Seriously though, it is easy to take a good parallel code and murder it by going for a Beowulf option. I assume that your CFD code is a heavilly parallel beast. From what I have seen
from the links available here
one must be very careful about the relative importance of the node communication bandwidth and the node latency on the code in question.
In my opinion this decision boils down to the following:
How interdependent is the parallel code? If the code is trivially parallel, as many of mine are, then a cluster of the cheapest ix86 machines running Linux is an excellent bet.
Are you certain that the code will work on the platforms in question? Perhaps invest a couple of kiloquid (thousand pounds) or so in a few ix86 machines for a feasibility study, if it doesn't then you have cut your losses and have a few machines to play Quake on;-). For the ultra cheap option, if your Prof is unhappy about gambing any dosh (money) on buying new stuff, stick Linux on the NT machines and try those.
How far through your PhD are you? Quite seriously, if you're anything like me you may find that you end up spending most of your time tweaking computers rather than doing physics. Building your own supercomputer could easilly suck your entire PhD down the time toilet.
I expect you have done this already but ask the people at Fluent Inc. what experience they have on parallelizing their codes.
Finally, I would say while the support you may get from commercial *NIX vendors is generally good it can also be very expensive. While this is not necessarilly a problem in a commercial datacentre environment, in academia it can be crippling. If you and your co-workers are willing to expend time and effort into trying to get the code working on a Beowulf style cluster it could well be a good long term investment.
Good luck!
P.S. When I checked this post, www.beowulf.org was down so you may want to check the google cache.
I'm not sure people really grasp the point here. This "war on terrorism" is, from the perspective of the people, being waged in a conventional way.
It is however being waged against an unconventional enemy.
The problem is that the "enemy" in this case is faceles, it's not/bin/laden, it's not the Taleban it's the thousands of people distributed across the world working in their terrorist cells and capable of acting autonomously. Surely you're aware that conventional warfare against an enemy that can practice Gurillea warfare is next to pointless, that lesson I'm sure your military learned from the Vietnam conflict.
Sure, you can knock out the Taleban, string up/bin/laden and leave Afganistan to rot, waiting for the next bunch of totalitarian madmen to take over, but it won't solve the long term problem. That problem is that people with the will can kill thousands, they are living in your towns, not in Afganistan, they speak your language, they look the same as you and they act the same as you.
I hasten to add that infiltration, inteligence gathering and the planting of agents will almost certainly be more effective in this on going conflict. You won't see it, you won't here about it, it won't be on TV, but it will be there and it will be more efficient at saving lives than a cruise missile.
Ultimatly of course the only way this kind of war can really be won in the long run is by unleashing the worlds most potent, deployable and self-reproducing weapon,
education....
....if someone can be persuaded to hate and to kill they can also be persuaded to love and to cherish
I've managed to kill loads of semiconductor stuff by being charged, including memory, power transistors, graphics cards etc..
It is quite easy to build up a static charge so that your body is over >200V. I've even measured myself as reaching ~3KV, though I was trying quite hard...
Now I ALWAYS make sure I'm earthed before touching ports, or the inside of machines. If there is any inadequate earthing of an external component this could allow the ports to get zapped.
I wouldn't be supprised if the plantifs win. Since we now live in a society which has "Caution, contents hot" written on coffee cups to avoid litigation, equipment would then come with "Caution, do not apply high tension electrical sources to port connectors":)
On the other hand this kind of thing is always written in manuals for motherboards and PCs.. I guess they will have their day in court...
There is of course one simple solution to what may be an increasing problem. Standardised optical interconnects or optical isolation of ports...
A simple IR port, which I understand a lot of machines come with these days, means no physical connection and therefore no discharge possibilty..
Perhaps in light of this your politzi will try to prosecute the people that recorded them "interrogating" Rodney King........the law is an odd planet.
Hmm.... Cameras-cameras everywhere....
This remindes me of a missed opportunity.
A friend of mine is making a film and they were practicing some of there scenes in Hyde Park which involved the use of BB guns (toys that fire plastic pellets).
Now as the more astute of you may be aware, hand guns were made illegal over here a while ago and the police (rightly in my opinion) take a pretty hard line with threats to the public.
So when someone phoned them and said "There's a man with orange hair shooting people in Hyde Park", what were they supposed to do?
Well what they did was to send in a heavilly armed anti-terrorist-style unit, complete with helicopter to "take-down" these people who, by that time, had begun to play frisbee and were completely unaware that within the space of 10 seconds they would find there faces in the ground with guns pointed at them...
If only they'd got that on film! It could have been great as part of the story line!
Err.. How did you work this one out? It makes no sense at all.
How can any single nuclear reaction (by which I mean just one fission or fusion) _EVER_ release over a _kilojoule_ of energy? I think you have make a mistake between KeV (kiloelectronvolts) and KJ (kilojoules)... There's a seriously big difference!
How many of these scientists and engineers would be where they're at without computers?
Who do you think invented what we now call the computer in the first place? I think "computers" and "scientists and engineers" are interchangable in the above sentence.
In "field" envy of this sort it always appears that every one tries to take other people out of context;-)
-ed
Ok, so I can't see someone building there own semiconductor plant in the kitchen (at least not for the current state of the art chips).. BUT!!!
How about FPGAs??
Field Programmable Gate Arrays....
Basically these babies are programmable hardware! So instead of compiling code to run on a particular processor you "compile" the hardware!!
In a nutshell they work by having building blocks on the chip, such as individual gates or more complex thangs, that you can connect together in a similar way to writing to an EPROM.
Applications?
- Customisable co-processors for
- 3d Graphics
- Scientific computing...
Anything that's slow in s/w but too expensive to implement yet in dedicated h/w.
This would of course leave the actual chip manufaturing to the experts, but allow "chip compilers" to flourish in an open community. A reasonable starting comprimise?
Do you know what UK organisation did the evaluation of Win2K source?
Was it the Defense Evaluation Research Agency (DERA)?
(There also known as "dearer"........ because they are)
(b.t.w. "dear" = "expensive")
If it was them, I'd throw their evaluations out the window and run away....
Or perhaps it's a secret way of the UK regaining control of the US military;)
Heh heh....
Hmm if memory serves, our (still quite monopolistic) British Telecom (BT) was toying with the idea of fibre to the home, as was our government...
However, since they (BT) spent so long dragging their feet with unbundling the local loop ADSL I wouldn't bank on bits of glass ever poking through the skirting board.....
While they can make money feeding us incremental improvements why do anything else?
Slashdot Mirror
Seriously though, it is easy to take a good parallel code and murder it by going for a Beowulf option. I assume that your CFD code is a heavilly parallel beast. From what I have seen from the links available here one must be very careful about the relative importance of the node communication bandwidth and the node latency on the code in question.
In my opinion this decision boils down to the following:
How interdependent is the parallel code? If the code is trivially parallel, as many of mine are, then a cluster of the cheapest ix86 machines running Linux is an excellent bet.
Are you certain that the code will work on the platforms in question? Perhaps invest a couple of kiloquid (thousand pounds) or so in a few ix86 machines for a feasibility study, if it doesn't then you have cut your losses and have a few machines to play Quake on ;-). For the ultra cheap option, if your Prof is unhappy about gambing any dosh (money) on buying new stuff, stick Linux on the NT machines and try those.
How far through your PhD are you? Quite seriously, if you're anything like me you may find that you end up spending most of your time tweaking computers rather than doing physics. Building your own supercomputer could easilly suck your entire PhD down the time toilet.
I expect you have done this already but ask the people at Fluent Inc. what experience they have on parallelizing their codes.
Finally, I would say while the support you may get from commercial *NIX vendors is generally good it can also be very expensive. While this is not necessarilly a problem in a commercial datacentre environment, in academia it can be crippling. If you and your co-workers are willing to expend time and effort into trying to get the code working on a Beowulf style cluster it could well be a good long term investment.
Good luck!
P.S. When I checked this post, www.beowulf.org was down so you may want to check the google cache.
-edIt is however being waged against an unconventional enemy.
The problem is that the "enemy" in this case is faceles, it's not
Sure, you can knock out the Taleban, string up
I hasten to add that infiltration, inteligence gathering and the planting of agents will almost certainly be more effective in this on going conflict. You won't see it, you won't here about it, it won't be on TV, but it will be there and it will be more efficient at saving lives than a cruise missile.
Ultimatly of course the only way this kind of war can really be won in the long run is by unleashing the worlds most potent, deployable and self-reproducing weapon,
education....
....if someone can be persuaded to hate and to kill they can also be persuaded to love and to cherish
Damn, I sound like a hippy!
To the point where I get scared of metal doors in certain rooms when the weather is dry because I used to get very painful electrostatic discharges.
I guess it could just be my electric personality
groan....
Or more likely a dangerous polyester/viscose/trainers interaction
The upshot of it is that it's always fairly rare, but when it happens it's expensive and painful, so why take the risk?
I've managed to kill loads of semiconductor stuff by being charged, including memory, power transistors, graphics cards etc..
It is quite easy to build up a static charge so that your body is over >200V. I've even measured myself as reaching ~3KV, though I was trying quite hard...
Now I ALWAYS make sure I'm earthed before touching ports, or the inside of machines. If there is any inadequate earthing of an external component this could allow the ports to get zapped.
I wouldn't be supprised if the plantifs win. Since we now live in a society which has "Caution, contents hot" written on coffee cups to avoid litigation, equipment would then come with "Caution, do not apply high tension electrical sources to port connectors" :)
On the other hand this kind of thing is always written in manuals for motherboards and PCs.. I guess they will have their day in court...
There is of course one simple solution to what may be an increasing problem. Standardised optical interconnects or optical isolation of ports...
A simple IR port, which I understand a lot of machines come with these days, means no physical connection and therefore no discharge possibilty..
Unless of course you get hit by lightning.... :-)
This remindes me of a missed opportunity.
A friend of mine is making a film and they were practicing some of there scenes in Hyde Park which involved the use of BB guns (toys that fire plastic pellets).
Now as the more astute of you may be aware, hand guns were made illegal over here a while ago and the police (rightly in my opinion) take a pretty hard line with threats to the public.
So when someone phoned them and said "There's a man with orange hair shooting people in Hyde Park", what were they supposed to do?
Well what they did was to send in a heavilly armed anti-terrorist-style unit, complete with helicopter to "take-down" these people who, by that time, had begun to play frisbee and were completely unaware that within the space of 10 seconds they would find there faces in the ground with guns pointed at them...
If only they'd got that on film! It could have been great as part of the story line!
Err.. How did you work this one out? It makes no sense at all.
How can any single nuclear reaction (by which I mean just one fission or fusion) _EVER_ release over a _kilojoule_ of energy? I think you have make a mistake between KeV (kiloelectronvolts) and KJ (kilojoules)... There's a seriously big difference!
Who do you think invented what we now call the computer in the first place? I think "computers" and "scientists and engineers" are interchangable in the above sentence. In "field" envy of this sort it always appears that every one tries to take other people out of context ;-)
-ed
How about FPGAs??
Field Programmable Gate Arrays....
Basically these babies are programmable hardware! So instead of compiling code to run on a particular processor you "compile" the hardware!!
In a nutshell they work by having building blocks on the chip, such as individual gates or more complex thangs, that you can connect together in a similar way to writing to an EPROM.
Applications?
- Customisable co-processors for
- 3d Graphics
- Scientific computing...
Anything that's slow in s/w but too expensive to implement yet in dedicated h/w.
This would of course leave the actual chip manufaturing to the experts, but allow "chip compilers" to flourish in an open community. A reasonable starting comprimise?
Do you know what UK organisation did the evaluation of Win2K source? Was it the Defense Evaluation Research Agency (DERA)? (There also known as "dearer".... .... because they are)
(b.t.w. "dear" = "expensive")
If it was them, I'd throw their evaluations out the window and run away....
Or perhaps it's a secret way of the UK regaining control of the US military ;)
Heh heh....
Hmm if memory serves, our (still quite monopolistic) British Telecom (BT) was toying with the idea of fibre to the home, as was our government...
However, since they (BT) spent so long dragging their feet with unbundling the local loop ADSL I wouldn't bank on bits of glass ever poking through the skirting board.....
While they can make money feeding us incremental improvements why do anything else?