True, but perhaps the compilers need to change with the different CPUs. Different types of for loops that mean different things? Or use the standard loop but don't program with pointers so the compilers can see what can be done to parallelise the loops for you.
I watched an in-depth Australian report on the candidates for the presidential election where MIKE GREEN, FORMER BUSH FOREIGN POLICY ADVISER was talking about Senator John McCain and his relationship with the Republican party. He said, possibly quite stupidly :
... Senator McCain but he is probably the most
independent Republican in the Senate.
How is this a good thing? Or is it my lack of knowledge of American politics? In Australia, if a member of a party distanced themselves by saying this, they would not last too long in that party. If you don't have the support of your party there'll be big problems from day one in being able to form a government, let alone being able to run that government effectively.
Now that Bill Gates is retired from Microsoft, the editors should get with the times and lose that dated, painfully unfunny logo they use for Microsoft.
The first batch of CF lights I got were that cold white light that I used to hate, but after having them, I've gotten used to them. I recently bought another one that was 'warm' and I gotta say, is a very comfortable light. As good as the old type of bulbs. In fact, when you put the covers over them, you can't tell the difference between it and ye olde type.
Thankfully I don't live in a cold climate. I think I can understand needing warmth anyway that you can get it.
Holy crap! I think people like yourself overload the power grid. Why do you need so many lights on?
Mate I think you should put just a few cycles aside to consider replacing some light bulbs with the energy efficient ones. Not all at once, just one every now and then. Whenever your spare change jar hits enough. Start with the most frequently used lights. They are actually pretty good. I was extremely skeptical at first until our government bought everybody an energy efficient bulb to dismiss the skepticism.
I've rearranged all the electrical goods' power so I can turn off, at the switch, the non-essentials when I go to bed. It's a mentality change. My power bill dropped dramatically.
Ok, so I didn't know that this posting was actually you, and by "Most, if not all people who work here" you were referring to your work colleagues and not/.. I'm slow but I get there in the end.
Wanna fill your spare cycles? Go to your Mathematics department and have a chat to the post-graduate students in non-linear algebra... they are a really good start. Tell them you've got a machine available. Next, go see your physicists. The particle physicists and medical guys will definitely keep a machine like that busy with open source software. Another group that can fill cycles are the document searching/classification people in the IT faculty. Are there any medical institutes nearby or affiliated with your university? Every university has an Engineering faculty. How do they get to do large scale simulations? There are some nice open source packages for that as well.
If you've got Matlab licenses, get to your university's Matlab mail list and advertise your machine for batch submission. People can be slow to adopt this, but once they've done it a few times, they won't do it any other way.
Make sure you're ready for the extra accounts... that you can say for every account, how many compute seconds were used. Ask people that if they publish a paper or present at a conference, they acknowledge your group. This makes great ammunition when needing upgrades/replacements.
You are missing the point. Sorry, I thought we were talking about high performance computing on a 500+ core system.
On systems like this, with vendor support and appropriate compilers, you're looking at about US$0.50 per compute hour for the lifetime of the machine - and that's underestimating everything, not including people. You are willing to accept a 30% performance hit? On my simple 50.9 compute year example, your 30% becomes US$67,000 and that is for a small job. In some circles, that is an unacceptable loss.
Oh, and thanks for the Hadoop link. Looks interesting but seems a little too young. I'll never use something that does a performance benchmark on a cluster that tells me how many nodes but absolutely nothing in regards to the interconnect, the processor architecture, the arrangement of the processors, etc, etc. Without that information, it means nothing as I can't reproduce or replicate using my libraries in order to evaluate whether to use it or not. Sure I could invest time, but until they put up specs that are more meaningful, I have other things to investigate.
That's exactly how it works at my University. My job is to optimise the larger batches of code and perform parallelism where appropriate. If they get over the 'mine' mentality they could do some real good.
Because the Java codes I have to optimise to run on our cluster (or SMP machine) still run like lame, wet dogs. I'm sorry, but Java's performance is really, really shitty and I'm considering removing Java and Javac from the systems altogether. When you compare another object oriented language, like C++, at least with that you can write your codes to make use of multiple instructions per cycle... and then with a good compiler and appropriate hardware, you can vectorize the instructions themselves. When you have codes that for a single processor, would run for about 50.9 years (an example I pulled from just today) those 'simple' little things go a long way. Now where can you tweak that sort of thing in Java?... Oh, that's right, YOU CAN'T.
But that's object oriented languages (which is getting better since I'm learning how to make myArray[i] run twice as fast), I prefer to use C and fortran. Those babies are low memory and totally rock. You don't spend all your time making method calls, which chew cycles for each call (let alone being in large nested loops).
I don't mind punching out Java code; it's extremely good for some tasks, but it is not a language I'd use on a high performance system. Just in my experience. I could be wrong and would love to be proven wrong.
I recently went home to visit my Dad and found that he was cleaning out the house, ridding it of junk because his partner wanted it cleaned (quite justified). I was having a good riffle through the stuff they were going to throw out when I came across my two, 50 litre buckets of Lego. I very quickly put these in my car to take back to my house for 'archival purposes'.
I'll have kids fairly soon that will be of an age to start playing with these things. I'll be encouraging the play of Lego more than watching TV or playing computer/console games. Sure, computer games can teach you problem solving techniques, but so does Lego, as well as having tactile response.
Your entire rant has been debunked by "machines running Windows CE ver 2020". Didn't you know that it's "the year of the Linux desktop" every year until then.
Too many bad assumptions. Yes, the media companies hire some smart people, but there is a much, much larger community of far more intelligent people punching out code to circumvent these restrictions.
Not the post but when I look at every message you've ever posted.
Out of 132 posts, dating back to 17th January 2005, every message has been modded -1. All except for your first one which was 0, but at least it was flamebait.
That's quite impressive. At least you're consistent.
I too run codes on top500 machines. Although, these days I run on lesser machines that are more than capable.
I have to comment on one of your perfectly valid points:
same for redundancy. If part of a supercomputer goes down, anything running on it is lost. Bad news for a scientist running on it, he needs to restart his work. Impossible news for a Wall Street data processing machine. If you're running codes on these machines that are using a large amount of compute power, you *need* to be check-pointing your code. You're an idiot if you think that the larger the machine, the less likely the downtime. Machines fail, the more machines, the greater the possibility of a failure of some type.
I first noticed that something was a little different when I woke up one morning (approx 7.30am). It wasn't painful, just something was happening that was different. It got more painful during the day, at 4pm I went to hospital. I didn't have private health insurance, so I had to wait about 20 hours before the surgery actually happened.
It happened *very* quickly. I used to wear underwear that was 20% or so polyester. The infection is internal, you don't actually see it.
Slashdot Mirror
There, I finally upgraded to FF3 on my FC8 box. You happy now?
But if your child read it your family contribution is -1
True, but perhaps the compilers need to change with the different CPUs. Different types of for loops that mean different things? Or use the standard loop but don't program with pointers so the compilers can see what can be done to parallelise the loops for you.
It's uncanny the parallels of your statements and Australian politics. Not the $300,000,000,000.00, that'd buy the country.
... Senator McCain but he is probably the most independent Republican in the Senate.
How is this a good thing? Or is it my lack of knowledge of American politics? In Australia, if a member of a party distanced themselves by saying this, they would not last too long in that party. If you don't have the support of your party there'll be big problems from day one in being able to form a government, let alone being able to run that government effectively.
Source: http://www.abc.net.au/insiders/
Now that Bill Gates is retired from Microsoft, the editors should get with the times and lose that dated, painfully unfunny logo they use for Microsoft.
It's legacy code.
Agreed .... very, very old news. I saw one of these things working about 8 years ago.
Cool ... does that mean I don't have to hit the breaks for jaywalkers?
The first batch of CF lights I got were that cold white light that I used to hate, but after having them, I've gotten used to them. I recently bought another one that was 'warm' and I gotta say, is a very comfortable light. As good as the old type of bulbs. In fact, when you put the covers over them, you can't tell the difference between it and ye olde type.
Thankfully I don't live in a cold climate. I think I can understand needing warmth anyway that you can get it.
Holy crap! I think people like yourself overload the power grid. Why do you need so many lights on?
Mate I think you should put just a few cycles aside to consider replacing some light bulbs with the energy efficient ones. Not all at once, just one every now and then. Whenever your spare change jar hits enough. Start with the most frequently used lights. They are actually pretty good. I was extremely skeptical at first until our government bought everybody an energy efficient bulb to dismiss the skepticism.
I've rearranged all the electrical goods' power so I can turn off, at the switch, the non-essentials when I go to bed. It's a mentality change. My power bill dropped dramatically.
Ok, so I didn't know that this posting was actually you, and by "Most, if not all people who work here" you were referring to your work colleagues and not /.. I'm slow but I get there in the end.
... they are a really good start. Tell them you've got a machine available. Next, go see your physicists. The particle physicists and medical guys will definitely keep a machine like that busy with open source software. Another group that can fill cycles are the document searching/classification people in the IT faculty. Are there any medical institutes nearby or affiliated with your university? Every university has an Engineering faculty. How do they get to do large scale simulations? There are some nice open source packages for that as well.
... that you can say for every account, how many compute seconds were used. Ask people that if they publish a paper or present at a conference, they acknowledge your group. This makes great ammunition when needing upgrades/replacements.
... share it.
Wanna fill your spare cycles? Go to your Mathematics department and have a chat to the post-graduate students in non-linear algebra
If you've got Matlab licenses, get to your university's Matlab mail list and advertise your machine for batch submission. People can be slow to adopt this, but once they've done it a few times, they won't do it any other way.
Make sure you're ready for the extra accounts
Don't keep the machine to yourselves
Sorry, I thought we were talking about high performance computing on a 500+ core system.
On systems like this, with vendor support and appropriate compilers, you're looking at about US$0.50 per compute hour for the lifetime of the machine - and that's underestimating everything, not including people. You are willing to accept a 30% performance hit? On my simple 50.9 compute year example, your 30% becomes US$67,000 and that is for a small job. In some circles, that is an unacceptable loss.
Oh, and thanks for the Hadoop link. Looks interesting but seems a little too young. I'll never use something that does a performance benchmark on a cluster that tells me how many nodes but absolutely nothing in regards to the interconnect, the processor architecture, the arrangement of the processors, etc, etc. Without that information, it means nothing as I can't reproduce or replicate using my libraries in order to evaluate whether to use it or not. Sure I could invest time, but until they put up specs that are more meaningful, I have other things to investigate.
Ahhh, how I miss Irix 6.5
Forgive me while I give a moment of silence.
Really? I though it was mathematics.
Obligatory http://xkcd.com/435/
That's exactly how it works at my University. My job is to optimise the larger batches of code and perform parallelism where appropriate. If they get over the 'mine' mentality they could do some real good.
Because the Java codes I have to optimise to run on our cluster (or SMP machine) still run like lame, wet dogs. I'm sorry, but Java's performance is really, really shitty and I'm considering removing Java and Javac from the systems altogether. When you compare another object oriented language, like C++, at least with that you can write your codes to make use of multiple instructions per cycle ... and then with a good compiler and appropriate hardware, you can vectorize the instructions themselves. When you have codes that for a single processor, would run for about 50.9 years (an example I pulled from just today) those 'simple' little things go a long way. Now where can you tweak that sort of thing in Java? ... Oh, that's right, YOU CAN'T.
But that's object oriented languages (which is getting better since I'm learning how to make myArray[i] run twice as fast), I prefer to use C and fortran. Those babies are low memory and totally rock. You don't spend all your time making method calls, which chew cycles for each call (let alone being in large nested loops).
I don't mind punching out Java code; it's extremely good for some tasks, but it is not a language I'd use on a high performance system. Just in my experience. I could be wrong and would love to be proven wrong.
I recently went home to visit my Dad and found that he was cleaning out the house, ridding it of junk because his partner wanted it cleaned (quite justified). I was having a good riffle through the stuff they were going to throw out when I came across my two, 50 litre buckets of Lego. I very quickly put these in my car to take back to my house for 'archival purposes'.
I'll have kids fairly soon that will be of an age to start playing with these things. I'll be encouraging the play of Lego more than watching TV or playing computer/console games. Sure, computer games can teach you problem solving techniques, but so does Lego, as well as having tactile response.
Your entire rant has been debunked by "machines running Windows CE ver 2020". Didn't you know that it's "the year of the Linux desktop" every year until then.
Too many bad assumptions. Yes, the media companies hire some smart people, but there is a much, much larger community of far more intelligent people punching out code to circumvent these restrictions.
Here's another way to test the quality of code ...
http://www.osnews.com/story/19266/WTFs_m
// Possibly a more efficient way of doing this ... cycle count?
.
.
.
.
coffee.add(sugar)
coffee.add(milk)
OutOfMemoryException
Not the post but when I look at every message you've ever posted.
Out of 132 posts, dating back to 17th January 2005, every message has been modded -1. All except for your first one which was 0, but at least it was flamebait.
That's quite impressive. At least you're consistent.
Jeeves? Escort this young man to the door. When you are done, warm my platinum throne will you.
I have to comment on one of your perfectly valid points: same for redundancy. If part of a supercomputer goes down, anything running on it is lost. Bad news for a scientist running on it, he needs to restart his work. Impossible news for a Wall Street data processing machine.
If you're running codes on these machines that are using a large amount of compute power, you *need* to be check-pointing your code. You're an idiot if you think that the larger the machine, the less likely the downtime. Machines fail, the more machines, the greater the possibility of a failure of some type.
Just keeping my great, great grandparents proud. History is a bitch to live with, you know.
Cheers for your concern.
I first noticed that something was a little different when I woke up one morning (approx 7.30am). It wasn't painful, just something was happening that was different. It got more painful during the day, at 4pm I went to hospital. I didn't have private health insurance, so I had to wait about 20 hours before the surgery actually happened.
It happened *very* quickly. I used to wear underwear that was 20% or so polyester. The infection is internal, you don't actually see it.