I don't know what the specifics are for other countries, but here in Australia, a law is being drafted that any research that is done that in some way receives government money (grant, paper, etc), must retain all data (both non-simulated and simulated) *forever* and be publicly available. The time between publication and release of data is what is taking time to draft.
It will be a little while yet before it becomes mandatory. Clearly we are talking about major infrastructure as well as significant cultural change.
Fair point. But what about csv files that are > 2TB each in size that describe integers, floats and unsigned shorts? XML will increase this size by what factor? An order of magnitude?
There is also The Visualization Handbook which covers all the different classifications of data and the appropriate techniques to visualise. Lots of theory and algorithms under the hood.
I think here in Aus, people have not been looking down at the trades for a few years now.
I earn AU$78000 now which I'm quite happy about, but when I go to a BBQ with mates, a couple of them are carpenters and plumbers. The carpenters earn a little over twice what I do and the plumber is almost three times. They are also quite happy with their jobs. Sure, they had to get through some shit when they were at the bottom (who didn't?) but they are probably the most relaxed people I know.
When I gutted and built a new bathroom during my last holidays, I remember thinking what would have happened if my parents hadn't forced me to go to university (and do Mathematics and Computer Science degrees) and had done a trade instead. I enjoyed that holiday.
And apparently, chicks dig 'real men'. You know, scars and good with the hands and all that.
When I have kids, I think I'll just let them develop whatever skills they want to. Sure I'll make sure they perform as school (and that they finish it) but as for career selection, that is can be their choice.
Don't just look at China. Getting tax-payer funded data out of any government in Australia is impossible. Each government department just protects their own interest.
I applaud this effort as well. Credit where credit is due.
Except, culturally, things work a little differently here. Or, at least they used to.
When you do something wrong, you own up to it. You admit your mistake. To deny you did something that you did actually do is seen as cowardice.
It goes back to the playground rules when you were at school.
But the times, they are a changing. More often, people are choosing to get a lawyer (goon) and hide behind them and make up bullshit lines instead of owning up and admitting what they did was wrong.
The innocent until proven guilty line only works for me when the accused is actually innocent. The guy in the story was just being a non coward.
Yup, right tool for the job. What you describe though is not what I class as HPC. We need *maximum* floating point throughput. That includes extremely efficient memory and cache management. I know I'm in the ballpark when the remote power monitors to the supercomputers show significant increase in consumption.
I check each new java release. Still nowhere near an Intel compiler on Intel hardware.
I have to rewrite code that contains dynamic memory structures in C++ (stl) and remove those bastards. They are simply not good enough... Replacing a stl::vector with the standard array will achieve a 500x speed improvement on a simple vector multiply (E54xx, 11.0.074). Some will argue that it is not needed. I beg to differ. What is better? Assuming zero network lag and instant communication, one processor can achieve what 500 do in exactly the same time. Which is more cost effective?
I like Java, don't get me wrong. I reach for it whenever I need cross network/client communication among many other things. It is extremely good at what *I* think it was designed for. At this very moment I'm punching out some Java code. But it's for an app for an entirely different purpose. A much more suitable one.
How I wish we could get something like this in Australia. You have more luck getting blood from a stone than getting data out of the federal and state agencies here. They protect their own interests and taxpayer funded data acquisition.
I generate a lot of visualisations on raw data and they all love it. But does that mean they give me the newer or full datasets? Fuck no. And you can see them making up excuses in meetings as the dodge it.
I hope that one day our government learns from what the US is doing. I love crunching large volumes of data, manipulating and displaying in what is hopefully providing new insight to those who can formulate the patterns. Currently we live in the dark ages.
I bash Java, but when I need to knock up things quickly, it is the first language I turn to. Usual old "right tool for the job" argument. A direct boot JVM would enable quite a few more classes of jobs.
Actually, now I think about it, wasn't there this?. Have you ever tried it? I haven't. Although I don't think that was the one I was thinking of.
I realise that java code looks elegant. That things are also easy to program. But have you ever examined the callstack? Do yourself a favour sometime and see why Java is slow as dogs balls, memory hungry and anything but elegant.
Part of what I do is manage the applications running on a couple of HPC machines and when things are really busy, guess which jobs get pushed aside for resource hogging?
Is that for single or double precision work? Which Xeon exactly? Which compiler? How was the code written for the compiler? Which compiler flags?
Although I don't dispute your claims, writing to get max performance out the newer xeons is *hard* and you need to be very careful. The 256 bit wide registers on the 54xx can be extremely handy for codes written the right way.
I currently have a client that needs to run a lot of this and so far, I have the single cpu version running 10x faster than the parallel version running on 8 cores (single node). Only simple changes thus far although there is a particularly nasty data structure in there that is next for the chopping block.
I mean, you see a half naked pic of your wife and you also know that she goes off in the sack. Jolie, on the other hand, looks pretty but you just have no idea, or memory, of how much fun she is.
My wife also gets me excited. And really, that's no surprise. I know what she really likes and I really, really like it too. Everytime I think about it, it gives me a smile.
Results are no shock. Jolie is just eye-candy. That it.
At work we were going through our storeroom and came across three SGI O2's, 4 Octanes, 2 indigos and an Onyx (Ok, it's a big storeroom). The boss wanted them to be thrown out. I managed to convince him not too. That these are the sort of machines that may start to increase in price.
I used the Onyx 6 years ago when I first started as a remote X display (or was it an Indigo)? But that's about it. All of them still work.
Given that you're a UQ follower, I would have thought you'd be for giving the cash to the scramjet (Hypersonics) lab. Those guys can make little money go a very long way. It would certainly be deserved.
Such a shame the original dude had a stroke. His understudy shows promise though.
Kinda like tuning your codes to run really fast, then the poor monkeys in the basement send you a IR photo of some power couplings and ask what you're doing. I've always been going for flops but it has been getting a little more difficult now that I bring in power consumption into the optimisation cycle.
This starts to go into that really gray area called motivation. Don't try and guess what motivates a single person. It's impossible to guess.
Me? After examining many different things, I hope that I have an idea that might change the way things are done and will want to work on it day and night. Not for profit, but just for kicks and being able to point back at it when I'm rocking in my cane chair waiting to move on.
I doubt the OP is going to find any answers here. Actually, I bet he/she goes Masters because they'll get pushed into it by the profs.
Yeah, I realise that the older you get, the lower the learning capability. But the older you get the wiser you get. I'm planning to go the more philosophical approach when I get older.
I'm 31 and probably have another 34 working years ahead of me. That is a shitload. And the thing is; I don't even know what I want to know everything about yet. There are a *lot* of things that I want to know everything about. (I believe many younger people get pressured into doing post-graduate study early... not because that shine more than others... but because schools have a quota to meet)
See, I have a great job right now. I have to learn new stuff all the time and do lots of cool things. Why would I do extra study now? It would only slow things down (you know, that thesis won't write itself). But later on, probably 38-45 period, I'll start looking into one thing in depth. It may finish at masters, it may finish with PhD. Who knows? But I think my current path feels right for me.
So far I like: Nonlinear algebra Parallel rendering GPU -vs- FPGA -vs- CPU Haptics Filesystems Data Management Rendering algorithms
Are you going to pick a topic at age 20? I can't pick one yet.
Work gave me a Macbook pro, but I still use my Thinkpad X60 cause Fedora on that thing is pretty much a perfect mobile computing solution for me.
The battery is 3 years old and I still get 5 hours constant use out of it (provided no big cpu stuff) on wireless. Being able to turn off things I don't need (usb, etc) makes this possible. Do that on windows or osx.
Slashdot Mirror
I don't know what the specifics are for other countries, but here in Australia, a law is being drafted that any research that is done that in some way receives government money (grant, paper, etc), must retain all data (both non-simulated and simulated) *forever* and be publicly available. The time between publication and release of data is what is taking time to draft.
It will be a little while yet before it becomes mandatory. Clearly we are talking about major infrastructure as well as significant cultural change.
I was more of the opinion that the patents were global. Playing them in space is not a breach.
I say this with my extensive knowledge of law. Yeah. It's the vibe.
Fair point. But what about csv files that are > 2TB each in size that describe integers, floats and unsigned shorts? XML will increase this size by what factor? An order of magnitude?
For small files ... absolutely.
There is also The Visualization Handbook which covers all the different classifications of data and the appropriate techniques to visualise. Lots of theory and algorithms under the hood.
I think here in Aus, people have not been looking down at the trades for a few years now.
I earn AU$78000 now which I'm quite happy about, but when I go to a BBQ with mates, a couple of them are carpenters and plumbers. The carpenters earn a little over twice what I do and the plumber is almost three times. They are also quite happy with their jobs. Sure, they had to get through some shit when they were at the bottom (who didn't?) but they are probably the most relaxed people I know.
When I gutted and built a new bathroom during my last holidays, I remember thinking what would have happened if my parents hadn't forced me to go to university (and do Mathematics and Computer Science degrees) and had done a trade instead. I enjoyed that holiday.
And apparently, chicks dig 'real men'. You know, scars and good with the hands and all that.
When I have kids, I think I'll just let them develop whatever skills they want to. Sure I'll make sure they perform as school (and that they finish it) but as for career selection, that is can be their choice.
Don't just look at China. Getting tax-payer funded data out of any government in Australia is impossible. Each government department just protects their own interest.
I applaud this effort as well. Credit where credit is due.
Except, culturally, things work a little differently here. Or, at least they used to.
When you do something wrong, you own up to it. You admit your mistake. To deny you did something that you did actually do is seen as cowardice.
It goes back to the playground rules when you were at school.
But the times, they are a changing. More often, people are choosing to get a lawyer (goon) and hide behind them and make up bullshit lines instead of owning up and admitting what they did was wrong.
The innocent until proven guilty line only works for me when the accused is actually innocent. The guy in the story was just being a non coward.
Yup, right tool for the job. What you describe though is not what I class as HPC. We need *maximum* floating point throughput. That includes extremely efficient memory and cache management. I know I'm in the ballpark when the remote power monitors to the supercomputers show significant increase in consumption.
I check each new java release. Still nowhere near an Intel compiler on Intel hardware.
I have to rewrite code that contains dynamic memory structures in C++ (stl) and remove those bastards. They are simply not good enough ... Replacing a stl::vector with the standard array will achieve a 500x speed improvement on a simple vector multiply (E54xx, 11.0.074). Some will argue that it is not needed. I beg to differ. What is better? Assuming zero network lag and instant communication, one processor can achieve what 500 do in exactly the same time. Which is more cost effective?
I like Java, don't get me wrong. I reach for it whenever I need cross network/client communication among many other things. It is extremely good at what *I* think it was designed for. At this very moment I'm punching out some Java code. But it's for an app for an entirely different purpose. A much more suitable one.
You sneaky little bastard. Keep up the good work :)
How I wish we could get something like this in Australia. You have more luck getting blood from a stone than getting data out of the federal and state agencies here. They protect their own interests and taxpayer funded data acquisition.
I generate a lot of visualisations on raw data and they all love it. But does that mean they give me the newer or full datasets? Fuck no. And you can see them making up excuses in meetings as the dodge it.
I hope that one day our government learns from what the US is doing. I love crunching large volumes of data, manipulating and displaying in what is hopefully providing new insight to those who can formulate the patterns. Currently we live in the dark ages.
That would be something I'd like to see.
I bash Java, but when I need to knock up things quickly, it is the first language I turn to. Usual old "right tool for the job" argument. A direct boot JVM would enable quite a few more classes of jobs.
Actually, now I think about it, wasn't there this?. Have you ever tried it? I haven't. Although I don't think that was the one I was thinking of.
Just firing random thoughts.
Go elegant.
I realise that java code looks elegant. That things are also easy to program. But have you ever examined the callstack? Do yourself a favour sometime and see why Java is slow as dogs balls, memory hungry and anything but elegant.
Part of what I do is manage the applications running on a couple of HPC machines and when things are really busy, guess which jobs get pushed aside for resource hogging?
Here.
Is that for single or double precision work? Which Xeon exactly? Which compiler? How was the code written for the compiler? Which compiler flags?
Although I don't dispute your claims, writing to get max performance out the newer xeons is *hard* and you need to be very careful. The 256 bit wide registers on the 54xx can be extremely handy for codes written the right way.
I currently have a client that needs to run a lot of this and so far, I have the single cpu version running 10x faster than the parallel version running on 8 cores (single node). Only simple changes thus far although there is a particularly nasty data structure in there that is next for the chopping block.
Just saying.
I mean, you see a half naked pic of your wife and you also know that she goes off in the sack. Jolie, on the other hand, looks pretty but you just have no idea, or memory, of how much fun she is.
My wife also gets me excited. And really, that's no surprise. I know what she really likes and I really, really like it too. Everytime I think about it, it gives me a smile.
Results are no shock. Jolie is just eye-candy. That it.
At work we were going through our storeroom and came across three SGI O2's, 4 Octanes, 2 indigos and an Onyx (Ok, it's a big storeroom). The boss wanted them to be thrown out. I managed to convince him not too. That these are the sort of machines that may start to increase in price.
I used the Onyx 6 years ago when I first started as a remote X display (or was it an Indigo)? But that's about it. All of them still work.
That would have to be very quick ... I remember adapting to WinME
What do you mean?
Given that you're a UQ follower, I would have thought you'd be for giving the cash to the scramjet (Hypersonics) lab. Those guys can make little money go a very long way. It would certainly be deserved.
Such a shame the original dude had a stroke. His understudy shows promise though.
This one will definitely cause autism.
Quick! Develop it but don't dare use it!
Sorry, that new performance measurement should be
flopspa
Good that there is thought to this.
Kinda like tuning your codes to run really fast, then the poor monkeys in the basement send you a IR photo of some power couplings and ask what you're doing. I've always been going for flops but it has been getting a little more difficult now that I bring in power consumption into the optimisation cycle.
New measurement for the top500
flopspw ?
This starts to go into that really gray area called motivation. Don't try and guess what motivates a single person. It's impossible to guess.
Me? After examining many different things, I hope that I have an idea that might change the way things are done and will want to work on it day and night. Not for profit, but just for kicks and being able to point back at it when I'm rocking in my cane chair waiting to move on.
I doubt the OP is going to find any answers here. Actually, I bet he/she goes Masters because they'll get pushed into it by the profs.
Yeah, I realise that the older you get, the lower the learning capability. But the older you get the wiser you get. I'm planning to go the more philosophical approach when I get older.
I'm 31 and probably have another 34 working years ahead of me. That is a shitload. And the thing is; I don't even know what I want to know everything about yet. There are a *lot* of things that I want to know everything about. (I believe many younger people get pressured into doing post-graduate study early ... not because that shine more than others ... but because schools have a quota to meet)
See, I have a great job right now. I have to learn new stuff all the time and do lots of cool things. Why would I do extra study now? It would only slow things down (you know, that thesis won't write itself). But later on, probably 38-45 period, I'll start looking into one thing in depth. It may finish at masters, it may finish with PhD. Who knows? But I think my current path feels right for me.
So far I like:
Nonlinear algebra
Parallel rendering
GPU -vs- FPGA -vs- CPU
Haptics
Filesystems
Data Management
Rendering algorithms
Are you going to pick a topic at age 20? I can't pick one yet.
Work gave me a Macbook pro, but I still use my Thinkpad X60 cause Fedora on that thing is pretty much a perfect mobile computing solution for me.
The battery is 3 years old and I still get 5 hours constant use out of it (provided no big cpu stuff) on wireless. Being able to turn off things I don't need (usb, etc) makes this possible. Do that on windows or osx.