Actually, I really liked the ending of Cryptonomicon. The last three pages move incredibly fast (in my hardcover ed.) have no dialogue, and in my head this translated into a sort of fugue, a lot like the ending of Apocalypse Now. Lots of cuts and dramatic music and bad lighting, especially the last couple of paragraphs.
Netzero is Windows Only. Thats the part that really bugs me.
It's much worse than you think.
The free service offers ten hours per month, which should be sufficient to get you through any short-term outages. This service does not currently support MAC, Windows 2000 or XP.
Not only do they not support other operating systems, they don't even support all the different varieties of Windows. Hmmph.
There's an old adage that sooner or later every argument will devolve to the point where one side or the other compares the subject of the argument to Hitler. At that point, the argument is over.
(Too lazy to look up a source for that right now. If you know who said that first, share and enjoy!)
Similarly, every discussion of the merits of one programming language or practice versus another will devolve until eventually somebody plays the "well-designed system" card.
"In a well-designed system, changes would be easy."
"In a well-designed system, implementations would be both opaque and portable."
"In a well-designed system, code would be self-documenting."
Now that we're through counting the merits of well-designed systems, could we please get back to talking about real-world experiences?
I'm sure any intelligent game programmer won't try to push for 100 fps on a 29.97 NTSC output, but you never know and it's still a possibility.
Actually, it isn't. If your display's vertical refresh isn't an integer divisor of your application frame rate, you'll actually end up dropping frames on the output. At some point, determined by the ratio of display refresh rate to app frame rate, the two cycles will be sufficiently out of sync that the display refreshes when a new frame isn't available in the framebuffer. Poof. Dropped frame.
In hard-real-time applications, like flight simulation for instance, making sure the vertical sync signal given to your graphics hardware is used as the basis for your application frame clock is very, very important. When you're dealing with the relatively low refresh rates of analog TV, it's important even for game consoles.
I don't see why there won't be a mixture of standards. IDE/ATAPI for joe consumer, SCSI for us discrimating desktop/server buyers, and FC for people who have too much money and like buzzwords.
Don't underestimate fibre channel. It's a very fast interconnect that's easy to implement over long distances with optical cabling, and that supports 127 units per loop.
I have about a terabyte of FC disk in my lab on various FC loops, and the stuff works so well it's almost funny. And our customers have, combined, many, many TB of FC JBOD. Super reliable.
And that's without even getting into switched storage fabrics. Personally I think cross-platform storage-sharing schemes like Sanergy-- and others whose names I forget-- are pretty kludgey, but shared stored in a single-OS environment works really well.
And even without shared storage, the ability to put all your storage-- both disk and tape-- and all your computers on a big FC switch and dynamically move devices from machine to machine just by unmounting over there and mounting over here... well, that's just plain cool.
I know this is completely off-topic, but here it is anyway.
Your understanding of the prefix "meta-" is incomplete. In addition to indicating syntactic self-reference (see Hofstader), it can also indicate semantic self-reference (see... well, Hofstader; he talks about this, too, in his discussion of GOD: God Over Djinn).
SGI has a device for connecting crossbar routers together to form large single-system-image computers. It's called a metarouter:a router for routers.
Likewise, a cluster of clusters would be properly called a metacluster. Since "Beowulf" is commonly synonymous with "cluster," the term "meta-beowulf" is pretty much correct, even though it makes me cringe.
I believe that the universe has always existed, and will continue to exist forever. It is also infinitely large in every direction. The only reason we can't see the rest is because the light hasn't gotten here.
Hmm.
The universe is infinitely large, and infinitely old.
Since the universe is infinitely old, light from luminous objects in all parts of the universe has been radiating for infinite time.
But you think light from distant objects "hasn't gotten here yet?"
I wish I'd had you on my side during my last tax audit!
The result of this is more flicker, not less. Especially if you have sharp horizontal lines in your image, the interlacing flicker can be appalling.
Actually, I think you're referring to aliasing, or the shimmering moire patterns one sometimes sees in interlaced images. This happens when a very small spot, or pattern of small spots, seems to jump back and forth between even and odd fields as the image refreshes.
The only example I can think of off the top of my head is in the VHS tape version of Braveheart. Near the beginning there's a shot of a soldier in chain mail, and the sharply detailed pattern in his armor shimmers on a regular TV like nobody's business.
The reason that I find 1080i interesting is that it is quoted as an interlaced spec. Many people think that 1080i is better than 720p for this reason. This is blatantly false. A 720p image has more data.
Sorry, you're not correct. 1080i is 1920x1080 at 30 frames per second, and 720p is 1280x720 at 60 frames per second.
1280*720*60
55296000 pixels per second
1920*1080*30
62208000 pixels per second
Sorry, but in terms of pure bits per second, 1080i wins.
Interesting, the 1080i ration for HDTV (which can also carry a 540p signal) is interlaced.
(Disclosure: the company I work for has a broadcast integration arm, so I'm always hearing people yak yak yak about the latest TV technology. And we also play with HDTV a lot in our lab.)
Interlacing provides one very imporant feature that you didn't mention: flicker reduction.
I don't have science to back this up, but it seems that a 60 Hz progressive display-- one in which every scan line is redrawn 60 times per second-- has a noticable flicker to most people. (Set your computer monitor to 60 Hz and see. No, really. Go ahead. I'll wait.)
A 60-Hz interlaced display, however, flickers less, because only *half* the scan lines are being redrawn at any given time. Your eye perceives a clean, flicker-free image, although admittedly this can result in some tearing or blurring when the scene moves quickly.
We've known for years and years that refresh rate is more important than frame rate for visually pleasing, flicker-free pictures. Analog film is projected at 24 frames per second, which is a really low frame rate by digital standards. We get away with it because the film projector gates (that is, projects on the screen) each frame twice, meaning the screen flashes 48 times per second. Less flicker for the same frame rate.
Interlacing uses the same principle but in a different way. Instead of refreshing the screen faster than we update it, we only update half the scan lines each time through the raster, leaving the other half lit. This works because phosphors on TV tubes continue to glow after they've been excited, so we basically get half of our scan lines for free every refresh.
My point here is that you might find a 60 Hz progressive scan display more pleasing than 60 Hz interlaced scan in some cases, but it's not universally true to say that progressive is always better than interlaced.
We should just give all cases of all words no matter what?
I think the point is that it's Microsoft doing the choosing. We have established, respected organizations that make dictionaries and thesaur... thes... uh, dictionaries, just dictionaries.
Anyway, my point is that Microsoft really should just license Webster's or the OED or something and stop making these kinds of decisions.
Yeah right. Instead, they can simply spin it as "terrorism".
Actually, I think the whole computer-crime-as-terrorism thing is a pretty useful analogy.
When the bad stuff happened last month, the FAA responded by completely shutting down all air travel in the US until major policy changes could be instituted. Did it have a serious impact on the security of the US air travel system? Dunno. Maybe. The point is, the FAA acted, and acted fast, doing the best job they could think of. We'll never know, thankfully, if they saved lives by doing so.
When nimda happened, Microsoft responded by... um. Actually, how did they respond? Exactly what swift, decisive measures did MS take to lessen the impact of that problem, and prevent future problems?
Re:Please inact a subscription service!
on
Slashdot Updates
·
· Score: 3, Interesting
Oh, yeah, excellent idea. The higher your karma, the fewer/smaller your ads are.
It makes sense, too. In theory, better content means more readers will visit the site, thereby increasing the "effectiveness" of the ads on the site. (I abstain from the argument of whether ads are effective at all.)
IRIX runs on a single hardware design, so it is always customized for it.
Actually, that's not right, either. IRIX 6.2 scaled about the same on the Challenge-series architecture (up to 36 processors) as IRIX 6.5 does on the Origin (up to 512 processors), two radically different designs.
It really has more to do with operating system architecture and scheduler design than it does with hardware.
If you follow the Linux kernel development, and read around, you'd notice that scaling to a 2-way or 4-way machine is a big leap in performance. Throw Linux or any other OS on a 6-way or 8-way machine and you will watch that increase in performance degrade (ie a 2-way machine isnt x 2 the performance of a single CPU machine, and an 8-way system isnt x 2 the performance of a 4-way machine).
This, of course, is crap. To say that "any other OS" has the same scalability problem that Linux has is simply not true.
Take IRIX, for instance. I wrote some image processing code that runs on Origin servers. The 8-processor server in my lab runs my code about four times faster than my 2-p servers. And, surprise, the 32-p server in my friend's lab runs my code about four times faster than my 8-p machine.
To generalize the problems you see on Linux and Windows to "any other" operating system is simply hogwash. Your point about Windows scalability is well taken, though.
Anything simple and obvious shouldn't be patentable.
"It is a rare mind indeed that can render the hitherto non-existent blindingly obvious. The cry 'I could have thought of that' is a very popular and misleading one, for the fact is that they didn't, and a very significant and revealing fact it is too."
Douglas Adams, Dirk Gently's Holistic Detective Agency
An already-existing attempt at fiber interconnect is called "Fibre Channel".
Fibre channel, of course, has nothing whatsoever to do with fibre optics. You can run fibre channel over either fibre optic cables or four-wire copper cables. I've got about a 50/50 mix of each in my lab. The only difference between the optical and the copper interconnects is that you can string the fibre optics farther than the copper cables. The speeds and latencies are the same.
This info is a little out of date-- it comes from Practical Unix Programming by Robbins and Robbins, published in '96.
It's a table of access times, scaled so 10 ns is equal to 1 second.
Processor cycle: 1 second
Cache access: 3 seconds
Memory access: 20 seconds
Context switch: 166 minutes
Disk access: 11 days
Notice that this table doesn't discuss bus bandwidths. The reason is simple: latency is more important than bus bandwidth for these kinds of comparisons. It doesn't matter if you can suck in 800 MB per second from RAM to CPU if getting that first byte still takes many nanoseconds.
In short, for normal server or desktop tasks, bus bandwidth isn't a serious bottleneck at all. But for traditional HPC applications, where a processor takes a huge chunk of data (measured at least in 10s of megabytes) and operates on it serially, from front to back, your bus and memory bottlenecks start to show through.
It's kind of analogous to having a car with a top speed of 250 MPH and a 0-60 time of four minutes. On the highway, once you get up to speed, you'll cruise along nicely. (Think of that as big serial computations.) But in stop-and-go traffic in the city, you're sucking. (Typical branching programs that depend on user input.)
If I could get a OSX native copy of Quark, Photoshop & Illustrator we would switch all of our OS9 desktops to OSX immediatly.
Illustrator X, which is either Cocoa or Carbon-- not sure which-- is available now. I've used the beta; it's nifty.
But I'm an old-timer. I use Photoshop 5.0, Quark 4, and Illustrator 8.0.1 in Classic mode under X. They're actually a lot more reliable under X than they are under 9. And under 10.1 on my Macs, Classic apps feel just as fast as native apps.
I'd advise you to save about 150,000 of those pennies and buy an iBook instead. My best friend has a PowerBook G4, and I have an iBook, and while the big screen on the G4 is nice, we both agree that my iBook is a better laptop.
That nice titanium case on the G4 scratches and scuffs incredibly easily, and it gets very very hot. Not to mention the fact that the slight flex in the G4's case makes it all too likely for a spinning CD or DVD to grind against the inside of the drive; it happens to my friend about once every other day.
My iBook, on the other hand, is a dream. I'd consider it to be *almost* good enough for an only machine, and perfect for a second machine.
Oh, and another thing. (feeble attempt to get back on topic) My iBook and my friend's PBG4 feel just about the same under OS 10.1 with 384 MB of RAM each. Both very, very usable.
I thought in the good US of A that all projects that the government does the people of the US of A had access to the source [u]nless [i]t was deemed that it was endangering security of the nation [o]r [i]t was contracted out to a company and then they had all the IP
Both of those things is usually true.
For two years I worked as a consultant for a company that built training simulators for the USAF, the Air Force Reserve, and several foreign military services (countries like Denmark, South Korea, Jordan, Egypt, and so on). This company's two big products were an F-16 tactical simulator and an F-18 mission simulator.
Some fairly significant parts of the simulator runtime code are classified. As an example, some configurations of the F-16 can be equipped with the AIM-120 AMRAAM radar-guided missile. The code that handles the capabilities of the AMRAAM, and its interactions with the mission control systems, and its dynamics in the air is all classified.
(The details of this code is classified; the existence of it isn't. At least, I hope not. Otherwise, I'm in a shit-load of trouble right now.)
So obviously classified code can't be open-sourced.
The code that wasn't classified (a lot-- if you replaced classified modules with stubs, it was possible to run the whole F-16 load in unclassified mode; we did that a lot for visiting suits and stuff) was proprietary. In some cases, it was highly proprietary.
My example here is the F-16's mission control computer. The source code for this computer's programs was provided to us by Lockheed; we translated it line-by-line, mostly by hand, from assembly language into Ada-83 and compiled it to run on the sim's SGI Onyx host computer. This module was basically the core of the simulator, and it was of course a closely guarded commercial secret, even though it wasn't technically classified by the DOD.
We did something similar with the F-18's mission computer programs, but instead of translating them, we ran them natively in a Motorola processor emulator on the SGI host. This was kind of a cluster f*ck; it took 17 MIPS CPUs to emulate the two Motorola processors and the one 1553 mux bus controller in the MCC in real time. But somebody decided it was cheaper to throw hardware at it than to translate Boeing's code.
The other distinctive thing about US military flight sims-- at least the two I worked closely on, and also the F-22 tactical sim with which I worked a little-- is that they're not generic flight-dynamics simulators. Rather than taking the programmed characteristics of a wing or an airframe, like it sounds like X-Plane does, these sims were built with the full knowledge of the aircraft's flight characteristics. So it would be completely impossible to take the F-16 Block 30 code, change a data file, and have an F-16 Block 42 sim, much less a space shuttle sim or a 767 sim or whatever. These apps just weren't built like that.
A lot more goes into a tactical or mission training sim than just flight dynamics, anyway. I'd guess that maybe one out of five modules in the F-16 sim dealt with flying the plane; less than that in the F-18 sim. The rest was cockpit interface drivers (we had a real cockpit, with hundreds of individual hardware devices, wired into the sim; the serial mux control code was impressive) and inter-sim communication (DIS [defense information systems] and HLA [high-level architecture] protocols) and image generation and tactical DCS (distributed coordinate system) databases and the operator/instructor interface and it goes on and on and on. These things would only be relevant in context of a military tactical or mission sim, flown by military pilots in training, in a military installation with military instructors and other military sims connected over the military's encrypted wide-area training network.
I hope that answered your question, at least in part.
I saw this monitor last November in Dallas, at SuperComputing 2000. It was sort of stashed away inside the IBM booth behind some of their big iron. It was big and bright and sharp, and I got the story of its origin from an IBM guy standing nearby.
Warning: the person who told me this may have been a salesman. I can't claim to know this to be absolutely true.
According to the IBM guy, the folks from Livermore National Labs wanted, for some reason related to monitoring or surveillence or something like that, a monitor that could display four HDTV-resolution images in a tile. IBM tiled four 1920x1200 images (HDTV's 1920x1080 fits nicely inside the 1920x1200 display standard) on one monitor and sold bunches of them to LLNL for a red-blooded American fortune.
At that time, IBM called the monitor "Big Bertha." That was the official name and everything; they had data sheets printed up to hand out at the show.
And everything everybody has said so far is true: at that kind of resolution, your desktop icons are about a quarter of an inch across. And xterms? Forget it. You've gotta set the font size to 36 points just to be able to read it comfortably!
But then they IBM guy opened up a full-color satellite image of some city or other-- I forget which one. He full-screened it, and then used the mouse to pan and rotate around it. I actually got dizzy; it was like looking through a window. It was AMAZING. I've never, ever seen anything like that before.
Of course, to push about 10 million full-color pixels around in real-time like that required something more that a $99 graphics card; the monitor was hooked up to an SP node or something similarly impressive.
Slashdot Mirror
(Topic? What topic?)
Actually, I really liked the ending of Cryptonomicon. The last three pages move incredibly fast (in my hardcover ed.) have no dialogue, and in my head this translated into a sort of fugue, a lot like the ending of Apocalypse Now. Lots of cuts and dramatic music and bad lighting, especially the last couple of paragraphs.
Then again, that may just be me.
Netzero is Windows Only. Thats the part that really bugs me.
It's much worse than you think.
The free service offers ten hours per month, which should be sufficient to get you through any short-term outages. This service does not currently support MAC, Windows 2000 or XP.
Not only do they not support other operating systems, they don't even support all the different varieties of Windows. Hmmph.
"We're cross-platform! NT and '95!"
(This from the FAQ.)
There's an old adage that sooner or later every argument will devolve to the point where one side or the other compares the subject of the argument to Hitler. At that point, the argument is over.
(Too lazy to look up a source for that right now. If you know who said that first, share and enjoy!)
Similarly, every discussion of the merits of one programming language or practice versus another will devolve until eventually somebody plays the "well-designed system" card.
"In a well-designed system, changes would be easy."
"In a well-designed system, implementations would be both opaque and portable."
"In a well-designed system, code would be self-documenting."
Now that we're through counting the merits of well-designed systems, could we please get back to talking about real-world experiences?
I'm sure any intelligent game programmer won't try to push for 100 fps on a 29.97 NTSC output, but you never know and it's still a possibility.
Actually, it isn't. If your display's vertical refresh isn't an integer divisor of your application frame rate, you'll actually end up dropping frames on the output. At some point, determined by the ratio of display refresh rate to app frame rate, the two cycles will be sufficiently out of sync that the display refreshes when a new frame isn't available in the framebuffer. Poof. Dropped frame.
In hard-real-time applications, like flight simulation for instance, making sure the vertical sync signal given to your graphics hardware is used as the basis for your application frame clock is very, very important. When you're dealing with the relatively low refresh rates of analog TV, it's important even for game consoles.
I don't see why there won't be a mixture of standards. IDE/ATAPI for joe consumer, SCSI for us discrimating desktop/server buyers, and FC for people who have too much money and like buzzwords.
Don't underestimate fibre channel. It's a very fast interconnect that's easy to implement over long distances with optical cabling, and that supports 127 units per loop.
I have about a terabyte of FC disk in my lab on various FC loops, and the stuff works so well it's almost funny. And our customers have, combined, many, many TB of FC JBOD. Super reliable.
And that's without even getting into switched storage fabrics. Personally I think cross-platform storage-sharing schemes like Sanergy-- and others whose names I forget-- are pretty kludgey, but shared stored in a single-OS environment works really well.
And even without shared storage, the ability to put all your storage-- both disk and tape-- and all your computers on a big FC switch and dynamically move devices from machine to machine just by unmounting over there and mounting over here... well, that's just plain cool.
(Goodbye, karma.)
I know this is completely off-topic, but here it is anyway.
Your understanding of the prefix "meta-" is incomplete. In addition to indicating syntactic self-reference (see Hofstader), it can also indicate semantic self-reference (see... well, Hofstader; he talks about this, too, in his discussion of GOD: God Over Djinn).
SGI has a device for connecting crossbar routers together to form large single-system-image computers. It's called a metarouter:a router for routers.
Likewise, a cluster of clusters would be properly called a metacluster. Since "Beowulf" is commonly synonymous with "cluster," the term "meta-beowulf" is pretty much correct, even though it makes me cringe.
I believe that the universe has always existed, and will continue to exist forever. It is also infinitely large in every direction. The only reason we can't see the rest is because the light hasn't gotten here.
Hmm.
The universe is infinitely large, and infinitely old.
Since the universe is infinitely old, light from luminous objects in all parts of the universe has been radiating for infinite time.
But you think light from distant objects "hasn't gotten here yet?"
I wish I'd had you on my side during my last tax audit!
The result of this is more flicker, not less. Especially if you have sharp horizontal lines in your image, the interlacing flicker can be appalling.
Actually, I think you're referring to aliasing, or the shimmering moire patterns one sometimes sees in interlaced images. This happens when a very small spot, or pattern of small spots, seems to jump back and forth between even and odd fields as the image refreshes.
The only example I can think of off the top of my head is in the VHS tape version of Braveheart. Near the beginning there's a shot of a soldier in chain mail, and the sharply detailed pattern in his armor shimmers on a regular TV like nobody's business.
HDTV is 1920x1080 at a few different frame rates.
Actually, there are six HDTV formats, according to the Advanced Television Standards Committee:
1920x1080 @ 24 fps (1080p24, or just "24p")
1920x1080 @ 30 fps (1080p)
1920x1080 @ 60 fps (1080i)
1280x720 @ 24 fps
1280x720 @ 30 fps
1280x720 @ 60 fps
The ATSC has also approved 12 (!!) formats for digital standard definition TV.
More info here: http://www.atsc.org/press/PR_Def.html
The reason that I find 1080i interesting is that it is quoted as an interlaced spec. Many people think that 1080i is better than 720p for this reason. This is blatantly false. A 720p image has more data.
Sorry, you're not correct. 1080i is 1920x1080 at 30 frames per second, and 720p is 1280x720 at 60 frames per second.
1280*720*60
55296000 pixels per second
1920*1080*30
62208000 pixels per second
Sorry, but in terms of pure bits per second, 1080i wins.
Interesting, the 1080i ration for HDTV (which can also carry a 540p signal) is interlaced.
(Disclosure: the company I work for has a broadcast integration arm, so I'm always hearing people yak yak yak about the latest TV technology. And we also play with HDTV a lot in our lab.)
Interlacing provides one very imporant feature that you didn't mention: flicker reduction.
I don't have science to back this up, but it seems that a 60 Hz progressive display-- one in which every scan line is redrawn 60 times per second-- has a noticable flicker to most people. (Set your computer monitor to 60 Hz and see. No, really. Go ahead. I'll wait.)
A 60-Hz interlaced display, however, flickers less, because only *half* the scan lines are being redrawn at any given time. Your eye perceives a clean, flicker-free image, although admittedly this can result in some tearing or blurring when the scene moves quickly.
We've known for years and years that refresh rate is more important than frame rate for visually pleasing, flicker-free pictures. Analog film is projected at 24 frames per second, which is a really low frame rate by digital standards. We get away with it because the film projector gates (that is, projects on the screen) each frame twice, meaning the screen flashes 48 times per second. Less flicker for the same frame rate.
Interlacing uses the same principle but in a different way. Instead of refreshing the screen faster than we update it, we only update half the scan lines each time through the raster, leaving the other half lit. This works because phosphors on TV tubes continue to glow after they've been excited, so we basically get half of our scan lines for free every refresh.
My point here is that you might find a 60 Hz progressive scan display more pleasing than 60 Hz interlaced scan in some cases, but it's not universally true to say that progressive is always better than interlaced.
We should just give all cases of all words no matter what?
I think the point is that it's Microsoft doing the choosing. We have established, respected organizations that make dictionaries and thesaur... thes... uh, dictionaries, just dictionaries.
Anyway, my point is that Microsoft really should just license Webster's or the OED or something and stop making these kinds of decisions.
Yeah right. Instead, they can simply spin it as "terrorism".
Actually, I think the whole computer-crime-as-terrorism thing is a pretty useful analogy.
When the bad stuff happened last month, the FAA responded by completely shutting down all air travel in the US until major policy changes could be instituted. Did it have a serious impact on the security of the US air travel system? Dunno. Maybe. The point is, the FAA acted, and acted fast, doing the best job they could think of. We'll never know, thankfully, if they saved lives by doing so.
When nimda happened, Microsoft responded by... um. Actually, how did they respond? Exactly what swift, decisive measures did MS take to lessen the impact of that problem, and prevent future problems?
Oh, yeah, excellent idea. The higher your karma, the fewer/smaller your ads are.
It makes sense, too. In theory, better content means more readers will visit the site, thereby increasing the "effectiveness" of the ads on the site. (I abstain from the argument of whether ads are effective at all.)
IRIX runs on a single hardware design, so it is always customized for it.
Actually, that's not right, either. IRIX 6.2 scaled about the same on the Challenge-series architecture (up to 36 processors) as IRIX 6.5 does on the Origin (up to 512 processors), two radically different designs.
It really has more to do with operating system architecture and scheduler design than it does with hardware.
If you follow the Linux kernel development, and read around, you'd notice that scaling to a 2-way or 4-way machine is a big leap in performance. Throw Linux or any other OS on a 6-way or 8-way machine and you will watch that increase in performance degrade (ie a 2-way machine isnt x 2 the performance of a single CPU machine, and an 8-way system isnt x 2 the performance of a 4-way machine).
This, of course, is crap. To say that "any other OS" has the same scalability problem that Linux has is simply not true.
Take IRIX, for instance. I wrote some image processing code that runs on Origin servers. The 8-processor server in my lab runs my code about four times faster than my 2-p servers. And, surprise, the 32-p server in my friend's lab runs my code about four times faster than my 8-p machine.
To generalize the problems you see on Linux and Windows to "any other" operating system is simply hogwash. Your point about Windows scalability is well taken, though.
Anything simple and obvious shouldn't be patentable.
"It is a rare mind indeed that can render the hitherto non-existent blindingly obvious. The cry 'I could have thought of that' is a very popular and misleading one, for the fact is that they didn't, and a very significant and revealing fact it is too."
Douglas Adams, Dirk Gently's Holistic Detective Agency
An already-existing attempt at fiber interconnect is called "Fibre Channel".
Fibre channel, of course, has nothing whatsoever to do with fibre optics. You can run fibre channel over either fibre optic cables or four-wire copper cables. I've got about a 50/50 mix of each in my lab. The only difference between the optical and the copper interconnects is that you can string the fibre optics farther than the copper cables. The speeds and latencies are the same.
how much of a bottleneck is the bus right now?
This info is a little out of date-- it comes from Practical Unix Programming by Robbins and Robbins, published in '96.
It's a table of access times, scaled so 10 ns is equal to 1 second.
Processor cycle: 1 second
Cache access: 3 seconds
Memory access: 20 seconds
Context switch: 166 minutes
Disk access: 11 days
Notice that this table doesn't discuss bus bandwidths. The reason is simple: latency is more important than bus bandwidth for these kinds of comparisons. It doesn't matter if you can suck in 800 MB per second from RAM to CPU if getting that first byte still takes many nanoseconds.
In short, for normal server or desktop tasks, bus bandwidth isn't a serious bottleneck at all. But for traditional HPC applications, where a processor takes a huge chunk of data (measured at least in 10s of megabytes) and operates on it serially, from front to back, your bus and memory bottlenecks start to show through.
It's kind of analogous to having a car with a top speed of 250 MPH and a 0-60 time of four minutes. On the highway, once you get up to speed, you'll cruise along nicely. (Think of that as big serial computations.) But in stop-and-go traffic in the city, you're sucking. (Typical branching programs that depend on user input.)
It's not that a hot PowerBook is a source of worry; it's that it hurts!
I swear, the only way it could be more painful is if Apple had decided to make the case out of copper....
If I could get a OSX native copy of Quark, Photoshop & Illustrator we would switch all of our OS9 desktops to OSX immediatly.
Illustrator X, which is either Cocoa or Carbon-- not sure which-- is available now. I've used the beta; it's nifty.
But I'm an old-timer. I use Photoshop 5.0, Quark 4, and Illustrator 8.0.1 in Classic mode under X. They're actually a lot more reliable under X than they are under 9. And under 10.1 on my Macs, Classic apps feel just as fast as native apps.
Bye-bye, karma. I know this is off topic, but...
I'm saving my pennies for a new 667MHz tiBook.
I'd advise you to save about 150,000 of those pennies and buy an iBook instead. My best friend has a PowerBook G4, and I have an iBook, and while the big screen on the G4 is nice, we both agree that my iBook is a better laptop.
That nice titanium case on the G4 scratches and scuffs incredibly easily, and it gets very very hot. Not to mention the fact that the slight flex in the G4's case makes it all too likely for a spinning CD or DVD to grind against the inside of the drive; it happens to my friend about once every other day.
My iBook, on the other hand, is a dream. I'd consider it to be *almost* good enough for an only machine, and perfect for a second machine.
Oh, and another thing. (feeble attempt to get back on topic) My iBook and my friend's PBG4 feel just about the same under OS 10.1 with 384 MB of RAM each. Both very, very usable.
(Sorry in advance for the long post.)
I thought in the good US of A that all projects that the government does the people of the US of A had access to the source [u]nless [i]t was deemed that it was endangering security of the nation [o]r [i]t was contracted out to a company and then they had all the IP
Both of those things is usually true.
For two years I worked as a consultant for a company that built training simulators for the USAF, the Air Force Reserve, and several foreign military services (countries like Denmark, South Korea, Jordan, Egypt, and so on). This company's two big products were an F-16 tactical simulator and an F-18 mission simulator.
Some fairly significant parts of the simulator runtime code are classified. As an example, some configurations of the F-16 can be equipped with the AIM-120 AMRAAM radar-guided missile. The code that handles the capabilities of the AMRAAM, and its interactions with the mission control systems, and its dynamics in the air is all classified.
(The details of this code is classified; the existence of it isn't. At least, I hope not. Otherwise, I'm in a shit-load of trouble right now.)
So obviously classified code can't be open-sourced.
The code that wasn't classified (a lot-- if you replaced classified modules with stubs, it was possible to run the whole F-16 load in unclassified mode; we did that a lot for visiting suits and stuff) was proprietary. In some cases, it was highly proprietary.
My example here is the F-16's mission control computer. The source code for this computer's programs was provided to us by Lockheed; we translated it line-by-line, mostly by hand, from assembly language into Ada-83 and compiled it to run on the sim's SGI Onyx host computer. This module was basically the core of the simulator, and it was of course a closely guarded commercial secret, even though it wasn't technically classified by the DOD.
We did something similar with the F-18's mission computer programs, but instead of translating them, we ran them natively in a Motorola processor emulator on the SGI host. This was kind of a cluster f*ck; it took 17 MIPS CPUs to emulate the two Motorola processors and the one 1553 mux bus controller in the MCC in real time. But somebody decided it was cheaper to throw hardware at it than to translate Boeing's code.
The other distinctive thing about US military flight sims-- at least the two I worked closely on, and also the F-22 tactical sim with which I worked a little-- is that they're not generic flight-dynamics simulators. Rather than taking the programmed characteristics of a wing or an airframe, like it sounds like X-Plane does, these sims were built with the full knowledge of the aircraft's flight characteristics. So it would be completely impossible to take the F-16 Block 30 code, change a data file, and have an F-16 Block 42 sim, much less a space shuttle sim or a 767 sim or whatever. These apps just weren't built like that.
A lot more goes into a tactical or mission training sim than just flight dynamics, anyway. I'd guess that maybe one out of five modules in the F-16 sim dealt with flying the plane; less than that in the F-18 sim. The rest was cockpit interface drivers (we had a real cockpit, with hundreds of individual hardware devices, wired into the sim; the serial mux control code was impressive) and inter-sim communication (DIS [defense information systems] and HLA [high-level architecture] protocols) and image generation and tactical DCS (distributed coordinate system) databases and the operator/instructor interface and it goes on and on and on. These things would only be relevant in context of a military tactical or mission sim, flown by military pilots in training, in a military installation with military instructors and other military sims connected over the military's encrypted wide-area training network.
I hope that answered your question, at least in part.
I saw this monitor last November in Dallas, at SuperComputing 2000. It was sort of stashed away inside the IBM booth behind some of their big iron. It was big and bright and sharp, and I got the story of its origin from an IBM guy standing nearby.
Warning: the person who told me this may have been a salesman. I can't claim to know this to be absolutely true.
According to the IBM guy, the folks from Livermore National Labs wanted, for some reason related to monitoring or surveillence or something like that, a monitor that could display four HDTV-resolution images in a tile. IBM tiled four 1920x1200 images (HDTV's 1920x1080 fits nicely inside the 1920x1200 display standard) on one monitor and sold bunches of them to LLNL for a red-blooded American fortune.
At that time, IBM called the monitor "Big Bertha." That was the official name and everything; they had data sheets printed up to hand out at the show.
And everything everybody has said so far is true: at that kind of resolution, your desktop icons are about a quarter of an inch across. And xterms? Forget it. You've gotta set the font size to 36 points just to be able to read it comfortably!
But then they IBM guy opened up a full-color satellite image of some city or other-- I forget which one. He full-screened it, and then used the mouse to pan and rotate around it. I actually got dizzy; it was like looking through a window. It was AMAZING. I've never, ever seen anything like that before.
Of course, to push about 10 million full-color pixels around in real-time like that required something more that a $99 graphics card; the monitor was hooked up to an SP node or something similarly impressive.
But damn, what a show.
I would say to Salon's editors: you sold out.
I guess by "sold out" you mean, "you're placing too much of an emphasis on feeding your families and not enough on entertaining ME."