware to run Windows so I can run the SDK's tools.
Pros:
- The device OS is itself written in Java, so your Java apps are not second class citizens
- API is intelligently designed and easy to work with
- Good API documentation
- Simulator runs the actual OS code
- Apps can be loaded using a USB cable
- Apps can be debugged as they run on the device itself through that USB cable
- SDK downloadable for free in a few clicks
- Lots of people have BlackBerries
- No retarded NDA!
Cons:
- Device OS is not open
- SDK is Windows-only
- SDK's IDE is garbage (but you can use Eclipse)
It is a question of what is more important to you: 100% hardware support or freedom.
Software "GNU/Freedom" is great but I care less about that issue than I do about getting my work done. I think there is a knock-on effect at work here that indirectly hurts adoption of distros like gNewSense, even among people who agree with its goals.
I need 3D acceleration on my Linux box to do my work, so I use Ubuntu with the nvidia binary blob. Because I like the GNU/FSF ideals, I could use gNewSense on other machines that don't require the binary blob, but that means I have to learn how to administer that distro in addition to Ubuntu. I don't have time for that, especially when I know Ubuntu works, so I just use that.
(Replace "Ubuntu" with "Windows" and "gNewSense" with "Linux" for a parallel argument.)
From the site:
10.
1024×600.
1.1Kg.
4 hours.
Mandriva Linux.
G-Key.
Gayaplex.
I understood the first six things on the list. But what the heck is a "Gayaplex"? I'm going to assume that it doesn't mean what its spelling implies...
I'm not sure what 9.4T MRI scanners cost these days, but you really need to get to that resolution to diagnose anything other than the most coarse-grained of stuff.
Correct me if I'm wrong, but you don't use magnets that strong on humans. The most powerful clinical MRI I've seen is 3T. (Granted I haven't paid attention in a few years, but still, 9.4T is quite a lot.)
Penny Arcade made a very good point a few years ago, when Thompson was threatening them:
Now that it is even more obvious that Thompson is a nutcase, who is going to take his place? Someone competent in that role? That is a scary thought.
But to be fair to them, my CFO asked a little while ago if the power problems we had were a result of her sending an email to Iceland. After all, it must take a lot more power to push the message that far than to push it across the street.
This is the funniest thing I've read all day. I can't blame the CFO for not knowing better; after all it's not her area of competence - presumably that's why she employs you. But still...imagine if she sent *two* emails to Iceland!
Should have called the cops on him anyway. Any med student who steals laptops is dishonest, and therefore doesn't deserve to become a doctor.
1. Don't steal someone else's copyright. Just don't. I know it doesn't sound like much, but the way I see it, an author's book, for whatever reason they choose to write it, is their intellectual heart.
Every textbook is not so much an original work as it is a collection of knowledge in the field. Are you compensating the authors of the countless papers whose hard-won research results you're aggregating and selling? Didn't think so. So don't be so precious about "intellectual heart".
4. The publishing model is growing outdated, but until you can get your professors to choose books that are online only, or to embrace the digital age, we've hit a wall.
This is happening - and the idea of using a new textbook, online or physical, is being rejected outright. Surprisingly few of my grad school classes at major research institutions used current textbooks - instead, the professors assigned papers to read and/or provided their own notes.
The amount of freely-available, high-quality content that can be used to replace textbooks will only grow. The days of expensive textbooks are numbered.
"The Design and Evolution of C++" by Stroustrup is a must-read if you are interested in why C++ is the way it is.
After reading it, I really hated C++. It's the classic example of a project that gets ruined by too many people working on it, all with their own goals, and the book tells you exactly why this happened. C++ now is a hideously complex monstrosity that is popular because it is all things to all people, not because it is a good language.
Anyway, if you disagree with me, have a look at the book. It is a testament to Stroustrup's objectivity that I came to the conclusion I did, and that you may come to the exact opposite conclusion as me after reading it.
Can someone explain why anyone could possibly want Windows on a scientific computing cluster? What does Windows offer that Linux doesn't?
Much of my work involves running molecular dynamics simulations. By HPC standards these are tiny calculations (in my case, usually 32 CPUs at a time). All science HPC software I'm aware of is Unix-oriented, and everything runs on Linux. At my institution we have an OS X cluster and we are in the process of purchasing a Linux cluster. We didn't even consider Windows - given the difficulties we've experienced administering Windows on the desktop, a Windows cluster just seems like an expensive exercise in frustration.
For example, when you visit someone you obviously are going to be with them, and they are going to see what you do.
Would it be OK if I had this running, and therefore could listen to your conversations? I wouldn't like it, but I would have no basis to complain about it. Anything sent over the wire or air in plaintext is visible to everyone, so I can't possibly expect privacy through an open access point. You aren't the only one I have to trust not to sniff my plaintext data - I also have to trust everyone who has access to each router at every hop between me and the destination.
Being a nerd I knew this already, but getting the general public to understand this concept is a bit harder...
But then again, I'm not a lawyer.
I, too, have a Gmail account and get very little spam. Hooray for Google.
Who cares whether they are manipulative, dishonest, or unaccountable? Let it stay accessible and let the public decide for themselves.
I really dislike Fox News, but even though they are a poor excuse for a news organization, they shouldn't be shut down. We all have the same right to free speech.