True, OED was his first civilian gig after the War. I should have just left it as "his well known gig". Certainly his main gig was writing LotR, especially as reflected in his final net worth (AFAIK).
the difference between Dune and Dune: House Atriedes
Good analogy. The difference between, say, The Fellowship of the Ring and any Christopher Tolkien followup (except perhaps the Silmarillion) is about as big.
JRR Tolkien and Frank Herbert were visionaries. Their books are legendary because they're so complete, so consistent, they're practically holographic. While those authors were also brilliant editors, especially Tolkien whose main gig was (as is well known) Oxford English Dictionary editor. Their (genetic, and thereby literary) heirs are undistinguished from a vast host of other second or lower tier of "visionary" authors, and have no special editing talent - nor have acquired any at their cashin publishers. While they also operate at a disadvantage while writing outside the original cultural contexts that produced those seminal works for a different audience.
Ironically, both Middle Earth and Dune are epic tales of the original forefathers of our times (Dune less obviously, sorry for the spoiler). A magical time when a unique individual arrived to set the worlds on the path that led to today's mundane, if relatively safe, existence. Both Tolkien and Herbert themselves portrayed themselves as mere humble quoters of the original stories, originally told by the great actors themselves. Their stories resonate with generations of the public partly because we understand that great storytellers are part of great stories which are part of great ages, come once in a long while, and cannot bequeath their talents and opportunities to their children.
On the bright side, both The Lord of the Rings and the Dune trilogies are so good that they can be reread often over a lifetime, delivering new rewards each time. Reading those later "extensions" is a waste of time that could better be spent rereading the original.
Of course a wristPC is what everyone's been waiting for since Dick Tracy. I even got one towards the end of the last millennium, but no one ever made it run Linux.
Making the WristPC dream come true will require 3 things: it should do nothing but wirelessly deliver a multimedia terminal to networked computation. Its videophone should enable voice recognition. And its screen should fold out, preferably automatically, to larger screens, at least 4x5" and 8.5x11".
Something like VNC+shoutcast makes the "2-way radio" part a reality today. Cameraphones are small and almost smart enough. The real trick is opening the screen on demand from 2x4". That's probably going to require nanotech materials a few years down the road.
Is there a way to use a paint material like this to capture the extra radio energy not among the tiny fraction actually received by an antenna? Sure, it's a tiny wattage, but if it could be collected for periodic recharging the wireless antenna, then the whole system could run a lot longer off batteries - if it were efficient.
The point of the 1.6Tbps EIB bandwidth citation is that it's the speed of IO for computation, just as you say. In fact, due to the crossconnects between SPEs, it's even faster, depending on which SPEs are linked. The RSX bandwidth isn't relevant, because the RSX is unavailable to Linux under the Hypervisor. But the FlexIO bandwidth is certainly relevant. 50GBps/200Gbps is a lot faster than the single GigE. Even getting data to the HDMI output, an unproven stunt, would at least get 10Gbps of the 200. The CompactFlash and SDIO ports could each support a GigE card, ganged together with maybe (cropped to 480Mbps) 1-4 USB GigE adapters, which could mean 3.5-6Mbps/in + 10.2Mbps/out. With a lot of fancy programming. Maybe some kind of waveguide connecting the 802.11g to an AP for another measly 54Mbps, and maybe some crazy SATA protocol for another 1.3Mbps.
Or maybe I have to solder a PCIe onto a bus somewhere (if only);). Or spend $8K for each blade, without any other HW, plus its backplane, probably from Mercury. At $600 each PS3, it's worth trying something to widen the bottleneck.
The mouse was a teaching aide to introduce users to the idea of moving the cursor on the screen the way they'd move a real object on their desk. Touchscreen tech was too crude in the early 1970s, or even in the 1980s, to introduce for direct pointing.
But now it works. Over a decade of PDA touchscreens has funded R&D that can put a precise, stable point just above the fingernail or stylus of any user.
Why do I have to use even a little trackpad in short strokes for indirect control of the cursor, when I could just point directly at that cursor? And why can't I use multiple fingers to describe lines, polygons, movement directions, multiple selections, and everything else I do with real objects on my real desktop?
No one said it was an "official" ad, just that it looked like it came from Obama's campaign. Just look at some of the press, like the WP blog linked from this article's summary. Even though it's the Washington Post, an "official" news org, and they found out that the culprit didn't work for Obama, was fired by his boss who had only an unrelated Obama connection, and Obama's campaign consistently denied any connection - no evidence there was a connection - the WP writer insisted in conclusion that it was still possibly Obama.
In other words, De Vellis' stunt worked, including backlash on Obama.
And even if he quit, it's only to preempt being fired. The difference isn't even academic.
I don't really care about the score. And I'm in a good mood. Because I like seeing anonymous attacks discredited early in the game, but still possible. It evolves the public, especially our suspicion of the "official" news media as automatically correct, or even responsible. And gets us closer to some way to make even "little guys" accountable for public communications, like we used to do with traditional libel.
Oh yeah, it's pretty simple. I hate loudmouth jerks who launch anonymous attacks. And that's a general rule, if not a "categorical imperative". Don't try to play with Kant when we're not talking about epistemology.
Despite your poindexter pretense, that's not a "natural law", even if it's generally true.
I did indeed respond to your question, pointing out that of course my general statement is a general statement. I'm not going to further dignify your bruiting about terms like "natural law" you evidently don't even understand. And if you don't understand how being obnoxious and supercilious while anonymous, you can't blame that on being a foreigner. Nobody likes someone like that.
A general rule that probably is a natural law, too.
It was so pro-Obama, including his logo on the "star's" shirt, that the reasonable impression that everyone had was that it came from Obama.
That's the entire point of a "false flag" operation. A good one has builtin "plausible deniability". In this case, there's no evidence to believe that Obama's campaign was involved in any way.
By being an overzealous "supporter", he thereby framed Obama. And thereby caused Obama damage.
He was fired, though he claimed he quit, as it says in the WP article linked from the story summary.
He framed the video as coming from Obama: the woman running to smash the screen has the Obama logo on her shirt. That was what caused people to believe it came from Obama, as it also says in the articles linked from the story summary.
Now you are 0 for 2. You don't even RTFA. Don't bother trying again.
Why would I grasp that, when it's not true? Of course the bus is to handle IO: it's connected directly to the FlexIO on the Cell to handle all the IO on an equal basis.
But then, you don't seem familiar with the PS3 architecture: the Cell is not the GPU, that's the (even faster and more esoteric) nVidia RSX. And I'm not even programming games, I'm using the Sony provided feature of installing Linux and programming this beast as a general purpose computer. Starting with its embedded Power RISC, then adding its parallel DSPs. So I return to what I said in my original post: if this thing only had more IO, I'd get even more use out of the Cell (and maybe the RSX, too).
Does anyone who wants to respond in this subthread have any acutal knowledge of the Cell, preferably of ways to increase the IO? Or is it just a Slashdot orgy of ignorance?
You don't win any arguments with "open your eyes" and nothing backing it up. What part of 204GFLOPS don't you understand? What part of the serial nature of IO, whether over SATA, USB, or ethernet don't you understand? What is it about addition of multiple channels for a total don't you understand?
When modern GPUs are as programmable as in the Cell in the PS3, as demonstrated most recently by the work in the story we're now discussing, then yes, they're supercomputers.
Except that more IO with such a fast bus means that the pipeline could refill the SPE local storages every time they're used up. And most of the apps that I want to process with a (hotrodded) PS3 are fine in single-precision float, which would need a supercomputer to crunch anyway. I want a supercomputer for nonscientific apps. Otherwise, I'd have more than $600 to spend on it.
Except that MS uses its announcements to "support XML" to confer the cloak of respectability as an open standard. Then they use XML to keep it proprietary, while saying they're not.
If people can't follow the story enough to track how MS abuses everything it touches, so I can just refer to it, then I'm really not interested in discussing it with them. I'm not writing a book here - I'm looking for discussion with people familiar with the issues.
And besides, since your argument claims that it's fair for MS to say they're using XML, I find your post to be more like a "concern troll". I don't think anyone who isn't already sympathetic to MS propaganda would find my perfectly legitimate accusations about MS and XML in this AJAX group context to support MS in any way. So thanks for your concern, but I'll just keep on.
What the hell are you talking about, Anonymous babbler Coward?
When a Republican runs a "false flag" operation like this to frame their clients and attack political opponents of course they should be fired. It should happen a lot more often, but it typically happens under orders by their Republican bosses, not in secret or on their own time. So they get a raise, or promotion up the "dirty tricks" ladder.
Only an anonymous Republican like you would project their partisan preference fears onto me for no reason. Damn, you can't even get "couched in generalized evenhanded axiomatic terms" right, you're struggling so hard not to say "but Republicans should be able to do this". You people are pathetic.
OK, I'm on the wrong side of people who retain employees who fraudulently frame clients with attacks they didn't make. Or I'm against the people who do that kind of fraud, that scares the clients.
Or maybe I'm just against the Anonymous idiot Cowards who think they're everything, but are really just the wrong side.
De Vellis was fired because he made a video attacking Clinton, fraudulently crediting it to the Obama campaign, while the Obama campaign was an actual (if tangential) customer where he actually works.
If he had not signed it "Obama", he might not have been fired. If his boss hadn't had Obama as a client, he might not have gotten fired.
This guy is a jerk. He's got the right to publish whatever video he comes up with, except when he lies in it. He has no right to frame Obama with that attack ad. And his boss has the right to fire a guy who pisses off the clients.
Not only did I not imply that, I explicitly mentioned that the CPU:IO throughput means 2000 instructions per IO.
You don't seem to know much about DSP. The vast majority of DSP is not logic, but arithmetic - the logic isn't usually that fast (except sometimes zero-overhead looping), but the arithmetic is extremely fast. The entire game in DSP is keeping the pipeline full. 2000:1 keeps the compute pipeline, the critical link, empty much of the time.
Moreover, there's no time for cache fetches in DSP loops - it's all working registers (the SPEs have many) and signal data from the bus.
Really, your comment demonstrates ignorance of both DSP and my post. Don't try to tell me about doing "everything in an amateur way" when you are the amateur and I am the expert.
Unless you have something to tell us about increasing the PS3 IO, which your comment leads me to doubt.
Anonymous troll Coward, what the hell has ESR got to do with anything, except collecting the Jargon File, and getting you to reveal that you're a troll?
The part of building my own Cell machine that's not "hacking" is that it's really system design, while getting the PS3 to do things its designers didn't expect is what hacking is all about. Using the existing device to exploit all its pent-up power is the soul of hacking.
I'm not bitching about what Sony didn't do - I even pointed out that they did quite a lot. I just asked the rest of the Slashdotters if they had other ideas for hacking more IO into the limited offering of the PS3.
Apparently you don't even know that Sony has officially inserted Linux support into the PS3 release from the beginning. Why don't you go bother someone who cares what you think about things about which you know nothing, instead of just serving as an excuse for me to bash an anonymous troll who doesn't understand hacking or community collaboration?
Because it's a toy supercomputer. If I find a way to expand its IO, I'll have a $600 supercomputer, scalable into a supercomputer cluster.
If I listen to you, Anonymous defeatist Coward, and just cry "waaaahhh, I'm too dumb to hack a toy into a tool", then I'll just have a really cool toy.
Allow me to introduce you to the term hack, which is what Slashdotters used to do before we were mostly posers.
How about a print array head of maybe a few dozen of these "full width" heads stacked up to print whole bands at once? Eventually a full "sheet at once" printer.
Then they can get really fancy, micropositioning the "print face" at subdot distances for even higher resolution...
Meanwhile, I'd like to use the printface with a video sensor for registration against the "last pass" for grafitti. Color, hi-res grafitti. Bombing by remote-control micro-helicopter...
Slashdot Mirror
True, OED was his first civilian gig after the War. I should have just left it as "his well known gig". Certainly his main gig was writing LotR, especially as reflected in his final net worth (AFAIK).
Good analogy. The difference between, say, The Fellowship of the Ring and any Christopher Tolkien followup (except perhaps the Silmarillion) is about as big.
JRR Tolkien and Frank Herbert were visionaries. Their books are legendary because they're so complete, so consistent, they're practically holographic. While those authors were also brilliant editors, especially Tolkien whose main gig was (as is well known) Oxford English Dictionary editor. Their (genetic, and thereby literary) heirs are undistinguished from a vast host of other second or lower tier of "visionary" authors, and have no special editing talent - nor have acquired any at their cashin publishers. While they also operate at a disadvantage while writing outside the original cultural contexts that produced those seminal works for a different audience.
Ironically, both Middle Earth and Dune are epic tales of the original forefathers of our times (Dune less obviously, sorry for the spoiler). A magical time when a unique individual arrived to set the worlds on the path that led to today's mundane, if relatively safe, existence. Both Tolkien and Herbert themselves portrayed themselves as mere humble quoters of the original stories, originally told by the great actors themselves. Their stories resonate with generations of the public partly because we understand that great storytellers are part of great stories which are part of great ages, come once in a long while, and cannot bequeath their talents and opportunities to their children.
On the bright side, both The Lord of the Rings and the Dune trilogies are so good that they can be reread often over a lifetime, delivering new rewards each time. Reading those later "extensions" is a waste of time that could better be spent rereading the original.
Of course a wristPC is what everyone's been waiting for since Dick Tracy. I even got one towards the end of the last millennium, but no one ever made it run Linux.
Making the WristPC dream come true will require 3 things: it should do nothing but wirelessly deliver a multimedia terminal to networked computation. Its videophone should enable voice recognition. And its screen should fold out, preferably automatically, to larger screens, at least 4x5" and 8.5x11".
Something like VNC+shoutcast makes the "2-way radio" part a reality today. Cameraphones are small and almost smart enough. The real trick is opening the screen on demand from 2x4". That's probably going to require nanotech materials a few years down the road.
Is there a way to use a paint material like this to capture the extra radio energy not among the tiny fraction actually received by an antenna? Sure, it's a tiny wattage, but if it could be collected for periodic recharging the wireless antenna, then the whole system could run a lot longer off batteries - if it were efficient.
The point of the 1.6Tbps EIB bandwidth citation is that it's the speed of IO for computation, just as you say. In fact, due to the crossconnects between SPEs, it's even faster, depending on which SPEs are linked. The RSX bandwidth isn't relevant, because the RSX is unavailable to Linux under the Hypervisor. But the FlexIO bandwidth is certainly relevant. 50GBps/200Gbps is a lot faster than the single GigE. Even getting data to the HDMI output, an unproven stunt, would at least get 10Gbps of the 200. The CompactFlash and SDIO ports could each support a GigE card, ganged together with maybe (cropped to 480Mbps) 1-4 USB GigE adapters, which could mean 3.5-6Mbps/in + 10.2Mbps/out. With a lot of fancy programming. Maybe some kind of waveguide connecting the 802.11g to an AP for another measly 54Mbps, and maybe some crazy SATA protocol for another 1.3Mbps.
;). Or spend $8K for each blade, without any other HW, plus its backplane, probably from Mercury. At $600 each PS3, it's worth trying something to widen the bottleneck.
Or maybe I have to solder a PCIe onto a bus somewhere (if only)
The mouse was a teaching aide to introduce users to the idea of moving the cursor on the screen the way they'd move a real object on their desk. Touchscreen tech was too crude in the early 1970s, or even in the 1980s, to introduce for direct pointing.
But now it works. Over a decade of PDA touchscreens has funded R&D that can put a precise, stable point just above the fingernail or stylus of any user.
Why do I have to use even a little trackpad in short strokes for indirect control of the cursor, when I could just point directly at that cursor? And why can't I use multiple fingers to describe lines, polygons, movement directions, multiple selections, and everything else I do with real objects on my real desktop?
Read the definition. Then indulge your morbid side as much as you like. But not on my time. Goodbye.
Put the pipe down, Anonymous crackhead Coward. Try programming the PS3 instead. It's faster, and less delusional.
No one said it was an "official" ad, just that it looked like it came from Obama's campaign. Just look at some of the press, like the WP blog linked from this article's summary. Even though it's the Washington Post, an "official" news org, and they found out that the culprit didn't work for Obama, was fired by his boss who had only an unrelated Obama connection, and Obama's campaign consistently denied any connection - no evidence there was a connection - the WP writer insisted in conclusion that it was still possibly Obama.
In other words, De Vellis' stunt worked, including backlash on Obama.
And even if he quit, it's only to preempt being fired. The difference isn't even academic.
I don't really care about the score. And I'm in a good mood. Because I like seeing anonymous attacks discredited early in the game, but still possible. It evolves the public, especially our suspicion of the "official" news media as automatically correct, or even responsible. And gets us closer to some way to make even "little guys" accountable for public communications, like we used to do with traditional libel.
Oh yeah, it's pretty simple. I hate loudmouth jerks who launch anonymous attacks. And that's a general rule, if not a "categorical imperative". Don't try to play with Kant when we're not talking about epistemology.
Despite your poindexter pretense, that's not a "natural law", even if it's generally true.
I did indeed respond to your question, pointing out that of course my general statement is a general statement. I'm not going to further dignify your bruiting about terms like "natural law" you evidently don't even understand. And if you don't understand how being obnoxious and supercilious while anonymous, you can't blame that on being a foreigner. Nobody likes someone like that.
A general rule that probably is a natural law, too.
It was so pro-Obama, including his logo on the "star's" shirt, that the reasonable impression that everyone had was that it came from Obama.
That's the entire point of a "false flag" operation. A good one has builtin "plausible deniability". In this case, there's no evidence to believe that Obama's campaign was involved in any way.
By being an overzealous "supporter", he thereby framed Obama. And thereby caused Obama damage.
He was fired, though he claimed he quit, as it says in the WP article linked from the story summary.
He framed the video as coming from Obama: the woman running to smash the screen has the Obama logo on her shirt. That was what caused people to believe it came from Obama, as it also says in the articles linked from the story summary.
Now you are 0 for 2. You don't even RTFA. Don't bother trying again.
Why would I grasp that, when it's not true? Of course the bus is to handle IO: it's connected directly to the FlexIO on the Cell to handle all the IO on an equal basis.
But then, you don't seem familiar with the PS3 architecture: the Cell is not the GPU, that's the (even faster and more esoteric) nVidia RSX. And I'm not even programming games, I'm using the Sony provided feature of installing Linux and programming this beast as a general purpose computer. Starting with its embedded Power RISC, then adding its parallel DSPs. So I return to what I said in my original post: if this thing only had more IO, I'd get even more use out of the Cell (and maybe the RSX, too).
Does anyone who wants to respond in this subthread have any acutal knowledge of the Cell, preferably of ways to increase the IO? Or is it just a Slashdot orgy of ignorance?
Fuck you. Stop fantasizing about me jerking off.
You don't win any arguments with "open your eyes" and nothing backing it up. What part of 204GFLOPS don't you understand? What part of the serial nature of IO, whether over SATA, USB, or ethernet don't you understand? What is it about addition of multiple channels for a total don't you understand?
When modern GPUs are as programmable as in the Cell in the PS3, as demonstrated most recently by the work in the story we're now discussing, then yes, they're supercomputers.
I know your disorder. You're a STUPID ASSHOLE.
Did I mention FUCK YOU? FUCK YOU.
Except that more IO with such a fast bus means that the pipeline could refill the SPE local storages every time they're used up. And most of the apps that I want to process with a (hotrodded) PS3 are fine in single-precision float, which would need a supercomputer to crunch anyway. I want a supercomputer for nonscientific apps. Otherwise, I'd have more than $600 to spend on it.
Except that MS uses its announcements to "support XML" to confer the cloak of respectability as an open standard. Then they use XML to keep it proprietary, while saying they're not.
If people can't follow the story enough to track how MS abuses everything it touches, so I can just refer to it, then I'm really not interested in discussing it with them. I'm not writing a book here - I'm looking for discussion with people familiar with the issues.
And besides, since your argument claims that it's fair for MS to say they're using XML, I find your post to be more like a "concern troll". I don't think anyone who isn't already sympathetic to MS propaganda would find my perfectly legitimate accusations about MS and XML in this AJAX group context to support MS in any way. So thanks for your concern, but I'll just keep on.
What the hell are you talking about, Anonymous babbler Coward?
When a Republican runs a "false flag" operation like this to frame their clients and attack political opponents of course they should be fired. It should happen a lot more often, but it typically happens under orders by their Republican bosses, not in secret or on their own time. So they get a raise, or promotion up the "dirty tricks" ladder.
Only an anonymous Republican like you would project their partisan preference fears onto me for no reason. Damn, you can't even get "couched in generalized evenhanded axiomatic terms" right, you're struggling so hard not to say "but Republicans should be able to do this". You people are pathetic.
OK, I'm on the wrong side of people who retain employees who fraudulently frame clients with attacks they didn't make.
Or I'm against the people who do that kind of fraud, that scares the clients.
Or maybe I'm just against the Anonymous idiot Cowards who think they're everything, but are really just the wrong side.
I'll take it.
De Vellis was fired because he made a video attacking Clinton, fraudulently crediting it to the Obama campaign, while the Obama campaign was an actual (if tangential) customer where he actually works.
If he had not signed it "Obama", he might not have been fired. If his boss hadn't had Obama as a client, he might not have gotten fired.
This guy is a jerk. He's got the right to publish whatever video he comes up with, except when he lies in it. He has no right to frame Obama with that attack ad. And his boss has the right to fire a guy who pisses off the clients.
All it would take would be including PCI-Express.
Not only did I not imply that, I explicitly mentioned that the CPU:IO throughput means 2000 instructions per IO.
You don't seem to know much about DSP. The vast majority of DSP is not logic, but arithmetic - the logic isn't usually that fast (except sometimes zero-overhead looping), but the arithmetic is extremely fast. The entire game in DSP is keeping the pipeline full. 2000:1 keeps the compute pipeline, the critical link, empty much of the time.
Moreover, there's no time for cache fetches in DSP loops - it's all working registers (the SPEs have many) and signal data from the bus.
Really, your comment demonstrates ignorance of both DSP and my post. Don't try to tell me about doing "everything in an amateur way" when you are the amateur and I am the expert.
Unless you have something to tell us about increasing the PS3 IO, which your comment leads me to doubt.
Anonymous troll Coward, what the hell has ESR got to do with anything, except collecting the Jargon File, and getting you to reveal that you're a troll?
The part of building my own Cell machine that's not "hacking" is that it's really system design, while getting the PS3 to do things its designers didn't expect is what hacking is all about. Using the existing device to exploit all its pent-up power is the soul of hacking.
I'm not bitching about what Sony didn't do - I even pointed out that they did quite a lot. I just asked the rest of the Slashdotters if they had other ideas for hacking more IO into the limited offering of the PS3.
Apparently you don't even know that Sony has officially inserted Linux support into the PS3 release from the beginning. Why don't you go bother someone who cares what you think about things about which you know nothing, instead of just serving as an excuse for me to bash an anonymous troll who doesn't understand hacking or community collaboration?
Moderation -1
100% Troll
If kicking an Anonymous troll Coward goodbye in the same style they slimed hello is a "Troll", then maybe I earned that one.
Because it's a toy supercomputer. If I find a way to expand its IO, I'll have a $600 supercomputer, scalable into a supercomputer cluster.
If I listen to you, Anonymous defeatist Coward, and just cry "waaaahhh, I'm too dumb to hack a toy into a tool", then I'll just have a really cool toy.
Allow me to introduce you to the term hack, which is what Slashdotters used to do before we were mostly posers.
How about a print array head of maybe a few dozen of these "full width" heads stacked up to print whole bands at once? Eventually a full "sheet at once" printer.
Then they can get really fancy, micropositioning the "print face" at subdot distances for even higher resolution...
Meanwhile, I'd like to use the printface with a video sensor for registration against the "last pass" for grafitti. Color, hi-res grafitti. Bombing by remote-control micro-helicopter...