Yawn. Yet more ranting from a paranoid lunatic. Okay, "mary_will_grow," you win. Business is evil. Businesses are trying to subjugate us and open up a new era of slavery to the corporations. There are no shades of gray here; business is just absolutely immoral and bad. We should all go back to being hunter-gatherers.
Just for the record, all USB devices I've ever encountered work equally well (or equally poorly, depending) connected directly to the computer or through a hub. The only exception arises when a USB device requires lots of power; if you chain too many power-hungry devices on the same bus, you can end up underpowering one or more of them. It's not a tragedy, it just means the device stops working until you give it more juice.
There are ways around this, too: powered USB hubs.
Well, according to Apple, the name of the OS is "Mac OS Ten". As in one more than OS 9.
Wrong. The name of the OS is "Mac OS X," which is pronounced "Mac OS Ten" because "X" is a Roman numeral. This is absolutely not "one more than OS 9." Why is this important?
Years ago, it was decided that Intel couldn't have a trademark on the expansion "80x86" or any integer x. They had "80286," "80386," and "80486," but they couldn't trademark those names, because they're just a sequential numbering. So Intel came up with an actual name-- as opposed to a number-- for their next major CPU release: Pentium. Ever since, we've had names-- trademarked names, protected by law-- for Intel's CPUs.
So the actual trademarked brand name of the operating system is "Mac OS X," with "X" being the Roman numeral for 10. That's not "as in one more than OS 9," because that would make it a serial number sequence and invalidate Apple's trademark claim.
The name of the operating system is separate and distinct from its version number. The name of the "classic" Mac operating system was simply "Mac OS." It had a version number, so the whole name would be "Mac OS 9.2.2," or simply "Mac OS 9" for short. Or even "OS 9" for shorter, but that was only acceptable in informal situations where everybody understood what you meant.
The name of this operating system is "Mac OS X," and its version number is (currently) 10.2.2. So if you want to say it all out loud and not be wrong, you say, "Mac OS X 10.2.2." And yeah, smart ass, the "X" is a numeral, so it's pronounced "ten."
If this is too complicated for you to understand, or if it just offends your sensibilities in some undefinable way, just keep calling it "Jaguar." That's a perfectly good name for it too. Either way, shut up about it, okay? If you continue to protest that you're unable to grasp this simple idea, you're going to start looking like an idiot. Just some friendly advice.
Yes, that's just what I said. Should I have made it explicitly clear that the "X" in "Mac OS X" is a Roman numeral? When you say it out loud, it's pronounced "Mac oh-ess ten." I just sort of assumed that everybody on Slashdot was familiar with the concept of Roman numerals. Maybe that was going too far, based on some of the replies I've gotten.
D'oh! Of course, you're right. I was talking about case-mapping, but I spoke so generally that my comment applied equally well to sorting, and that's wrong. Of course you want to be able to sort Japanese characters in dictionary order, and you (may or may not) want words starting with hiragana "ka" to appear next to words starting with katakana "ka." For that purpose, you have to be able to sort the two character sets together. But strictly speaking, a name spelled in katakana characters is different from a name spelled in hiragana characters, even if it's made up of the same syllables. So you wouldn't want to "case-fold" hiragana into katakana or vice versa.
The right answer here depends on a lot of stuff that you (the submitter, that's right, I'm talking to you) didn't tell us.
What's your friend's level of comfort with UNIX? IRIX is a very well documented OS, but there are few places you can go for help if you get stuck. You have to be prepared to sit down and read the manuals (techpubs.sgi.com).
Does your friend want to do anything else with the computer? Like, say, surf the web or read email? If so, he'll be happier with either a Windows XP machine (and you said he's "fed up") or a Mac. Doing basic stuff on an IRIX box-- like word processing or email-- is like chewing on tin foil. Most people I know keep a PC or a Mac next to their SGI system for doing everything other than what they use their SGI system for.
Is a new machine the right answer for this guy? You can get used Octanes pretty reasonably. Fuel comes with V12 graphics (I think V10 is also an option, but I'm not positive), and that's pretty serious stuff, but you can probably get your hands on an MXE or similar system for considerably less than a new Fuel. Consider contacting SGI's remanufacturing division; they sell older systems at a substantially reduced cost. Although I don't know if they sell to overseas customers. Might be US-only.
All in all, I think the best choice is probably going to be a top-o-the-line Power Mac G4 with the fastest graphics card available. Right now, I think that's the GeForce 4 Titanium, for about $400. It's no V12, but it'll do the job. The dual processors are a nice bonus, letting you work at full speed while doing test renders in the background (I use that feature all the time on my dual 1 GHz "Speed Holes" model.) And because OS X is UNIX in all but name, you get all the advantages of running UNIX on your desktop while still being able to run stuff like Microsoft Office should you need it.
I guess what I'm saying is that you should think about the questions I asked, carefully weigh all the factors, and then buy a Mac.;-)
Your experience is not typical. In the absence of a serious hardware fault, IRIX never crashes. And the only way you can "screw something over so bad that you'd need to reinstall the OS" is by doing something incredibly dumb. My personal favorite happened once when I was trying to change the ownership of an entire directory of files. I typed "sudo chown foo.foo." The only thing is, the period and the slash keys are, like, right next to each other. So I ended up typing "sudo chown foo.foo/". By the time I slammed control-C, it had already made it through/bin and was working on/dev. I could have fixed it, but it was just easier to boot miniroot and reinstall the base OS pieces to reset the permissions and ownerships.
Contrary to your opinion, IRIX is one of the most stable and friendly OSs out there. Oh, and it's not "UNIX-based." It's UNIX. SGI has licensed the UNIX trademark from The Open Group for IRIX, so it's a full-fledged UNIX operating system.
If you're writing Turkish, you're going to be using the ISO-8859-9 character mapping. The capital letter corresponding to the lowercase "i" (U+0069) is U+0130 ("LATIN CAPITAL LETTER I WITH DOT ABOVE"), while the capital letter corresponding to lowercase U+0131, "LATIN SMALL LETTER DOTLESS I," is I (U+0049).
Because your system will chose a character mapping for displaying and sorting strings based on your locale setting, if your locale is set incorrectly, string comparisons for case and sorting won't be done properly.
I'm not a native speaker of Japanese or Chinese, but I'm a little educated in both. Hiragana and katakana are different alphabets with different purposes. The katakana character for the syllable "ka" is not equivalent to the hiragana character for the syllable "ka." There's no circumstance under which you'd want corresponding hiragana and katakana characters to be considered equal.
As far as Chinese goes, traditional Chinese characters are used in Taiwan, while simplified Chinese characters are used in mainland China. Again, they're not equivalent. So you wouldn't need or want to map between them.
The uppercase-lowercase thing is pretty much unique to Latin and Latin-derived alphabets. Some languages have contextual forms-- for example, an initial character in Arabic looks different from the same character in medial or final position in the word-- but that's a rendering issue, not an encoding issue.
I actually think it would be quite straightforward to design a Unicode-based system that's case-insensitive with respect to alphabets that have distinct cases. More work than doing so for ASCII, of course, but not insurmountably more.
They should just scrap the term hacker and call him a terrorist, because thats what breaking into the US millitary is, terrorism.
The term "terrorist" has certainly been overused in the past year or so, but what many people don't realize is that it actually has a strict legal definition. (Well, actually several strict legal definitions, depending on the jurisdiction you're paying attention to at the time.)
Way back in 1937, the League of Nations defined terrorism as, "All criminal acts directed against a State and intended or calculated to create a state of terror in the minds of particular persons or a group of persons or the general public." So under that definition, an act is terrorism only if it's specifically intended to create a state of terror. September 11, yes. This guy, no.
In 1999, the UN defined terrorism this way: "Reiterates that criminal acts intended or calculated to provoke a state of terror in the general public, a group of persons or particular persons for political purposes are in any circumstance unjustifiable, whatever the considerations of a political, philosophical, ideological, racial, ethnic, religious or other nature that may be invoked to justify them." So here to we have the idea that the act must be specifically intended to invoke a feeling of terror. So by that definition, too, this incident is not terrorism.
The USDOD defines terrorism to be, "The calculated use of violence or the threat of violence to inculcate fear; intended to coerce or to intimidate governments or societies in the pursuit of goals that are generally political, religious, or ideological." Once again we have the idea that the act must be calculated to cause fear. If an act merely incidentally causes fear or terror, it's not strictly terrorism.
Since 9/11, laws have sprung up in several US jurisdictions making it a crime to plan, enact, or carry out any act designed to produce a fear response in the population. In fact, the DC sniper suspects are being indicted in Maryland under just such a law. But all of these also have the same basic thread: that the act must have been done with the specific and deliberate intent of causing fear.
So no, what this loser did isn't technically terrorism.
At worst, it could be considered an act of war from the country where the hacker originated against the country that was hacked.
Not really. In order to make the leap from crime to act of war, there has to be an element of direct or indirect state sponsorship. An individual acting on his own to carry out a criminal act-- even a horrible or devastating one-- in another country does not automatically constitute an act of war. But if another government sponsors the act, that's a different story. The basic idea here is that war is a state of armed conflict between nations, not between groups or individuals. Rhetorical shorthand aside, the United States could never be in a state of war against al Qaeda, or against Osama bin Laden personally. The concept of war can't be applied to those sorts of conflicts in any meaningful way.
Yeah, I was actually corrected on this matter in a private email just a little while ago. The product known to the world as CrayLink originated at SGI, not at Cray at all. My time with SGI started in '96, which was after the acquisition, so I was just repeating what I was told. My bad.
In light of that revelation, I have to agree. It seems like SGI didn't really get much, if anything, out of the Cray acquisition. Except those cool purple tee shirts that said, "My other computer is a Cray." I've still got mine around here someplace....
no, I don't buy the post hoc redefinition of the acronym
Not to be rude, but so what? I don't recall anybody asking you if you agreed with the current common usage of the term "RAID." It means what it means: a number of hard drives concatenated, striped, or otherwise combined to behave as a single logical device. Whether the combination of drives is redundant or not, or whether the individual drives is inexpensive or not, isn't important these days. The various implementations of RAID have been well defined, and are understood by everybody in the field. Quibbling over definitions that have long since been rendered moot is a waste of time.
In the future, i claim that software and faster interconnect hardware (be it some new ethernet or myrinet or something similar) will let you do the same on commodity hardware.
That's fine. But unless something drastic happens, this "in the future" scenario won't happen for many years-- even the fastest external interconnect is a long way from CrayLink/NUMAlink's specs. Assuming that we're both right, that it will happen eventually but not anytime soon, then by the time your scenario comes to pass, SGI or some other supercomputer manufacturer will have responded to customer needs by making some other large-scale single-system-image implementation that is to this new mystery interconnect as CrayLink was to 100BASE-T in 1995.
Basically, you've got to spend 33% more on your storage to use 0+1 instead of RAID 5.
No, you've got it backwards. To use RAID 5, you'll have to spend up to 33% more on disks than you would if you used no RAID protection at all. If you bought two 100 GB drives (to keep the numbers round) you'd spend $200 (again to keep the numbers round). To RAID-protect those two 100 GB drives with RAID-3 or RAID-5, you'd have to buy one more 100 GB drive, meaning you spend $300 to buy 200 GB of storage. You spent 33% more.
To create a RAID-0+1 set out of those two drives, you'd have two choices. You could just mirror the drives by themselves, and have 100 GB of storage space for $200-- 100% more money than you would have spent if you'd just bought one 100 GB disk. Or, if you have to have 200 GB of usable space, you could buy two more 100 GB drives and create a RAID-0+1 set out of them, giving you 200 GB of space for $400-- again, 100% more than you would have spent for just the two disks alone.
So the difference in cost between RAID-3/5 and RAID-0+1 starts at 66% above the cost of the data disks themselves. But you can create a RAID-3 or a RAID-5 set out of any number of drives. You could have bought ten 100 GB drives, build a RAID-3 or a RAID-5 set out of them, had 900 GB of usable space, and only spent 10% more on storage than you would have spent without RAID protection. (The other way of looking at it is that you sacrifice a fraction of your usable storage up to 33% when using RAID-3 or RAID-5. Either way, the fraction never exceeds 33% unless you start counting hot spares, mirrored parity drives [for RAID-3], and things of that nature.)
On the other hand, to use RAID 0+1, you have to have one mirror disk for every data disk, meaning you have no choice at all but to spend 100% more on storage than you otherwise would have. The difference in cost between RAID-3/5 and RAID-0+1 starts at 66% of the value of your usable storage and goes up from there.
RAID-3/5 actually makes more sense in all applications that aren't seriously affected by storage performance. RAID-3/5 are inherently slower than RAID 1 because they involve the generation of parity data, and that takes some measurable amount of time during writes. Depending on your RAID software, you may do a parity check on reads as well as a preventative measure, and in that case your reads slow down as well. But RAID-3/5 will always be less expensive than RAID-1 for four drives or more. (You can't do RAID-3/5 with two drives, and you can't do RAID-1 with three drives, so one has to start comparing them at four.)
Now, in some specific cases, RAID-0+1 gives you a level of functionality you can't get with RAID-3/5, and it makes sense to use it for that reason. For example, in an enterprise environment with (say) 20 TB of data in a big array, I would consider setting up a three-way stripe/mirror set. That way, you can detach one of the mirrors and do an offline back-up to tape or whatever, while the production filesystem continues to run with its other mirror protecting it. When the backup or whatever is complete, simply reattach the offline mirror and re-synchronize it. Storage arrays like the HDS 9900 series use this sort of trick all the time, and in that environment it makes a lot of sense.
But absent of outside factors, just considering pure cost of investment alone, RAID-3/5 will always be cheaper than RAID-0+1 for four drives or more.
Let's be clear. These are not clusters. These are single system images. Clusters are composed of a number of separate computers-- each running its own kernel, with its own address space and its own storage-- that can be used together for certain purposes. The computers you're talking about at NASA are single system images, with one running kernel and one address space. A program running on a 1,024-processor system can use one processor, or some, or all 1,024 at the same time merely by spawning sprocs or pthreads, or by calling library code that spawns sprocs or pthreads.
The chief advantage to the developer of programming for a single system image rather than a cluster is that one doesn't have to use an abstraction tool like PVM or MPI to parallelize one's code. The other big advantage is that single system images are always faster than clusters of identical capacity and processor count.
This is such a dead horse. The name of the operating system is "Mac OS X." That's the brand name. The version number is currently 10.2.2. When enough time has passed, the version number will be 11.something. At that time, the full name and version of the OS will be "Mac OS X 11.something."
It's really not that hard to wrap your head around this idea, y'all. It's not necessary to make a lot of noise about it every single time OS X comes up on Slashdot.
With the low cost of storage these days, RAID 5 is basically obsolete. Spend the extra few gigabytes, and use RAID 0+1
That's an overstatement. ATA/IDE/whatever storage is pretty cheap, but SCSI and Fibre Channel disks are still pricey. In order to protect a 1 TB filesystem with RAID 0+1, you'd have to have 2 TB worth of (let's say) Fibre Channel drives. That extra terabyte would cost you many thousands of dollars. But to protect the same filesystem with RAID 3 or RAID 5, you only have to have (at least) one spare drive. That's a lot cheaper than the 6 or 8 or 16 or whatever drives you'd have to buy to mirror the whole filesystem.
I'd say that for filesystems in the range of 0-500 GB using inexpensive disks, RAID 3 and RAID 5 are probably unnecessary. But outside that set of conditions, RAID 0+1 just isn't practical.
I'm not going to get into an argument about processors. Your assertion that the R10000 needs to be retired is unfounded and bogus, and not even worth discussing.
The much vaunted scalability of these systems is also quite questionable.
Okay, you're kind of talking out of your ass now. I've personally seen code slam a 768-processor machine. (It was nothing more complex than an image processing demonstration. SGI's ImageVision software library is hand-coded to parallelize across all available processors. Run ImageVision on a machine with 8 processors and you'll see it run on all 8. Run it on a machine with 768 and you'll see it run on all 768. In the demo, the program did a 5x5 convolution on an image of truly gigantic proportions, 100K by 100K pixels, or something.) Companies like SGI build giant-- or what we would consider giant-- computers because people need them.
MPI (which will run very nicely on a cluster), can get you into the hundreds of CPUs of scalability.
Maybe. But with a single system image, you can scale to tens of thousands of CPUs using nothing more complex than sprocs or pthreads. And you don't get shot in the butt by inter-node bandwidth or latency shortcomings.
Just because you can add processors, doesnt mean the system "scales". Scaling is one of those fast and furious marketing terms that has an actual meaning attached to it, but you have to wipe away the marketing fluff to find it.
I hear what you're saying. You seem to be implying that this system doesn't scale, by your definition. Yet you fail to explain what that definition is, exactly. Indicting the whole concept of single-system scalability isn't going to earn you any points here.
No, don't bother doing that comparison. The 12.8 GB/s figure is the aggregate memory bandwidth across the system topology. If you want an idea of how that translates into real-world numbers, find an Origin 3900 and run the STREAM TRIAD memory benchmark. I'm making an educated guess here, but I'd expect to see a result around or above the 100 GB/s mark, based on the results that SGI has published from similar systems to this one.
The change you've noticed is due in no small part to a shake-up at the very highest levels of management. Rick Belluzzo, while he was a decent enough guy personally, really had some screwed-up ideas about what the company ought to be doing. He saw SGI's competitors as being Dell and Compaq and, to a certain extent, Apple. In fact, SGI's competitors were, and are, companies like NEC and, in a way, Cray itself. Rick thought too small, and nearly killed the company in the process.
You can't just flippy-floppy words around and call the result the converse. That trick only works when the proposition is in the form of an "if-then" statement. The converse of "if x then y" is "if y then x."
This is, of course, aside from the fact that I was obviously using the expression "the converse is also true" in the colloquial sense, not the Boolean. Dork.
Now that the broadcasters have their hands on new full 6-MHz channels (free of charge), all this HDTV silliness will be forgotten....
Pure FUD. That's not what's happening at all. In fact, just the opposite is starting to unfold. The major networks-- ABC, NBC, CBS, Fox, PBS, UPN, WB-- are rolling out more HDTV content this season than ever before. I don't have the numbers in front of me, but there's some prime-time network programming available in HDTV every night of the week. And, with the sole exception of Fox, we're talking about high-bandwidth 1080i or 720p programming here. What you're describing could happen, but it hasn't so far, and there are no signs that it will, so talking about it is pure speculation.
Slashdot Mirror
Yawn. Yet more ranting from a paranoid lunatic. Okay, "mary_will_grow," you win. Business is evil. Businesses are trying to subjugate us and open up a new era of slavery to the corporations. There are no shades of gray here; business is just absolutely immoral and bad. We should all go back to being hunter-gatherers.
You first. I'll be right behind you.
Just for the record, all USB devices I've ever encountered work equally well (or equally poorly, depending) connected directly to the computer or through a hub. The only exception arises when a USB device requires lots of power; if you chain too many power-hungry devices on the same bus, you can end up underpowering one or more of them. It's not a tragedy, it just means the device stops working until you give it more juice.
There are ways around this, too: powered USB hubs.
Well, according to Apple, the name of the OS is "Mac OS Ten". As in one more than OS 9.
Wrong. The name of the OS is "Mac OS X," which is pronounced "Mac OS Ten" because "X" is a Roman numeral. This is absolutely not "one more than OS 9." Why is this important?
Years ago, it was decided that Intel couldn't have a trademark on the expansion "80x86" or any integer x. They had "80286," "80386," and "80486," but they couldn't trademark those names, because they're just a sequential numbering. So Intel came up with an actual name-- as opposed to a number-- for their next major CPU release: Pentium. Ever since, we've had names-- trademarked names, protected by law-- for Intel's CPUs.
So the actual trademarked brand name of the operating system is "Mac OS X," with "X" being the Roman numeral for 10. That's not "as in one more than OS 9," because that would make it a serial number sequence and invalidate Apple's trademark claim.
The name of the operating system is separate and distinct from its version number. The name of the "classic" Mac operating system was simply "Mac OS." It had a version number, so the whole name would be "Mac OS 9.2.2," or simply "Mac OS 9" for short. Or even "OS 9" for shorter, but that was only acceptable in informal situations where everybody understood what you meant.
The name of this operating system is "Mac OS X," and its version number is (currently) 10.2.2. So if you want to say it all out loud and not be wrong, you say, "Mac OS X 10.2.2." And yeah, smart ass, the "X" is a numeral, so it's pronounced "ten."
If this is too complicated for you to understand, or if it just offends your sensibilities in some undefinable way, just keep calling it "Jaguar." That's a perfectly good name for it too. Either way, shut up about it, okay? If you continue to protest that you're unable to grasp this simple idea, you're going to start looking like an idiot. Just some friendly advice.
Yes, that's just what I said. Should I have made it explicitly clear that the "X" in "Mac OS X" is a Roman numeral? When you say it out loud, it's pronounced "Mac oh-ess ten." I just sort of assumed that everybody on Slashdot was familiar with the concept of Roman numerals. Maybe that was going too far, based on some of the replies I've gotten.
D'oh! Of course, you're right. I was talking about case-mapping, but I spoke so generally that my comment applied equally well to sorting, and that's wrong. Of course you want to be able to sort Japanese characters in dictionary order, and you (may or may not) want words starting with hiragana "ka" to appear next to words starting with katakana "ka." For that purpose, you have to be able to sort the two character sets together. But strictly speaking, a name spelled in katakana characters is different from a name spelled in hiragana characters, even if it's made up of the same syllables. So you wouldn't want to "case-fold" hiragana into katakana or vice versa.
The right answer here depends on a lot of stuff that you (the submitter, that's right, I'm talking to you) didn't tell us.
;-)
What's your friend's level of comfort with UNIX? IRIX is a very well documented OS, but there are few places you can go for help if you get stuck. You have to be prepared to sit down and read the manuals (techpubs.sgi.com).
Does your friend want to do anything else with the computer? Like, say, surf the web or read email? If so, he'll be happier with either a Windows XP machine (and you said he's "fed up") or a Mac. Doing basic stuff on an IRIX box-- like word processing or email-- is like chewing on tin foil. Most people I know keep a PC or a Mac next to their SGI system for doing everything other than what they use their SGI system for.
Is a new machine the right answer for this guy? You can get used Octanes pretty reasonably. Fuel comes with V12 graphics (I think V10 is also an option, but I'm not positive), and that's pretty serious stuff, but you can probably get your hands on an MXE or similar system for considerably less than a new Fuel. Consider contacting SGI's remanufacturing division; they sell older systems at a substantially reduced cost. Although I don't know if they sell to overseas customers. Might be US-only.
All in all, I think the best choice is probably going to be a top-o-the-line Power Mac G4 with the fastest graphics card available. Right now, I think that's the GeForce 4 Titanium, for about $400. It's no V12, but it'll do the job. The dual processors are a nice bonus, letting you work at full speed while doing test renders in the background (I use that feature all the time on my dual 1 GHz "Speed Holes" model.) And because OS X is UNIX in all but name, you get all the advantages of running UNIX on your desktop while still being able to run stuff like Microsoft Office should you need it.
I guess what I'm saying is that you should think about the questions I asked, carefully weigh all the factors, and then buy a Mac.
Your experience is not typical. In the absence of a serious hardware fault, IRIX never crashes. And the only way you can "screw something over so bad that you'd need to reinstall the OS" is by doing something incredibly dumb. My personal favorite happened once when I was trying to change the ownership of an entire directory of files. I typed "sudo chown foo.foo ." The only thing is, the period and the slash keys are, like, right next to each other. So I ended up typing "sudo chown foo.foo /". By the time I slammed control-C, it had already made it through /bin and was working on /dev. I could have fixed it, but it was just easier to boot miniroot and reinstall the base OS pieces to reset the permissions and ownerships.
Contrary to your opinion, IRIX is one of the most stable and friendly OSs out there. Oh, and it's not "UNIX-based." It's UNIX. SGI has licensed the UNIX trademark from The Open Group for IRIX, so it's a full-fledged UNIX operating system.
all of my PCI slots are currently filled (second ethernet card, four port USB card, and SCSI card).
You know, there are these neat things now called USB hubs. You don't need to add PCI cards to add USB devices any more. Just thought you'd wanna know.
If you're writing Turkish, you're going to be using the ISO-8859-9 character mapping. The capital letter corresponding to the lowercase "i" (U+0069) is U+0130 ("LATIN CAPITAL LETTER I WITH DOT ABOVE"), while the capital letter corresponding to lowercase U+0131, "LATIN SMALL LETTER DOTLESS I," is I (U+0049).
Because your system will chose a character mapping for displaying and sorting strings based on your locale setting, if your locale is set incorrectly, string comparisons for case and sorting won't be done properly.
I'm not a native speaker of Japanese or Chinese, but I'm a little educated in both. Hiragana and katakana are different alphabets with different purposes. The katakana character for the syllable "ka" is not equivalent to the hiragana character for the syllable "ka." There's no circumstance under which you'd want corresponding hiragana and katakana characters to be considered equal.
As far as Chinese goes, traditional Chinese characters are used in Taiwan, while simplified Chinese characters are used in mainland China. Again, they're not equivalent. So you wouldn't need or want to map between them.
The uppercase-lowercase thing is pretty much unique to Latin and Latin-derived alphabets. Some languages have contextual forms-- for example, an initial character in Arabic looks different from the same character in medial or final position in the word-- but that's a rendering issue, not an encoding issue.
I actually think it would be quite straightforward to design a Unicode-based system that's case-insensitive with respect to alphabets that have distinct cases. More work than doing so for ASCII, of course, but not insurmountably more.
They should just scrap the term hacker and call him a terrorist, because thats what breaking into the US millitary is, terrorism.
The term "terrorist" has certainly been overused in the past year or so, but what many people don't realize is that it actually has a strict legal definition. (Well, actually several strict legal definitions, depending on the jurisdiction you're paying attention to at the time.)
Way back in 1937, the League of Nations defined terrorism as, "All criminal acts directed against a State and intended or calculated to create a state of terror in the minds of particular persons or a group of persons or the general public." So under that definition, an act is terrorism only if it's specifically intended to create a state of terror. September 11, yes. This guy, no.
In 1999, the UN defined terrorism this way: "Reiterates that criminal acts intended or calculated to provoke a state of terror in the general public, a group of persons or particular persons for political purposes are in any circumstance unjustifiable, whatever the considerations of a political, philosophical, ideological, racial, ethnic, religious or other nature that may be invoked to justify them." So here to we have the idea that the act must be specifically intended to invoke a feeling of terror. So by that definition, too, this incident is not terrorism.
The USDOD defines terrorism to be, "The calculated use of violence or the threat of violence to inculcate fear; intended to coerce or to intimidate governments or societies in the pursuit of goals that are generally political, religious, or ideological." Once again we have the idea that the act must be calculated to cause fear. If an act merely incidentally causes fear or terror, it's not strictly terrorism.
Since 9/11, laws have sprung up in several US jurisdictions making it a crime to plan, enact, or carry out any act designed to produce a fear response in the population. In fact, the DC sniper suspects are being indicted in Maryland under just such a law. But all of these also have the same basic thread: that the act must have been done with the specific and deliberate intent of causing fear.
So no, what this loser did isn't technically terrorism.
At worst, it could be considered an act of war from the country where the hacker originated against the country that was hacked.
Not really. In order to make the leap from crime to act of war, there has to be an element of direct or indirect state sponsorship. An individual acting on his own to carry out a criminal act-- even a horrible or devastating one-- in another country does not automatically constitute an act of war. But if another government sponsors the act, that's a different story. The basic idea here is that war is a state of armed conflict between nations, not between groups or individuals. Rhetorical shorthand aside, the United States could never be in a state of war against al Qaeda, or against Osama bin Laden personally. The concept of war can't be applied to those sorts of conflicts in any meaningful way.
Yeah, I was actually corrected on this matter in a private email just a little while ago. The product known to the world as CrayLink originated at SGI, not at Cray at all. My time with SGI started in '96, which was after the acquisition, so I was just repeating what I was told. My bad.
In light of that revelation, I have to agree. It seems like SGI didn't really get much, if anything, out of the Cray acquisition. Except those cool purple tee shirts that said, "My other computer is a Cray." I've still got mine around here someplace....
Thanks for correcting me, Angus.
no, I don't buy the post hoc redefinition of the acronym
Not to be rude, but so what? I don't recall anybody asking you if you agreed with the current common usage of the term "RAID." It means what it means: a number of hard drives concatenated, striped, or otherwise combined to behave as a single logical device. Whether the combination of drives is redundant or not, or whether the individual drives is inexpensive or not, isn't important these days. The various implementations of RAID have been well defined, and are understood by everybody in the field. Quibbling over definitions that have long since been rendered moot is a waste of time.
In the future, i claim that software and faster interconnect hardware (be it some new ethernet or myrinet or something similar) will let you do the same on commodity hardware.
That's fine. But unless something drastic happens, this "in the future" scenario won't happen for many years-- even the fastest external interconnect is a long way from CrayLink/NUMAlink's specs. Assuming that we're both right, that it will happen eventually but not anytime soon, then by the time your scenario comes to pass, SGI or some other supercomputer manufacturer will have responded to customer needs by making some other large-scale single-system-image implementation that is to this new mystery interconnect as CrayLink was to 100BASE-T in 1995.
It's a circle-of-life thing.
Basically, you've got to spend 33% more on your storage to use 0+1 instead of RAID 5.
No, you've got it backwards. To use RAID 5, you'll have to spend up to 33% more on disks than you would if you used no RAID protection at all. If you bought two 100 GB drives (to keep the numbers round) you'd spend $200 (again to keep the numbers round). To RAID-protect those two 100 GB drives with RAID-3 or RAID-5, you'd have to buy one more 100 GB drive, meaning you spend $300 to buy 200 GB of storage. You spent 33% more.
To create a RAID-0+1 set out of those two drives, you'd have two choices. You could just mirror the drives by themselves, and have 100 GB of storage space for $200-- 100% more money than you would have spent if you'd just bought one 100 GB disk. Or, if you have to have 200 GB of usable space, you could buy two more 100 GB drives and create a RAID-0+1 set out of them, giving you 200 GB of space for $400-- again, 100% more than you would have spent for just the two disks alone.
So the difference in cost between RAID-3/5 and RAID-0+1 starts at 66% above the cost of the data disks themselves. But you can create a RAID-3 or a RAID-5 set out of any number of drives. You could have bought ten 100 GB drives, build a RAID-3 or a RAID-5 set out of them, had 900 GB of usable space, and only spent 10% more on storage than you would have spent without RAID protection. (The other way of looking at it is that you sacrifice a fraction of your usable storage up to 33% when using RAID-3 or RAID-5. Either way, the fraction never exceeds 33% unless you start counting hot spares, mirrored parity drives [for RAID-3], and things of that nature.)
On the other hand, to use RAID 0+1, you have to have one mirror disk for every data disk, meaning you have no choice at all but to spend 100% more on storage than you otherwise would have. The difference in cost between RAID-3/5 and RAID-0+1 starts at 66% of the value of your usable storage and goes up from there.
RAID-3/5 actually makes more sense in all applications that aren't seriously affected by storage performance. RAID-3/5 are inherently slower than RAID 1 because they involve the generation of parity data, and that takes some measurable amount of time during writes. Depending on your RAID software, you may do a parity check on reads as well as a preventative measure, and in that case your reads slow down as well. But RAID-3/5 will always be less expensive than RAID-1 for four drives or more. (You can't do RAID-3/5 with two drives, and you can't do RAID-1 with three drives, so one has to start comparing them at four.)
Now, in some specific cases, RAID-0+1 gives you a level of functionality you can't get with RAID-3/5, and it makes sense to use it for that reason. For example, in an enterprise environment with (say) 20 TB of data in a big array, I would consider setting up a three-way stripe/mirror set. That way, you can detach one of the mirrors and do an offline back-up to tape or whatever, while the production filesystem continues to run with its other mirror protecting it. When the backup or whatever is complete, simply reattach the offline mirror and re-synchronize it. Storage arrays like the HDS 9900 series use this sort of trick all the time, and in that environment it makes a lot of sense.
But absent of outside factors, just considering pure cost of investment alone, RAID-3/5 will always be cheaper than RAID-0+1 for four drives or more.
Let's be clear. These are not clusters. These are single system images. Clusters are composed of a number of separate computers-- each running its own kernel, with its own address space and its own storage-- that can be used together for certain purposes. The computers you're talking about at NASA are single system images, with one running kernel and one address space. A program running on a 1,024-processor system can use one processor, or some, or all 1,024 at the same time merely by spawning sprocs or pthreads, or by calling library code that spawns sprocs or pthreads.
The chief advantage to the developer of programming for a single system image rather than a cluster is that one doesn't have to use an abstraction tool like PVM or MPI to parallelize one's code. The other big advantage is that single system images are always faster than clusters of identical capacity and processor count.
This is such a dead horse. The name of the operating system is "Mac OS X." That's the brand name. The version number is currently 10.2.2. When enough time has passed, the version number will be 11.something. At that time, the full name and version of the OS will be "Mac OS X 11.something."
It's really not that hard to wrap your head around this idea, y'all. It's not necessary to make a lot of noise about it every single time OS X comes up on Slashdot.
With the low cost of storage these days, RAID 5 is basically obsolete. Spend the extra few gigabytes, and use RAID 0+1
That's an overstatement. ATA/IDE/whatever storage is pretty cheap, but SCSI and Fibre Channel disks are still pricey. In order to protect a 1 TB filesystem with RAID 0+1, you'd have to have 2 TB worth of (let's say) Fibre Channel drives. That extra terabyte would cost you many thousands of dollars. But to protect the same filesystem with RAID 3 or RAID 5, you only have to have (at least) one spare drive. That's a lot cheaper than the 6 or 8 or 16 or whatever drives you'd have to buy to mirror the whole filesystem.
I'd say that for filesystems in the range of 0-500 GB using inexpensive disks, RAID 3 and RAID 5 are probably unnecessary. But outside that set of conditions, RAID 0+1 just isn't practical.
I'm not going to get into an argument about processors. Your assertion that the R10000 needs to be retired is unfounded and bogus, and not even worth discussing.
The much vaunted scalability of these systems is also quite questionable.
Okay, you're kind of talking out of your ass now. I've personally seen code slam a 768-processor machine. (It was nothing more complex than an image processing demonstration. SGI's ImageVision software library is hand-coded to parallelize across all available processors. Run ImageVision on a machine with 8 processors and you'll see it run on all 8. Run it on a machine with 768 and you'll see it run on all 768. In the demo, the program did a 5x5 convolution on an image of truly gigantic proportions, 100K by 100K pixels, or something.) Companies like SGI build giant-- or what we would consider giant-- computers because people need them.
MPI (which will run very nicely on a cluster), can get you into the hundreds of CPUs of scalability.
Maybe. But with a single system image, you can scale to tens of thousands of CPUs using nothing more complex than sprocs or pthreads. And you don't get shot in the butt by inter-node bandwidth or latency shortcomings.
Just because you can add processors, doesnt mean the system "scales". Scaling is one of those fast and furious marketing terms that has an actual meaning attached to it, but you have to wipe away the marketing fluff to find it.
I hear what you're saying. You seem to be implying that this system doesn't scale, by your definition. Yet you fail to explain what that definition is, exactly. Indicting the whole concept of single-system scalability isn't going to earn you any points here.
No, don't bother doing that comparison. The 12.8 GB/s figure is the aggregate memory bandwidth across the system topology. If you want an idea of how that translates into real-world numbers, find an Origin 3900 and run the STREAM TRIAD memory benchmark. I'm making an educated guess here, but I'd expect to see a result around or above the 100 GB/s mark, based on the results that SGI has published from similar systems to this one.
It must take a lot to cool it, which would make it pretty loud.
;-)
Yes, it's pretty loud. It's also six feet tall and bright purple. Wanna talk about what it smells like?
Some people have an uncanny way of zeroing in on the most irrelevant aspects of things.
and other Bigass(tm) vector machines
Heh. When I first scanned this, I read it as "BigBLAS(tm)." Funny thing is, either meaning makes perfect sense.
The change you've noticed is due in no small part to a shake-up at the very highest levels of management. Rick Belluzzo, while he was a decent enough guy personally, really had some screwed-up ideas about what the company ought to be doing. He saw SGI's competitors as being Dell and Compaq and, to a certain extent, Apple. In fact, SGI's competitors were, and are, companies like NEC and, in a way, Cray itself. Rick thought too small, and nearly killed the company in the process.
You can't just flippy-floppy words around and call the result the converse. That trick only works when the proposition is in the form of an "if-then" statement. The converse of "if x then y" is "if y then x."
This is, of course, aside from the fact that I was obviously using the expression "the converse is also true" in the colloquial sense, not the Boolean. Dork.
Now that the broadcasters have their hands on new full 6-MHz channels (free of charge), all this HDTV silliness will be forgotten....
Pure FUD. That's not what's happening at all. In fact, just the opposite is starting to unfold. The major networks-- ABC, NBC, CBS, Fox, PBS, UPN, WB-- are rolling out more HDTV content this season than ever before. I don't have the numbers in front of me, but there's some prime-time network programming available in HDTV every night of the week. And, with the sole exception of Fox, we're talking about high-bandwidth 1080i or 720p programming here. What you're describing could happen, but it hasn't so far, and there are no signs that it will, so talking about it is pure speculation.