It's been a while since I took my radiation safety course, but IIRC the problem with Pu is that it's a bone seeker, i.e. Pu will accumulate in the bone.
While the activity of Pu is low enough that you'd need to chew a truly massive amount to get some kind of acute radiation sickness (vomiting etc.), over the longer term even a relatively small amount of Pu can cause massive localized damage (Pu is an alpha emitter). As that Pu is then found in the bone as I already mentioned, it means leukemia. Not nice.
Also, the way Pu exposure happens is in practice from inhaling Pu dust, not digesting it.
Rubidium 87 has a half-life of 47 billion (10^9) years (our soloar system is not yet 5 billion years old). Uranium 238 has a half-life of 4.5 Billion (10^9) years, Plutonium 239 has a half-life of 25.000 years. Half-life means that after some billion years, you still have half of your nuclear waste happily emitting radioactivity, while the other half has decayed to other, possibly also radioactive elements.
Correct. OTOH, the longer the half-life the less intense will the radiation be, as there are of course fewer nuclei decaying per unit time per unit mass.
See, the danger in nuclear waste is not the extremely long aged compounds like Ru87 or U238, whose radiactivity is not very much higher than the background radiation. Also, short-lived compounds (i.e. the ones which in the short term constitude the vast majority of the radiation) aren't the most problematic either, since mankind can certainly contain them for the few decades required.
The real problem is the compounds of medium half-life, like Pu. These materials radiate strongly, and can present a serious danger to anyone coming into contact with them. Also, containing them requires containing them for longer than the time civilization has been around.
Lessons learned: PR people don't know anything about mathematics and radioactivity.
Lesson 2: Pundits on/. don't necessarily know better than said PR department.
I'd rather you not correct it with a half-truth like say, oh, picking an extremely long-lived compound whose contribution to the total radioactivity of the waste is minute at best.
Plutonium also has a very long half life.
24000 years, IIRC. As you admit yourself, after 10x the half-life (=240000 years, in the same ballpark as the 200 000 years claimed) most of the radiactivity from plutonium has disappeared.
I assume you wouldn't eat it for breakfast.
Of course not. I wouldn't want to eat any other heavy metal for breakfast either, they all tend to be quite toxic to biological life.
While it's good to see another neat/good idea, the problem is having a place to put it. Until such a site exists AND IS ALLOWED TO OPERATE, we're left twiddling our thumbs. Since nothing is 100% safe and secure, I'm not optimistic such a site will be operational.
Unfortunately, that is the political reality.
However, IMHO any reasonably well thought out burial method, flaws and all, is still orders of magnitude better than how nuclear waste is currently stored in the world.
In a way, this is just another case of the NIMBY crowd winning against the best interests of the rest of mankind.
Umm, Zoche is nothing more than a way to scam R&D money from the german government. The specs of the engine look fantastic and all, but zoche has been saying "FAA/JAA certification next year" for the past, oh, 15 years or so. So far he has shown nothing but empty promises.
As far as aviation diesels go, Thielert and SMA produce certified engines, flying today in certified installations (some new planes, some STC:s for Cessnas IIRC). There are other serious players too, Deltahawk is nearing certification, and perhaps Wilksch too.
Of course, they're pledging to continue selling Itanium servers.
In the longer run, IMHO it sounds somewhat problematic, considering that all the engineers developing software will be running on x86-64. I.e. the software will first be available on x86-64, more tested etc.
So why should the customer shell out money for an Itanium server instead of an x86-64 server which has better bang-per-buck and runs the software more reliably? In the short run HP can probably contain x86-64 in low end servers, keeping high end stuff reserved for Itanic. But in the long run, they'll have to start providing higher end x86-64 gear too, or their customers will move to a competitor that will.
Personally, I have an old computer with a couple of extra disks, running dirvish. It's cheap, and has enough capacity to backup my homedir and some other stuff as well, and dirvish offers snapshot like capability (uses hardlinks to save space).
Lets all just do the enormously satisfying deed of pissing off the politically correct crowd, and call it clitoris, as it was meant to be called.
And before you ask "WTF", let me explain.
Over here, the word for "mouse" is also a slang word for pussy. So what do you call the slightly protruding feature that you twiddle with you finger?;-)
> What's more difficult is continually > optimizing for the various machine architectures
Hm, that's interesting. Is that something that would be done in Fortran using some sort of pragma-ish hints? Or is it something the Fortran interpreter writers would be mostly concerned with?
Googling a bit reveals a couple of Fortran compilers... seems like that's where the per-architecture optimization would happen.
Yes, most of the optimizations for specific hardware concern only compiler writers. Just like C, Java or most other languages.
But maybe the "end-user" programmer needs to do some tweaking as well?
In some cases yes. As an example, the advent of heavily cache-based cpu designs in the 90:ies (e.g. all modern cpus) suddenly made iteration order very important. If you iterate through an big multidimensional array in the "wrong" order, the code will be orders of magnitude slower than if you do it in the right order. The right order being the same order as the array is laid out in memory.
That being said, in general Fortran has survived changing hardware architectures pretty well. This is largely because the Fortran standard mostly concerns itself with what a standard-conforming program should do, not the details of how it is actually done. This allows compilers writers considerable leeway in implementing features as they see fit, rather than the standard explicitly spelling out how it's must be done.
What's the maintenance burden like for a large body of Fortran code?
Roughly speaking, about half that of a comparable C program.
Speed-wise, Fortran is about as good as it gets unless you want to go into asm. Fortran aliasing rules allow more aggresive optimization than C/C++ (although in this regard C99 achieves the same thing with restricted pointers). Also, the current Fortran version, Fortran 95, has a slightly matlab-like array language, where you can express many computations directly as whole array operations instead of needing explicit loops. As you certainly can imagine, mapping these array operations onto vector processors is straightforward. Of course, even if you don't have a vector processor, the array language is seriously neat is it can cut down considerably on the code size.
The problem here is resource allocation. As GNOME and KDE (or any other project, for that matter) have finite resources (i.e. volunteer time), if they spend some time on improving performance, it's less time spent on something else (such as the features that attract people to kde/gnome in the first place).
IMHO, KDE and GNOME have their priorities pretty well laid out as it is. The nice thing with FOSS is that KDE and GNOME doesn't have to be everything for everyone as MS trie with windows. E.g. if you want features and eye candy, use KDE or GNOME. If you want speed, lack of bloat etc. for that old P133, there's plenty of choice in lightweight WM:s such as openbox.
And if you feel KDE or GNOME is slow, it's not like a new computer is exactly expensive. Say, an athlon xp 2600 with 0.5-1 Gb RAM that runs both desktops more than well costs what, EUR 500-1000?
The way the Solaris kernel is so scaleable across over 100 processors is not some clever hack, it's taken years of refinement of the kernel.
Well, I'd guess that Linux with the various SGI patches that run on the SGI 512 CPU systems aren't "some clever hack" either, for that matter if that's what you're trying to imply. It's the result of years of work SGI put into making IRIX scale that has been ported to Linux.
I would have thought that their SMP hardware would be interesting to geeks - unlike the compromised NUMA architecture that lesser Unix boxes run.
As others have said, the current generation of big iron Suns (the Sun Fire) is NUMA. You can find details in
Alan E. Charlesworth: The sun fireplane system interconnect, Proceedings of SC2001 (available online if you have an ACM subscription).
While Suns combination of snooping and directory based cache coherency is neat, it nothing revolutionary or so, that other companies would somehow be unable to implement if they wanted to.
If Sun tanks, the world will go on just fine without them.
Solaris should definitely be of interest to anyone interested in UNIX or Linux - unless features such as partitioning and scalability are dull?
For the vast majority of users who have no need for partitioning nor extreme scalability, yes those features are pretty dull. And those who need those features know that Sun isn't the only game in town.
If Sun tanks, the world will go on just fine without them.
There's no hard line between "cluster" and "networked independent computers". If you wan't to make some distinction, it could be that cluster nodes are pretty homogenous, and the cluster has a dedicated network instead of just using the office LAN.
Re:It's not just about speed and massively paralle
on
Linux Clustering
·
· Score: 1
To be considered a "supercomputer," it also needs enough CONTIGUOUS MEMORY SPACE to hold the massive amounts of data associated with true "supercomputing."
Well, that's one way of seeing it I guess. A way not shared by most people in supercomputing, I might add.
Though I guess if consumers keep buying them, then you can't really blame Lucas too much.
We have a saying over here that translated to english goes something like "the idiot is not the one who asks, but the one who pays".
E.g. if you think this is fantastic and all, go buy the set and enjoy it. If you think it's a ripoff and so on, well don't spend your money on it. It's not like you're losing out on anything by Lucas making yet another "rip em off" edition.
Slashdot Mirror
It's been a while since I took my radiation safety course, but IIRC the problem with Pu is that it's a bone seeker, i.e. Pu will accumulate in the bone.
While the activity of Pu is low enough that you'd need to chew a truly massive amount to get some kind of acute radiation sickness (vomiting etc.), over the longer term even a relatively small amount of Pu can cause massive localized damage (Pu is an alpha emitter). As that Pu is then found in the bone as I already mentioned, it means leukemia. Not nice.
Also, the way Pu exposure happens is in practice from inhaling Pu dust, not digesting it.
Rubidium 87 has a half-life of 47 billion (10^9) years (our soloar system is not yet 5 billion years old). Uranium 238 has a half-life of 4.5 Billion (10^9) years, Plutonium 239 has a half-life of 25.000 years. Half-life means that after some billion years, you still have half of your nuclear waste happily emitting radioactivity, while the other half has decayed to other, possibly also radioactive elements.
Correct. OTOH, the longer the half-life the less intense will the radiation be, as there are of course fewer nuclei decaying per unit time per unit mass.
See, the danger in nuclear waste is not the extremely long aged compounds like Ru87 or U238, whose radiactivity is not very much higher than the background radiation. Also, short-lived compounds (i.e. the ones which in the short term constitude the vast majority of the radiation) aren't the most problematic either, since mankind can certainly contain them for the few decades required.
The real problem is the compounds of medium half-life, like Pu. These materials radiate strongly, and can present a serious danger to anyone coming into contact with them. Also, containing them requires containing them for longer than the time civilization has been around.
Lessons learned: PR people don't know anything about mathematics and radioactivity.
Lesson 2: Pundits on
You'd rather I didn't correct it?
I'd rather you not correct it with a half-truth like say, oh, picking an extremely long-lived compound whose contribution to the total radioactivity of the waste is minute at best.
Plutonium also has a very long half life.
24000 years, IIRC. As you admit yourself, after 10x the half-life (=240000 years, in the same ballpark as the 200 000 years claimed) most of the radiactivity from plutonium has disappeared.
I assume you wouldn't eat it for breakfast.
Of course not. I wouldn't want to eat any other heavy metal for breakfast either, they all tend to be quite toxic to biological life.
While it's good to see another neat/good idea, the problem is having a place to put it. Until such a site exists AND IS ALLOWED TO OPERATE, we're left twiddling our thumbs. Since nothing is 100% safe and secure, I'm not optimistic such a site will be operational.
Unfortunately, that is the political reality.
However, IMHO any reasonably well thought out burial method, flaws and all, is still orders of magnitude better than how nuclear waste is currently stored in the world.
In a way, this is just another case of the NIMBY crowd winning against the best interests of the rest of mankind.
There will be some residual radioactivity in any nuclear waste forever
Of course, but after a few hundred thousand years it will IIRC be at about the same level as background radiation.
Umm, Zoche is nothing more than a way to scam R&D money from the german government. The specs of the engine look fantastic and all, but zoche has been saying "FAA/JAA certification next year" for the past, oh, 15 years or so. So far he has shown nothing but empty promises.
As far as aviation diesels go, Thielert and SMA produce certified engines, flying today in certified installations (some new planes, some STC:s for Cessnas IIRC). There are other serious players too, Deltahawk is nearing certification, and perhaps Wilksch too.
Good. It's probably optimized for flight simulators then.
Of course, they're pledging to continue selling Itanium servers.
In the longer run, IMHO it sounds somewhat problematic, considering that all the engineers developing software will be running on x86-64. I.e. the software will first be available on x86-64, more tested etc.
So why should the customer shell out money for an Itanium server instead of an x86-64 server which has better bang-per-buck and runs the software more reliably? In the short run HP can probably contain x86-64 in low end servers, keeping high end stuff reserved for Itanic. But in the long run, they'll have to start providing higher end x86-64 gear too, or their customers will move to a competitor that will.
Yeah, HP is turning into a bunch of vacuum cleaner salesmen, just like Dell.
Luckily some of the old HP spirit is left in Agilent.
Personally, I have an old computer with a couple of extra disks, running dirvish. It's cheap, and has enough capacity to backup my homedir and some other stuff as well, and dirvish offers snapshot like capability (uses hardlinks to save space).
Spin is basically a quantized angular momentum intrinsic to many particles (electrons are spin 1/2, photons are spin 1).
If you dig slightly deeper into the realm of relativistic quantum mechanics, it turns out that spin is nothing more than a relativistic effect.
Lets all just do the enormously satisfying deed of pissing off the politically correct crowd, and call it clitoris, as it was meant to be called.
;-)
And before you ask "WTF", let me explain.
Over here, the word for "mouse" is also a slang word for pussy. So what do you call the slightly protruding feature that you twiddle with you finger?
After RTFA, I'd guess those responsible would rather remain anonymous.
It means "So long, and thanks for all the tax money", to paraphrase D. Adams.
> What's more difficult is continually
> optimizing for the various machine architectures
Hm, that's interesting. Is that something that would be done in Fortran using some sort of pragma-ish hints? Or is it something the Fortran interpreter writers would be mostly concerned with?
Googling a bit reveals a couple of Fortran compilers... seems like that's where the per-architecture optimization would happen.
Yes, most of the optimizations for specific hardware concern only compiler writers. Just like C, Java or most other languages.
But maybe the "end-user" programmer needs to do some tweaking as well?
In some cases yes. As an example, the advent of heavily cache-based cpu designs in the 90:ies (e.g. all modern cpus) suddenly made iteration order very important. If you iterate through an big multidimensional array in the "wrong" order, the code will be orders of magnitude slower than if you do it in the right order. The right order being the same order as the array is laid out in memory.
That being said, in general Fortran has survived changing hardware architectures pretty well. This is largely because the Fortran standard mostly concerns itself with what a standard-conforming program should do, not the details of how it is actually done. This allows compilers writers considerable leeway in implementing features as they see fit, rather than the standard explicitly spelling out how it's must be done.
What's the maintenance burden like for a large body of Fortran code?
Roughly speaking, about half that of a comparable C program.
Speed-wise, Fortran is about as good as it gets unless you want to go into asm. Fortran aliasing rules allow more aggresive optimization than C/C++ (although in this regard C99 achieves the same thing with restricted pointers). Also, the current Fortran version, Fortran 95, has a slightly matlab-like array language, where you can express many computations directly as whole array operations instead of needing explicit loops. As you certainly can imagine, mapping these array operations onto vector processors is straightforward. Of course, even if you don't have a vector processor, the array language is seriously neat is it can cut down considerably on the code size.
The problem here is resource allocation. As GNOME and KDE (or any other project, for that matter) have finite resources (i.e. volunteer time), if they spend some time on improving performance, it's less time spent on something else (such as the features that attract people to kde/gnome in the first place).
IMHO, KDE and GNOME have their priorities pretty well laid out as it is. The nice thing with FOSS is that KDE and GNOME doesn't have to be everything for everyone as MS trie with windows. E.g. if you want features and eye candy, use KDE or GNOME. If you want speed, lack of bloat etc. for that old P133, there's plenty of choice in lightweight WM:s such as openbox.
And if you feel KDE or GNOME is slow, it's not like a new computer is exactly expensive. Say, an athlon xp 2600 with 0.5-1 Gb RAM that runs both desktops more than well costs what, EUR 500-1000?
IIRC, with Solaris 9 Sun moved to a 1:1 thread model, just like LinuxThreads and NPTL.
Previously they had a M:N model.
The way the Solaris kernel is so scaleable across over 100 processors is not some clever hack, it's taken years of refinement of the kernel.
Well, I'd guess that Linux with the various SGI patches that run on the SGI 512 CPU systems aren't "some clever hack" either, for that matter if that's what you're trying to imply. It's the result of years of work SGI put into making IRIX scale that has been ported to Linux.
I would have thought that their SMP hardware would be interesting to geeks - unlike the compromised NUMA architecture that lesser Unix boxes run.
As others have said, the current generation of big iron Suns (the Sun Fire) is NUMA. You can find details in
Alan E. Charlesworth: The sun fireplane system interconnect, Proceedings of SC2001 (available online if you have an ACM subscription).
While Suns combination of snooping and directory based cache coherency is neat, it nothing revolutionary or so, that other companies would somehow be unable to implement if they wanted to.
If Sun tanks, the world will go on just fine without them.
Solaris should definitely be of interest to anyone interested in UNIX or Linux - unless features such as partitioning and scalability are dull?
For the vast majority of users who have no need for partitioning nor extreme scalability, yes those features are pretty dull. And those who need those features know that Sun isn't the only game in town.
If Sun tanks, the world will go on just fine without them.
Well, duh...
Most clusters are used to run MPI applications.
There's no hard line between "cluster" and "networked independent computers". If you wan't to make some distinction, it could be that cluster nodes are pretty homogenous, and the cluster has a dedicated network instead of just using the office LAN.
To be considered a "supercomputer," it also needs enough CONTIGUOUS MEMORY SPACE to hold the massive amounts of data associated with true "supercomputing."
Well, that's one way of seeing it I guess. A way not shared by most people in supercomputing, I might add.
Though I guess if consumers keep buying them, then you can't really blame Lucas too much.
We have a saying over here that translated to english goes something like "the idiot is not the one who asks, but the one who pays".
E.g. if you think this is fantastic and all, go buy the set and enjoy it. If you think it's a ripoff and so on, well don't spend your money on it. It's not like you're losing out on anything by Lucas making yet another "rip em off" edition.
Since we usually like to tell stories that show ourselves in a good light rather than bad, I'm guessing 10k. ;-)
Ah yes, anyone who claims that the allmighty Religion Of Capitalism (as opposed to the economic theory of Capitalism) has limits, is a Marxist.
I'm not saying that capitalism is perfect, it certainly has its share of problems.
Anyway, the point being, you're not "pointing out flaws in capitalism", you're foaming at the mouth in blind hatred of it.