I like the current design, thank you very much, although Unicode support and support for editing posts would be appreciated. Oh, and it'd be nice if the front page didn't spontaneously reload while I'm reading a summary.
I think what we have seen in the nearly 1000 comments above is that people don't like or want the new site, and that it'd be a bad idea to change it. When slashdot moved over to the current 2.0 design, there was a lot of moaning but not nearly as much as the rabid hate the 3.0 design is generating; personally I liked the 2.0 design when it was introduced.
Actually I also thought that was the case; but it certainly wasn't Asimov, which is why I posted this. Further research (if you can call it research) reveals the wikipedia page which states:
'Although Watson is well known for his alleged 1943 statement, "I think there is a world market for maybe five computers", there is scant evidence he made it.'
Without math, it's impossible to convey what you're trying to convey. The press is way too dumbed down already, and many times I've read science stories that are just plain misleading as they try to simplify the message.
Putting equations into news stories means that some people won't understand them, but most importantly it will encourage some of those people to investigate further, and learn how to read equations. If there's no math in the popular press in the first place, then there's no incentive for people to improve themselves.
I've thought about mounting a gigantic railgun on the eastern flank of Mount Kilimanjaro. You'd need to lengthen the barrel partly into the earth to give a longer run-up. Keep the barrel evacuated -- you can probably use a plasma window at the exit point to keep the atmosphere out. (although I'm not certain a plasma windows would work terribly well with a large aperture -- can anyone tell me?)
The tricky part is you'd need to accelerate your projectile at over 5000 gravities for the ~ 0.3 seconds it would take to get up to escape velocity. (these calcs are about right to within an order of magnitude). That's quite a lot of acceleration.
Thanks for your response GTG, I appreciate your answers.
Let me give you my perspective: I am a very experienced dev, quite happy in ASM, C, C++, and C#, and other languages (but I hate Java with a passion!)
I currently have a problem set which is extremely amenable to being solved with GPU development. I have had a look this evening at OpenCL, and I must say I was very impressed with how simple it appeared to me, coming from a C background. I was just about to plunge into the dev on OpenCL, when you [very inconsiderately!] disrupted my plans by suggesting that OpenMP might be a better plan.
Since you were so rude as to disrupt my plans (!), may I please ask for your suggestions for toolchain + ide for OpenMP dev? (I'm on Windows generally, but happy with Linux also...)
Having researched both, OpenCL is definitively better by far. Granted, CUDA has the native advantage, but that's not always going to be there, and I think most would agree that vendor tie-in is a Very Bad Thing (tm)
While you are completely correct in what you say, you are not answering the question. The question is, how do I get in to massively parallel GPU processing easily -- not, 'is my application suited to this?'. It's assumed that the OP might be an idiot, however he is asking a completely valid question, and you should be responding accordingly.
Actually, this post has prompted me to re-evaluate my methodologies. I have an enormous image-processing project to complete in short order. I was previously intending to use CPU only, but this post made me think, and go and research OpenCL. What I've found is that not only is it cross-platform, very fast (in many cases), but is also very easy to use (if you already know C very well, as I do).
The answer is, 'High level languages are not appropriate for this problem set'.
So, while I must agree that the original question of how to easily use GPGPU techniques from a purely managed environment such as C# is actually pretty inane, the result has been fruitful in that it highlights that if you really want GPGPU performance then you really do need to invest the effort required into actually learning the specific languages necessary. That's a relevant answer, in my book.
Something is seriously wrong with your maths here. Don't quit your day job.
Thruster = 0.000236 N Mass = 2000 f=ma ; a=f/m=0.000236/2000=1.18e-7 (this is a VERY low acceleration) s=ut+(1/2)at^2 (u=0 so we ignore that part) assume t=24000 hrs=2e9 seconds (to turnaround point) s=0.5 * 1.18e-7 * 2e9 * 2e9=2.360e9 metres By turn-around you would have travelled 2.36 million kilometres. 1 light year is 9.461e15 metres, so the space craft would have travelled about 2.5 millionths of a light year.
Bzzt you are both wrong. The net acceleration due to this test is zero, because the ions ejected out of the engine are halted by the test chamber. Net result is zero force.
Nonsense and bullshit. In general, the whole planet has acquiesced to the US convention of 1 billion == 1 000 000 000. Those who don't agree are doomed to be misunderstood and sidelined.
Slashdot Mirror
Monty Python: Search for the Holy Grail
http://www.youtube.com/watch?v=y2R3FvS4xr4
What is the airspeed velocity of a swallow?
I like the current design, thank you very much, although Unicode support and support for editing posts would be appreciated. Oh, and it'd be nice if the front page didn't spontaneously reload while I'm reading a summary.
I think what we have seen in the nearly 1000 comments above is that people don't like or want the new site, and that it'd be a bad idea to change it. When slashdot moved over to the current 2.0 design, there was a lot of moaning but not nearly as much as the rabid hate the 3.0 design is generating; personally I liked the 2.0 design when it was introduced.
Actually I also thought that was the case; but it certainly wasn't Asimov, which is why I posted this. Further research (if you can call it research) reveals the wikipedia page which states:
'Although Watson is well known for his alleged 1943 statement, "I think there is a world market for maybe five computers", there is scant evidence he made it.'
This technology is available today, but hasn't really caught on.
"I think there is a world market for maybe five computers." -- Thomas Watson, chairman of IBM, 1943
Oops forgot the link: http://www.terrafugia.com/tfx-vision
Thanks but I'll wait for the TF-X when it comes out. No runway for me!
Damn straight. Their disastrous reactions were a lot worse than no reaction at all.
Oh, yes, thank you for generating another Slashdot meme in space
FTFY
Without math, it's impossible to convey what you're trying to convey. The press is way too dumbed down already, and many times I've read science stories that are just plain misleading as they try to simplify the message.
Putting equations into news stories means that some people won't understand them, but most importantly it will encourage some of those people to investigate further, and learn how to read equations. If there's no math in the popular press in the first place, then there's no incentive for people to improve themselves.
I've thought about mounting a gigantic railgun on the eastern flank of Mount Kilimanjaro. You'd need to lengthen the barrel partly into the earth to give a longer run-up. Keep the barrel evacuated -- you can probably use a plasma window at the exit point to keep the atmosphere out. (although I'm not certain a plasma windows would work terribly well with a large aperture -- can anyone tell me?)
The tricky part is you'd need to accelerate your projectile at over 5000 gravities for the ~ 0.3 seconds it would take to get up to escape velocity. (these calcs are about right to within an order of magnitude). That's quite a lot of acceleration.
JTG* sigh ... I'm so inattentive ...
Thanks for your response GTG, I appreciate your answers.
Let me give you my perspective: I am a very experienced dev, quite happy in ASM, C, C++, and C#, and other languages (but I hate Java with a passion!)
I currently have a problem set which is extremely amenable to being solved with GPU development. I have had a look this evening at OpenCL, and I must say I was very impressed with how simple it appeared to me, coming from a C background. I was just about to plunge into the dev on OpenCL, when you [very inconsiderately!] disrupted my plans by suggesting that OpenMP might be a better plan.
Since you were so rude as to disrupt my plans (!), may I please ask for your suggestions for toolchain + ide for OpenMP dev? (I'm on Windows generally, but happy with Linux also ...)
All the best :)
c
Whew, I was just about to launch into a tirade on how wrong you were! As it stands, I'm going to be a lot less tired than I'd thought I'd be!
Deprecated, and quite a while back at that. Do not use.
John,
While I have to defer to your position as being a teacher of these things, I have to question what you say.
a) OpenCL was intended as an open access API to GPGPU techniques. Has something changed to channel people into vendor-specific approaches?
b) What advantages do OpenACC and OpenMP 4 offer over previous techniques? Are these standards-based?
c) Which GPGPU language (if any) can one target in the sure knowledge that it is future-proof? In which ways is this superior to OpenCL?
These are genuine questions that I really want answers to.
Having researched both, OpenCL is definitively better by far. Granted, CUDA has the native advantage, but that's not always going to be there, and I think most would agree that vendor tie-in is a Very Bad Thing (tm)
While you are completely correct in what you say, you are not answering the question. The question is, how do I get in to massively parallel GPU processing easily -- not, 'is my application suited to this?'. It's assumed that the OP might be an idiot, however he is asking a completely valid question, and you should be responding accordingly.
Actually, this post has prompted me to re-evaluate my methodologies. I have an enormous image-processing project to complete in short order. I was previously intending to use CPU only, but this post made me think, and go and research OpenCL. What I've found is that not only is it cross-platform, very fast (in many cases), but is also very easy to use (if you already know C very well, as I do).
The answer is, 'High level languages are not appropriate for this problem set'.
So, while I must agree that the original question of how to easily use GPGPU techniques from a purely managed environment such as C# is actually pretty inane, the result has been fruitful in that it highlights that if you really want GPGPU performance then you really do need to invest the effort required into actually learning the specific languages necessary. That's a relevant answer, in my book.
Something is seriously wrong with your maths here. Don't quit your day job.
Thruster = 0.000236 N
Mass = 2000
f=ma ; a=f/m=0.000236/2000=1.18e-7 (this is a VERY low acceleration)
s=ut+(1/2)at^2 (u=0 so we ignore that part)
assume t=24000 hrs=2e9 seconds (to turnaround point)
s=0.5 * 1.18e-7 * 2e9 * 2e9=2.360e9 metres
By turn-around you would have travelled 2.36 million kilometres.
1 light year is 9.461e15 metres, so the space craft would have travelled about 2.5 millionths of a light year.
Bzzt you are both wrong. The net acceleration due to this test is zero, because the ions ejected out of the engine are halted by the test chamber. Net result is zero force.
http://en.wikipedia.org/wiki/Dawn_(spacecraft)
It was launched in 2007 with an ion thruster.
Fair point, but getting the software to play nice with such a setup is not currently viable ...
Also, your $50 tablet is actually a fully-fledged computing device. I wouldn't really call it a thin client. All it needs is an internet connection.
Shared networking, yes that's far more economical. Put a few ethernet ports in each apartment -- great!
But that is NOT what the OP is talking about. The thin client model is ludicrous.
Fully agreed.
Nonsense and bullshit. In general, the whole planet has acquiesced to the US convention of 1 billion == 1 000 000 000. Those who don't agree are doomed to be misunderstood and sidelined.