I've run the same simulations on Xeons and Itaniums and the AMD's win hands down in terms of raw floating point throughput.
Spec2000 fp scores for Itanium 2's with 9M of cache are 2712/2712 sustained/max for a single CPU. For Athlon64 FX-55 (the top AMD chip, IIRC), the results are 1695/1878. Perhaps you are using old Itaniums? Or perhaps specfp is a really bad model for the computations you are interested in?
some of my tests have shown single A64's to out-perform dual Xeons rated similarly (2.8 Ghz Xeons vs. 2800+ A64's).
Sure. Xeons suck badly. But if you have the money, I believe Itaniums don't.
No, they are not making up the numbers. However, it takes some time to set up a top-ranking supercomputer (most of the time is not building the thing - you can just use college students paid in pizza - but debugging the inevitable problems in the network that arise as soon as you try putting any significant load on the system; I've worked next to a guy who spent months debugging a puny 12-box beowulf cluster, and problems are exponentially worse for the large supercomputers). The setup times are even worse for government and military machines - I suppose they need to meet strignent quality specs. As a result, many of the processors on top-500 are not state of the art (e.g. 1.25 GHz Alphas) so it might not the best place to look if you want to find out what CPU to use for a future supercomputer. If you notice, some of the highest scores are attained by IBM's BlueGene - which uses massive numbers of slow embedded PowerPC's, but comes out on top because of its excellent, fault-tolerant networking.
If I had infinite $ and a very big room, I would order 64 SGI Altix systems with 512 Itanium2's each, running SGI's custom Linux distro, and link those babies up in a big cluster. 32768 Itanium2's > 32768 PowerPC 440's.
and now with Intel's interest moving to a better 64-bit system[...]
The Itanium is a nice chip. If you want to build a supercomputer for floating-point-heavy scientific computations, and cost is not a factor, you really have two options: the Itanium, or a NEC vector processor. Plus, there are the aesthetics -- the EPIC architecure, the predicates, the totally new bios system.
It really is too bad that Intel never achieved economies of scale with the Itanium. Now, it looks that regular consumers are stuck with ugly upgrades to the ugly x86 architecture for the next century or so... Athlon-128 in 2020, anyone?
Second, anyone who uses unencrypted email on a server they do no control, ESPECIALLY if it belongs to someone they are screwing, deserves to spend the rest of their productive years flipping burgers, or possibly stamping licence plates.
--
In my opinion Microsoft is essentially a state-sponsored monopoly and, as such, represents, little more than a tweaked version of the classic communist state entity.
It's a monopoly - but not a state-sponsored one. The worst you can accuse it of is using the usual lock-in practices to make sure some gov't departments keep on using windows. If you want to see a government-sponsored monopoly, look at something like AT&T in the 1970's, or your local cable provider, or your local power company before the days of deregulation. In all of these cases, there were laws protecting the monopolist from competition, and regulations that made the monopolist essentially an arm of the government. Microsoft, however, managed to achieve and hold on to monopoly status without government assistance - a monumental achievement (even though it's just as bad for the consumer and the industry as a traditional state-sponsored monopoly).
When we had communism in Poland, most shops had empty shelves -- and within just months after the communism's fall any shortages were just gone, as if by a wave of a magic wand.
That magic wand took a long time to work in Russia...
Re:Definition of sourcecode of graphics files.
on
Revising the GPL
·
· Score: 1
oes the license require me to distribute the layered gimp file along with my binaries?
The discussion on kde-look implies that for SVG images, the XML text is the source code, and the raster images produced from it are the object files. I would guess that similarly for 3D graphics, your model definition would be the source, since the render can't really be modified if you don't have the model.
However, I really don't know about raster-based images. An XCF or a PSD document is easier to use if you want to make changes, but it's certainly quite possible to just push pixels around, especially for something like a 48x48 icon. Arguably, a flat PNG is more human-editable than a PSD document (not everyone has access to photoshop). Consider a similar situation: you use a small perl script to auto-generate a part of your C code, but the resulting code is still perfectly human-readable and human-editable. Do you need to distribute the script if you use GPL? Is there a lawyer in the house?
What happens if that galaxy we didn't see in our left blind spot doesn't use it's turn signal and rams right into us
Well, the only large galaxy scheduled to collide with us in the near future is Andromeda - and that collision is scheduled at least several hundred billion years from now. By then, we will either be extinct, or in control of the entire Local Cluster. If I were you, I really wouldn't worry about galactic collisions.
In Gentoo (well, at least in ~x86), just edit your/etc/conf.d/rc and set RC_PARALLEL_STARTUP="yes", and now at boot, your daemons start in parallel if possible. Don't know if other distros can do this yet - it looks like Gentoo's own invention.
Linux programmers won't silently tolerate incomplete API's with weird behavior where 3rd party Windows application developers must accept and learn "whatever Win32 does now".
I disagree. First, if you are using anything like Perl or Python, you essentially agree to accept and learn "whatever Larry, Damien, and Guido like at the moment". And their opinion tends to change with minor version numbers... However, at least Perl and Python are documented (and in the case of Python, readably so). The bad side is that many parts of GNU/Linux not only have API's consisting of "whatever the dev likes at the moment" - the API isn't even documented! Case in point: can you figure out what the current fdo standard is for which applications to launch to open a file of a given mimetype? There are some webpages about the topic, but they are outdated. To figure out what's really going on, you need to read two mailing lists, and glean the API from between the lines. Scary.
But that's nothing compared to the kernel. To figure out the inotify api, you need to read Love's website. (how many kernel devs are there? do you know all their websites? how many of them do you want to check when you have a problem?) To figure out what's going on with scsi support this week, you need to diligently read the lkml and the fedora bugzilla (and the behavior of the scsi subsistem fluctuates a lot over time). Much of the documentation in/usr/src/linux/Documentation is old and obsolete, and the new documentation must be figured out from source, from mailing lists, or by asking the devs themselves.
If there is one thing Microsoft can be proud of, it's good documentation, all organized in one place.
they patch the kernel to notify the system when files are modified.
That's because dnotify sucks donkey balls, as has been eloquently explained many times by anyone who has been forced to use that misbegotten kernel feature. If you want to have a clean file modification notification mechanism, you need inotify, which means either using kernel 2.6.8 or newer, or patching an older kernel.
Hard to tell. On the one hand, water vapor is a much more effective greenhouse gas than CO2, so lowering out CO2 output and increasing water vapor output may make the greenhouse effect worse. On the other hand, a very high concentration of H2O vapor would form a thick cloud layer, which would reflect solar radiation. On the third hand, the atmosphere can only hold so much humidity, so the extra H2O is likely to fall as precipitation, especially as snow in cold regions, which would change the distribution of albedo on Earth's surface and alter climates locally. On the fourth hand, humid air has a higher heat capacity than dry air, which also might alter the climate. On the fifth hand, a more humid atmosphere with less CO2 in it might change the distribution of plant life on the planet, further altering the atmosphere in some direction. Basically, there is no way to tell what will happen unless you set up a bunch of equations and plug them into the Earth Simulator.
If this technology is as good as Aurora claims, it can be used to implement a virtual ID card - just scan someone's face, and you can bring up their info from a database, no need for them to carry a piece of plastic around.
Obviously that's a privacy concern - but how can you regulate face recognition? It's fundamentally no different from having a live cop recognize your mug.
Sure, Turkey is a great country. However, this discussion is not about whether Turks are nice people or whether Turkey is a good place to live. The point of this discussion is that it is silly to arrest someone just for collecting (potentially) terrorist propaganda. If the guy was making bombs, or doing intelligence work for the PPK - sure, he deserves jail time. But arresting him just for managing a website?
Turkey's criminal justice system is quite good by, say, the standards of North Africa (there, Coskun would just "disappear"). However, it's clearly not good enough by the standards of EU...
the price of the FPGA (xc3s1500) which is quoted at Avnet at 70$US for the 320 pins version and 115$US for the 676 pins version (in qty of 25-99)...
Well, first Avnet charges a markup to stay in business; and second, I presume Tech Source wants to sell more than 99 of these babies. Xilinx quotes the price of an xc3s1500 as "under $20" for quantities of 250 000. Pulling some numbers out of my ass, I presume that Tech Source is going to make something like 10 000 of these cards, so they would be getting a big volume discount - maybe not $20 each, but certainly way lower than $70.
the usual suspects port there games to Linux - you know ID and Blizzard come to mind
Blizzard doesn't port their games to Linux - only to Mac, unfortunately. (I would buy a Linux version of WC3/TFT, if anyone from Blizzard is reading this...) Perhaps you are thinking of Epic?
What about 3DFX? [...] Would there be anything wrong with utilizing these old resources to achieve this goal?
NVidia bought 3Dfx and assimilated its intellectual property. I don't think you could make a clone of a 3Dfx card without getting assaulted by NVidia's lawyers.
Hopefully the internet will continue to be unsafe, filthy, and represent all that is wrong with our species as a whole. It makes things more interesting and certainly more entertaining.
Absolutely. Humanity needs some venue where we can let it all hang out, and be as filthy and anarchic as we like. Personally, I would prefer than venue to be the Internet, rather than, say, my neighborhood's streets.
For a good overview of Battle Angel Alita, see Vaz's take on the main character. You have an ideal killing machine, a cyborg supersoldier discarded after an ancient interplanetary war, who is resurrected with her old skills but without old memories, who fundamentally cannot reconcile her bloodlust with her kindness, who is repeatedly used, abused, and thrown away by greater powers and who accepts it because she has nothing to live for, who repeatedly finds
places that she could call home or people that understand her and then loses them immediately, who is forced to learn one-limbed fighting from frequent experience, and who
ultimately sacrifices herself to save the people that hate her guts.
Personally, I would be surprised if Cameron can do justice to the series.
Slashdot Mirror
They are fixed in gentoo-dev-sources-2.6.10-r1 (and the problems described by the grsecurity guys are fixed in gentoo-dev-sources-2.6.10-r4).
I'm running gentoo kernels on my debian boxes simply because gentoo kernel maintainers respond to security issues instantaneously.
I've run the same simulations on Xeons and Itaniums and the AMD's win hands down in terms of raw floating point throughput.
Spec2000 fp scores for Itanium 2's with 9M of cache are 2712/2712 sustained/max for a single CPU. For Athlon64 FX-55 (the top AMD chip, IIRC), the results are 1695/1878. Perhaps you are using old Itaniums? Or perhaps specfp is a really bad model for the computations you are interested in?
some of my tests have shown single A64's to out-perform dual Xeons rated similarly (2.8 Ghz Xeons vs. 2800+ A64's).
Sure. Xeons suck badly. But if you have the money, I believe Itaniums don't.
No, they are not making up the numbers. However, it takes some time to set up a top-ranking supercomputer (most of the time is not building the thing - you can just use college students paid in pizza - but debugging the inevitable problems in the network that arise as soon as you try putting any significant load on the system; I've worked next to a guy who spent months debugging a puny 12-box beowulf cluster, and problems are exponentially worse for the large supercomputers). The setup times are even worse for government and military machines - I suppose they need to meet strignent quality specs. As a result, many of the processors on top-500 are not state of the art (e.g. 1.25 GHz Alphas) so it might not the best place to look if you want to find out what CPU to use for a future supercomputer. If you notice, some of the highest scores are attained by IBM's BlueGene - which uses massive numbers of slow embedded PowerPC's, but comes out on top because of its excellent, fault-tolerant networking.
If I had infinite $ and a very big room, I would order 64 SGI Altix systems with 512 Itanium2's each, running SGI's custom Linux distro, and link those babies up in a big cluster. 32768 Itanium2's > 32768 PowerPC 440's.
and now with Intel's interest moving to a better 64-bit system[...]
The Itanium is a nice chip. If you want to build a supercomputer for floating-point-heavy scientific computations, and cost is not a factor, you really have two options: the Itanium, or a NEC vector processor. Plus, there are the aesthetics -- the EPIC architecure, the predicates, the totally new bios system.
It really is too bad that Intel never achieved economies of scale with the Itanium. Now, it looks that regular consumers are stuck with ugly upgrades to the ugly x86 architecture for the next century or so... Athlon-128 in 2020, anyone?
You need to read Stanislaw Lem 'Non Serviam'.
And the Strugatsky brothers' "Hard to be god".
Second, anyone who uses unencrypted email on a server they do no control, ESPECIALLY if it belongs to someone they are screwing, deserves to spend the rest of their productive years flipping burgers, or possibly stamping licence plates.
--
Gmail invites available for needy cases
Oh, the irony...
In my opinion Microsoft is essentially a state-sponsored monopoly and, as such, represents, little more than a tweaked version of the classic communist state entity.
It's a monopoly - but not a state-sponsored one. The worst you can accuse it of is using the usual lock-in practices to make sure some gov't departments keep on using windows. If you want to see a government-sponsored monopoly, look at something like AT&T in the 1970's, or your local cable provider, or your local power company before the days of deregulation. In all of these cases, there were laws protecting the monopolist from competition, and regulations that made the monopolist essentially an arm of the government. Microsoft, however, managed to achieve and hold on to monopoly status without government assistance - a monumental achievement (even though it's just as bad for the consumer and the industry as a traditional state-sponsored monopoly).
When we had communism in Poland, most shops had empty shelves -- and within just months after the communism's fall any shortages were just gone, as if by a wave of a magic wand.
That magic wand took a long time to work in Russia...
oes the license require me to distribute the layered gimp file along with my binaries?
The discussion on kde-look implies that for SVG images, the XML text is the source code, and the raster images produced from it are the object files. I would guess that similarly for 3D graphics, your model definition would be the source, since the render can't really be modified if you don't have the model.
However, I really don't know about raster-based images. An XCF or a PSD document is easier to use if you want to make changes, but it's certainly quite possible to just push pixels around, especially for something like a 48x48 icon. Arguably, a flat PNG is more human-editable than a PSD document (not everyone has access to photoshop). Consider a similar situation: you use a small perl script to auto-generate a part of your C code, but the resulting code is still perfectly human-readable and human-editable. Do you need to distribute the script if you use GPL? Is there a lawyer in the house?
>losetup -e aes256 /dev/loop0 /dev/hdb
/dev/loop0 /dev/hdb
At least for kernel 2.6, you want
>losetup -e aes-256
Russian TV is just about the most horrid and bleak torrent of dementia ever to spring from the mind of man.
Speaking as a Russian who has lived in Britain and the US, I can tell you that TV sucks planet-wide.
What happens if that galaxy we didn't see in our left blind spot doesn't use it's turn signal and rams right into us
Well, the only large galaxy scheduled to collide with us in the near future is Andromeda - and that collision is scheduled at least several hundred billion years from now. By then, we will either be extinct, or in control of the entire Local Cluster. If I were you, I really wouldn't worry about galactic collisions.
That's ... impressive. However, some of the optimizations look rather risky. E.g. shouldn't netmount wait for net.eth0 to successfully finish?
In Gentoo (well, at least in ~x86), just edit your /etc/conf.d/rc and set RC_PARALLEL_STARTUP="yes", and now at boot, your daemons start in parallel if possible. Don't know if other distros can do this yet - it looks like Gentoo's own invention.
Linux programmers won't silently tolerate incomplete API's with weird behavior where 3rd party Windows application developers must accept and learn "whatever Win32 does now".
/usr/src/linux/Documentation is old and obsolete, and the new documentation must be figured out from source, from mailing lists, or by asking the devs themselves.
I disagree. First, if you are using anything like Perl or Python, you essentially agree to accept and learn "whatever Larry, Damien, and Guido like at the moment". And their opinion tends to change with minor version numbers... However, at least Perl and Python are documented (and in the case of Python, readably so).
The bad side is that many parts of GNU/Linux not only have API's consisting of "whatever the dev likes at the moment" - the API isn't even documented! Case in point: can you figure out what the current fdo standard is for which applications to launch to open a file of a given mimetype? There are some webpages about the topic, but they are outdated. To figure out what's really going on, you need to read two mailing lists, and glean the API from between the lines. Scary.
But that's nothing compared to the kernel. To figure out the inotify api, you need to read Love's website. (how many kernel devs are there? do you know all their websites? how many of them do you want to check when you have a problem?) To figure out what's going on with scsi support this week, you need to diligently read the lkml and the fedora bugzilla (and the behavior of the scsi subsistem fluctuates a lot over time). Much of the documentation in
If there is one thing Microsoft can be proud of, it's good documentation, all organized in one place.
they patch the kernel to notify the system when files are modified.
That's because dnotify sucks donkey balls, as has been eloquently explained many times by anyone who has been forced to use that misbegotten kernel feature. If you want to have a clean file modification notification mechanism, you need inotify, which means either using kernel 2.6.8 or newer, or patching an older kernel.
Hard to tell. On the one hand, water vapor is a much more effective greenhouse gas than CO2, so lowering out CO2 output and increasing water vapor output may make the greenhouse effect worse. On the other hand, a very high concentration of H2O vapor would form a thick cloud layer, which would reflect solar radiation. On the third hand, the atmosphere can only hold so much humidity, so the extra H2O is likely to fall as precipitation, especially as snow in cold regions, which would change the distribution of albedo on Earth's surface and alter climates locally. On the fourth hand, humid air has a higher heat capacity than dry air, which also might alter the climate. On the fifth hand, a more humid atmosphere with less CO2 in it might change the distribution of plant life on the planet, further altering the atmosphere in some direction. Basically, there is no way to tell what will happen unless you set up a bunch of equations and plug them into the Earth Simulator.
If this technology is as good as Aurora claims, it can be used to implement a virtual ID card - just scan someone's face, and you can bring up their info from a database, no need for them to carry a piece of plastic around.
Obviously that's a privacy concern - but how can you regulate face recognition? It's fundamentally no different from having a live cop recognize your mug.
Sure, Turkey is a great country. However, this discussion is not about whether Turks are nice people or whether Turkey is a good place to live. The point of this discussion is that it is silly to arrest someone just for collecting (potentially) terrorist propaganda. If the guy was making bombs, or doing intelligence work for the PPK - sure, he deserves jail time. But arresting him just for managing a website?
Turkey's criminal justice system is quite good by, say, the standards of North Africa (there, Coskun would just "disappear"). However, it's clearly not good enough by the standards of EU...
the price of the FPGA (xc3s1500) which is quoted at Avnet at 70$US for the 320 pins version and 115$US for the 676 pins version (in qty of 25-99)...
Well, first Avnet charges a markup to stay in business; and second, I presume Tech Source wants to sell more than 99 of these babies. Xilinx quotes the price of an xc3s1500 as "under $20" for quantities of 250 000. Pulling some numbers out of my ass, I presume that Tech Source is going to make something like 10 000 of these cards, so they would be getting a big volume discount - maybe not $20 each, but certainly way lower than $70.
without dividers, perspective interpolation is going to be pretty tough
Perhaps there exists a cheap ASIC divider/trig unit (a 487?) that they can use as a coprocessor...
the usual suspects port there games to Linux - you know ID and Blizzard come to mind
Blizzard doesn't port their games to Linux - only to Mac, unfortunately. (I would buy a Linux version of WC3/TFT, if anyone from Blizzard is reading this...) Perhaps you are thinking of Epic?
What about 3DFX? [...] Would there be anything wrong with utilizing these old resources to achieve this goal?
NVidia bought 3Dfx and assimilated its intellectual property. I don't think you could make a clone of a 3Dfx card without getting assaulted by NVidia's lawyers.
It's better to have a finished product that meets a limited set of goals than an over-engineered design that never gets properly implemented...
Hopefully the internet will continue to be unsafe, filthy, and represent all that is wrong with our species as a whole. It makes things more interesting and certainly more entertaining.
Absolutely. Humanity needs some venue where we can let it all hang out, and be as filthy and anarchic as we like. Personally, I would prefer than venue to be the Internet, rather than, say, my neighborhood's streets.
For a good overview of Battle Angel Alita, see Vaz's take on the main character. You have an ideal killing machine, a cyborg supersoldier discarded after an ancient interplanetary war, who is resurrected with her old skills but without old memories, who fundamentally cannot reconcile her bloodlust with her kindness, who is repeatedly used, abused, and thrown away by greater powers and who accepts it because she has nothing to live for, who repeatedly finds places that she could call home or people that understand her and then loses them immediately, who is forced to learn one-limbed fighting from frequent experience, and who ultimately sacrifices herself to save the people that hate her guts.
Personally, I would be surprised if Cameron can do justice to the series.