A question, but isn't the analysis rather simplistic, to be more realistic we'd have to look at terminal velocity, the fact that most of the mass of the rock would burn up upon reentry and a couple of other things.
Off the top of my head, and I didn't pay too much attention to the books descriptions, when I'd read it, so I may be wrong.
Re:you can't stop the doomsayers
on
LHC Success!
·
· Score: 1
Yes, I agree with what you said, however if you were to look at it this way. News channels should also not be allowed show strobing/flashing lights of the type that can trigger epileptic fits. Coupled with the fact that these guys were warned by the broadcast ministry to tone down on their fear mongering - I'd say the new channels do indeed bear a certain burden especially in this case. But yes, they aren't the only ones at blame.
Re:you can't stop the doomsayers
on
LHC Success!
·
· Score: 1
Well, a girl offed her self some time back in India, the theory is she did it cause of the doomsday predictions shown on "news" channels
I just looked it up, since I was working with projective geometry based implementations on hardware, I never quite worried about optimising the algorithms. I see that you have done work in Galvois fields so you'll know what I'm talking about.
Actually the sparse matrix based implementations are the ones that I've seen being researched so pardon my ignorance of the other side. The challenges involved are not inconsiderable, however the gains are much more interesting and this is where most VLSI implementations focus on - note the DVBS2 standard - it uses hamming codes with LDPC if I'm not mistaken - the implementations I've seen - are all type II PG based regular parity check matrix based. Still - thanks for the information now at least I'm aware of the fact that there is research being done in avoiding sparse matrix based implementations, for the record though it is these sparse matrix based implementations that give you best performance for a channel.
LDPC works for AWGN, I know because I have implemented the decoder. One can just perform min-sum rather than belief propagation, I've never heard of a "loopy" variant though. Now interestingly, research has been done on using LDPC for magnetic storage media - this was an year ago, so their error probability distribution should have been modelled I guess, I'm sure something more must be out.
The down side is to use LDPC it would make more sense if like in those research papers you were thinking of using it with backup magnetic tapes or something, since LDPC requires large message sizes and the way we normally write to the HDD isn't going to be the best for use of LDPC.
LDPC for those who didn't know is Low Density Parity Check Codes, they are a class of codes discovered in the 1960's and have been found to be asymptotically close to the Shannon limit of a channel. Wikipedia has a decent enough introduction for them - Also check out Turbo Codes, and Vitterbi codes - all of them more interesting to work with than Reed-Solomon
connecting them together and the fudge factor for making sure it's the sustained performance and not hte theoretical peak would add the rest of the 110 Million $.
ps: yes those are very expensive 'wires' used to 'connect' them.
Well I would like to argue that, though this one doesn't I know there are some people in IBM looking into the TESLA GPUS's they pack a mean punch and now come with inherent support for double precision arithmetic - the only down side is its power usage and its form factor. At each one able to perform at about 200 GFLOPS even a couple of hundred of them would be extremely useful.
U-Texas Austin is doing some work on an architecture called TRIPS, there was a prof. in U Maryland who was working on a 'Desktop Super Computer' so the point is it's possible and people are working on it right now.
The when is a little difficult but it won't be 30-40 years for a teraflop chip.
Most people who do reviewing are knowledgeable to the point of expertise in their field (This can be questioned but the reviewers are at least grad students who have published on the topic once). The people on slashdot - not so knowledgeable about high energy physics.
Like your Sibling post had said, Diamond is tetrahedrally bonded carbon, it doesn't have that extra electron that makes 'graphite' a good conductor. Diamond here is not being used as a semiconductor from what I have understood (why I feel it cannot be used as one, not being one after all - I have written later).
You are correct about mixing 3rd group with 5th group but diamond (Carbon) is 4th group. notice however that you can mix 4th group with 5th group to give you your n type, which I already said still requires that the 4th group be a semiconductor which diamond is not.
I know about dopants etc. having spent four years in Electronics, however not being material sciences or something I may still be not completely correct. But thanks for the info any ways. Nitrogen has an extra electron but is also extra stable N_2 being almost inert (please notice the _almost_ ). I think it's cause of the strength of the tripple P-P orbital bonds.
Diamond would not at all make a good semiconductor for conventional computers the simplest reason being - Diamond is not a semiconductor, the more complete answer involving lots of quantum physics and me having to explain fermi levels. Please do check this out on wikipedia or some place (I'd suggest a book but I guess you may not have access to books on semiconductors unless you are an EE or in material sciences or something).
we need to add impurities to silicon to get it to work the way we want it, these guys are adding impurities to carbon to get it to act the way they want.
The resemblance may be superficial but it is there, nature is symmetric after all.
Possibly, however no one in their right minds would process a matrix of that size without blocking. There are several 'Block SVD' algorithms, there are also some projects done by students on doing SVD on GPUs.
Agreed that the matrices that it is processing are large but what is the size of the blocks that it is processing - IIRC the 8600 that the guys in my lab are working on can have 32 threads running in parallel however each thread will be working on a very small (I've been told it's upto 32x32 double precision floating point and that's stretching it)
Thus -blocking- as I had said comes into play, look I'm not downplaying the usefulness of the GPU or anything, nor am I downplaying their research, it's just that it's to be expected and the usefulness to general applications is limited.
Well, consider it this way - high level to VHDL/Verilog sucks a lot. So you are left with optimising VHDL code for performing highly parallel computations. FPGAs only recently have started getting competitive with GPUs etc. with the advent of large FPGAs which clock all the way upto 500 MHz, and have enough resources to make your mouth water (Xilinx LX330 has 54K LUT-6 slices).
There was a great paper by a guy called Underwood, which showed that in around 2009-2010 FPGAs could manage to outperform CPUs at computationally intensive tasks, having worked on this for some time now I do find my self believing him.
That is not entirely true - they are 'amazingly fast' only for a set of matrix operations of some limited size. They are not particularly effective for very large sizes, fortunately as you said tomography is essentially image processing and image processing matrices are usually not very large (filtering etc. etc. all will be done by matrices of sizes - 3x3- 9x9 such like, by a large matrix I'm talking of the order of 5000x50,000)
Ok, now I know a little about this, so let me expand a little on what you said - specifically matrix vector.
in the case of a matrix vector operation Ax = b, the amount of reuse for a given vector x is limited to the number of rows of the matrix - now the costs you incur is essentially in data movement operations - the cost of bringing the Vector in and the cost of bringing the element of A in.
The cost of the vector is amortised over time due to the parallelism that we see, but the elements of A are reused only once, and this ! is a problem making matrix-vector a bandwidth limited operation rather than a truly parallelizable one, matrix-matrix on the other hand is memory limited and can be said to be more parallelizable.
In short - O(n^m) the higher the m - chances are the more parallelizable the operation, making it a better target for acceleration.
Agreed, but their interconnect topology is what should be interesting not just the hardware, after all with simple topologies etc., there is a limit to how it scales efficiently, I have been doing some work on parallel processing for supercomputers as my undergrad thesis and believe me the major thing that differs amongst the top some 100 super computers is their interconnect topology not just their hardware.
Also, their search algo is based on eigen values I think, a very very profitable algo to parallelize. what version of parallel libraries do they use ?
Well, not like the human brain, but yes the idea is correct. For the human brain there are other approaches that may be a lot more interesting and a lot more realistic to achieve.
Well I'm not sure about how they have used hybrid cores, but try multicores with one core being an - on the fly reconfigurable FPGA' these kind of things would be so totally awesome if used properly. I think IBM is moving along similar lines for it's CELL series, there may be a tie-up with Xilinx involved - that I'm not so sure of, but it can be used most interestingly.
And is not all that obvious to most of the people who are just keeping up with computers rather than computing.
It is also a rather stringent religion requiring that you marry within the community only, and converts from other religions are not exactly allowed (to the best of my knowledge). I only of about the parsi's cause they live here, I've never questioned them about their religion (who does that ?).
They are older than the Zen form of Buddhism from what I remember.
Actually, the superscalar nature of the Intel processors lends to some interestig low level optimisations that the hardware itself does with the instructions, out of order execution, branching pipeline etc. (although it sounds like jargon, it isn't when you ahve worked on it for optimising along the lines of the BLAS libraries, you end up having to deal with this stuff).
To reiterate:- compilers are sticking to optimising the CISC code- true. however that is becaues:- The SSC architecture causes very low level optimisation to take place 'autmoatically' (the hardware takes care of that).
Your willingness to jump to conclusions based on the slightest of 'observations' that you seem to have _made_ are even more so.
Seriously though, have you not read slashdot comments, yes there is a lot of 'anti-republican' bias but are you telling me you have never noticed that there are pro-republican posts too ? (not to forget Ron Paul).
Essentially I find it interesting that you who thought that you had to point out that I wasn't drawing my own conclusions, don't realise that there would be others like you who point out the political bias's on slashdot - which even if I do say so myself I am rather well aware of.
Wow, so this is like a throw back to the days of the Raj man, I can't believe we still have this. Some of these laws seem so old I'd have been pretty sure they weren't updated even though we got independence, but then the fact that this acknowledges the internet as a possible distribution mechanism means that it is indeed rather 'modern' (chronologically only). No wonder the Shiv Sena guys are able to get a moral highground - even with respect to free speech, they are actually doing something that is legal (despite it being a variation on vigilantism I guess). Wow, no wonder people try to leave the country(a completely subjective opinion on my part not a fact or flamebait or troll), so my next question, I have a rough idea about the US (go slashdot), hows the situation with respect to freepeech laws and all in Europe ?
Interestingly this could be right out of a star-treck episode, some directive (not the prime and I forget which one this is) states that - when in an alien planet their laws and customs are what should be respected even if you think they are illeagle or morally incorrect. Do a lot of Multinationals have problems like this ?
Slashdot Mirror
A question, but isn't the analysis rather simplistic, to be more realistic we'd have to look at terminal velocity, the fact that most of the mass of the rock would burn up upon reentry and a couple of other things.
Off the top of my head, and I didn't pay too much attention to the books descriptions, when I'd read it, so I may be wrong.
Yes, I agree with what you said, however if you were to look at it this way. News channels should also not be allowed show strobing/flashing lights of the type that can trigger epileptic fits. Coupled with the fact that these guys were warned by the broadcast ministry to tone down on their fear mongering - I'd say the new channels do indeed bear a certain burden especially in this case. But yes, they aren't the only ones at blame.
Well, a girl offed her self some time back in India, the theory is she did it cause of the doomsday predictions shown on "news" channels
http://timesofindia.indiatimes.com/India/Fearing_end_of_world_girl_kills_self/articleshow/3467519.cms
I just looked it up, since I was working with projective geometry based implementations on hardware, I never quite worried about optimising the algorithms. I see that you have done work in Galvois fields so you'll know what I'm talking about.
Actually the sparse matrix based implementations are the ones that I've seen being researched so pardon my ignorance of the other side. The challenges involved are not inconsiderable, however the gains are much more interesting and this is where most VLSI implementations focus on - note the DVBS2 standard - it uses hamming codes with LDPC if I'm not mistaken - the implementations I've seen - are all type II PG based regular parity check matrix based. Still - thanks for the information now at least I'm aware of the fact that there is research being done in avoiding sparse matrix based implementations, for the record though it is these sparse matrix based implementations that give you best performance for a channel.
LDPC works for AWGN, I know because I have implemented the decoder. One can just perform min-sum rather than belief propagation, I've never heard of a "loopy" variant though. Now interestingly, research has been done on using LDPC for magnetic storage media - this was an year ago, so their error probability distribution should have been modelled I guess, I'm sure something more must be out.
The down side is to use LDPC it would make more sense if like in those research papers you were thinking of using it with backup magnetic tapes or something, since LDPC requires large message sizes and the way we normally write to the HDD isn't going to be the best for use of LDPC.
LDPC for those who didn't know is Low Density Parity Check Codes, they are a class of codes discovered in the 1960's and have been found to be asymptotically close to the Shannon limit of a channel. Wikipedia has a decent enough introduction for them - Also check out Turbo Codes, and Vitterbi codes - all of them more interesting to work with than Reed-Solomon
AC Current - Alternating Current Current
Similarly DC Current.
AC voltage. DC Voltage.
I'm an EE can you tell ?
connecting them together and the fudge factor for making sure it's the sustained performance and not hte theoretical peak would add the rest of the 110 Million $.
ps: yes those are very expensive 'wires' used to 'connect' them.
Well I would like to argue that, though this one doesn't I know there are some people in IBM looking into the TESLA GPUS's they pack a mean punch and now come with inherent support for double precision arithmetic - the only down side is its power usage and its form factor. At each one able to perform at about 200 GFLOPS even a couple of hundred of them would be extremely useful.
U-Texas Austin is doing some work on an architecture called TRIPS, there was a prof. in U Maryland who was working on a 'Desktop Super Computer' so the point is it's possible and people are working on it right now.
The when is a little difficult but it won't be 30-40 years for a teraflop chip.
Most people who do reviewing are knowledgeable to the point of expertise in their field (This can be questioned but the reviewers are at least grad students who have published on the topic once). The people on slashdot - not so knowledgeable about high energy physics.
Agreed, which is why I said superficial.
Like your Sibling post had said, Diamond is tetrahedrally bonded carbon, it doesn't have that extra electron that makes 'graphite' a good conductor. Diamond here is not being used as a semiconductor from what I have understood (why I feel it cannot be used as one, not being one after all - I have written later).
You are correct about mixing 3rd group with 5th group but diamond (Carbon) is 4th group. notice however that you can mix 4th group with 5th group to give you your n type, which I already said still requires that the 4th group be a semiconductor which diamond is not.
I know about dopants etc. having spent four years in Electronics, however not being material sciences or something I may still be not completely correct. But thanks for the info any ways. Nitrogen has an extra electron but is also extra stable N_2 being almost inert (please notice the _almost_ ). I think it's cause of the strength of the tripple P-P orbital bonds.
Diamond would not at all make a good semiconductor for conventional computers the simplest reason being - Diamond is not a semiconductor, the more complete answer involving lots of quantum physics and me having to explain fermi levels. Please do check this out on wikipedia or some place (I'd suggest a book but I guess you may not have access to books on semiconductors unless you are an EE or in material sciences or something).
Well I'm not sure about that, but there is an interesting co-incidence here, silicon is in the 4th group, and so is diamond(carbon).
electrical(electronics) computers -> silicon
Quantum computers -> Diamonds (carbon)
we need to add impurities to silicon to get it to work the way we want it, these guys are adding impurities to carbon to get it to act the way they want.
The resemblance may be superficial but it is there, nature is symmetric after all.
Possibly, however no one in their right minds would process a matrix of that size without blocking. There are several 'Block SVD' algorithms, there are also some projects done by students on doing SVD on GPUs.
http://www.cs.unc.edu/~geom/Numeric/svd/
http://www.google.co.in/search?q=block+SVD&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-US:official&client=firefox-a
Agreed that the matrices that it is processing are large but what is the size of the blocks that it is processing - IIRC the 8600 that the guys in my lab are working on can have 32 threads running in parallel however each thread will be working on a very small (I've been told it's upto 32x32 double precision floating point and that's stretching it)
Thus -blocking- as I had said comes into play, look I'm not downplaying the usefulness of the GPU or anything, nor am I downplaying their research, it's just that it's to be expected and the usefulness to general applications is limited.
Well, consider it this way - high level to VHDL/Verilog sucks a lot. So you are left with optimising VHDL code for performing highly parallel computations. FPGAs only recently have started getting competitive with GPUs etc. with the advent of large FPGAs which clock all the way upto 500 MHz, and have enough resources to make your mouth water (Xilinx LX330 has 54K LUT-6 slices).
There was a great paper by a guy called Underwood, which showed that in around 2009-2010 FPGAs could manage to outperform CPUs at computationally intensive tasks, having worked on this for some time now I do find my self believing him.
LAPACK and LINPACK and SCALPACK, why measure using your own benchmark when you don't mention performance with the standard ones.
That is not entirely true - they are 'amazingly fast' only for a set of matrix operations of some limited size. They are not particularly effective for very large sizes, fortunately as you said tomography is essentially image processing and image processing matrices are usually not very large (filtering etc. etc. all will be done by matrices of sizes - 3x3- 9x9 such like, by a large matrix I'm talking of the order of 5000x50,000)
Ok, now I know a little about this, so let me expand a little on what you said - specifically matrix vector.
in the case of a matrix vector operation Ax = b, the amount of reuse for a given vector x is limited to the number of rows of the matrix - now the costs you incur is essentially in data movement operations - the cost of bringing the Vector in and the cost of bringing the element of A in.
The cost of the vector is amortised over time due to the parallelism that we see, but the elements of A are reused only once, and this ! is a problem making matrix-vector a bandwidth limited operation rather than a truly parallelizable one, matrix-matrix on the other hand is memory limited and can be said to be more parallelizable.
In short - O(n^m) the higher the m - chances are the more parallelizable the operation, making it a better target for acceleration.
Agreed, but their interconnect topology is what should be interesting not just the hardware, after all with simple topologies etc., there is a limit to how it scales efficiently, I have been doing some work on parallel processing for supercomputers as my undergrad thesis and believe me the major thing that differs amongst the top some 100 super computers is their interconnect topology not just their hardware.
Also, their search algo is based on eigen values I think, a very very profitable algo to parallelize. what version of parallel libraries do they use ?
Well, not like the human brain, but yes the idea is correct. For the human brain there are other approaches that may be a lot more interesting and a lot more realistic to achieve.
Well I'm not sure about how they have used hybrid cores, but try multicores with one core being an - on the fly reconfigurable FPGA' these kind of things would be so totally awesome if used properly. I think IBM is moving along similar lines for it's CELL series, there may be a tie-up with Xilinx involved - that I'm not so sure of, but it can be used most interestingly.
And is not all that obvious to most of the people who are just keeping up with computers rather than computing.
Actually, it does have good evil conflict.
[slightly off-topic]
It is also a rather stringent religion requiring that you marry within the community only, and converts from other religions are not exactly allowed (to the best of my knowledge). I only of about the parsi's cause they live here, I've never questioned them about their religion (who does that ?).
They are older than the Zen form of Buddhism from what I remember.
Actually, the superscalar nature of the Intel processors lends to some interestig low level optimisations that the hardware itself does with the instructions, out of order execution, branching pipeline etc. (although it sounds like jargon, it isn't when you ahve worked on it for optimising along the lines of the BLAS libraries, you end up having to deal with this stuff).
:- compilers are sticking to optimising the CISC code- true. :-
To reiterate
however that is becaues
The SSC architecture causes very low level optimisation to take place 'autmoatically' (the hardware takes care of that).
Your willingness to jump to conclusions based on the slightest of 'observations' that you seem to have _made_ are even more so.
Seriously though, have you not read slashdot comments, yes there is a lot of 'anti-republican' bias but are you telling me you have never noticed that there are pro-republican posts too ? (not to forget Ron Paul).
Essentially I find it interesting that you who thought that you had to point out that I wasn't drawing my own conclusions, don't realise that there would be others like you who point out the political bias's on slashdot - which even if I do say so myself I am rather well aware of.
Wow, so this is like a throw back to the days of the Raj man, I can't believe we still have this. Some of these laws seem so old I'd have been pretty sure they weren't updated even though we got independence, but then the fact that this acknowledges the internet as a possible distribution mechanism means that it is indeed rather 'modern' (chronologically only). No wonder the Shiv Sena guys are able to get a moral highground - even with respect to free speech, they are actually doing something that is legal (despite it being a variation on vigilantism I guess). Wow, no wonder people try to leave the country(a completely subjective opinion on my part not a fact or flamebait or troll), so my next question, I have a rough idea about the US (go slashdot), hows the situation with respect to freepeech laws and all in Europe ?
Interestingly this could be right out of a star-treck episode, some directive (not the prime and I forget which one this is) states that - when in an alien planet their laws and customs are what should be respected even if you think they are illeagle or morally incorrect. Do a lot of Multinationals have problems like this ?