What really puzzles me is that these intrustive ads clearly do anger readers, and don't seem to work very well...yet this arms race of distracting ads continues unabated.
The reason for this fact should be obvious. The only feedback available from an ad is positive feedback: if you click the link, the advertiser knows it. They don't know why you clicked it-- maybe it was for a product you liked, or maybe it was an accident on your part. But the advertiser knows you clicked it, so another tick mark is added to that ad's score column.
If you don't click on the ad... nothing happens. The advertiser has no way of knowing whether you didn't click because you're behind a filtering proxy, or because you were offended by the ad, or because your browser crashed. There's no negative feedback mechanism here at all.
Maybe if web ads were focus-grouped like TV commercials are, advertising companies might have a better idea of how the public at large is reacting to their ads.
On the other hand, if somebody could somehow demonstrate that pop-under (or whatever) ads actually have a measurable negative impact on company revenues, that'd be another story.
(Copied shamelessly from here in the hopes that some of you might read it before forming an opinion. Emphasis, where used, is mine.)
About our new ads
A note to readers
Sept. 24, 2001 | Today Salon introduces a new kind of advertisement -- a full-screen message that will show up in your browser when you click on a link, and will play briefly before moving you on to the page you requested. (The ad should only show up once per day per user, unless you have turned "cookies" off in your browser.)
As most of you know, this has been a difficult year for advertising-supported publications, online and off. Like many other companies we've responded by trying to innovate for our advertisers -- so we can remain financially healthy and continue to serve you. As with any innovation, we expect to learn from our experience over time, to keep what works and drop what doesn't.
We know that some Web users find this sort of ad intrusive. But before you send in that irate e-mail, we ask you to consider that the content you come to Salon for -- independent-minded, thought-provoking, unavailable elsewhere -- does not come free.
Today we have two ways to support our writers, editors and the rest of the staff that keeps Salon coming to you every day -- through advertising and through subscriptions. If sitting through one five-second ad before you can read an article is simply too much of a delay for you, we offer a Salon Premium subscription as a different way to support Salon -- you get access to exclusive content and the option to turn off most ads on the site. (For more information, click here.)
Our intention, as always, is to bring you the most intelligent, provocative, fearless coverage of news and culture available anywhere.
There are three ABIs: o32, n32, and 64. From the abi man page:
When an n32 program executes, IRIX sets the FR bit in the status register of the MIPS microprocessor to 1, just as it does for an n64 program. Under this setting, all 32 double precision floating point registers are enabled. This has the same performance and compatibility implications as with n64.
The only functional difference between n32 and 64 is the size of the pointer. Under n32, sizeof(char*) is 32 bits, and under 64 it's 64 bits. That means you can address more than 2 GB of memory with the 64 ABI.
Under both ABIs, you have access to the 128-bit "long double" type. Under n32, a "long" is an "int" (32 bits), while under 64 a "long" is a "long long" (64 bits).
What this means is this: if you're careless, and use the "long" type in your source code, the same program compiled for the n32 ABI will probably be faster than the 64 ABI version, because you'll be able to fit more "longs" into cache.
The moral of the story? Don't use the "long" type. If you need 32 bits, use an int, and if you need 64, use a long long.
You know, an iMac with OS X meets most of those criteria.
Doesn't have loud cooling fans
No fans in an iMac.
Doesn't have noisy disks
Okay, the hard drive in my iMac can be annoying, but there are two options. Either I could replace it with a quieter one, or (my choice) just set the disk to spin down when not in use. Silence is golden.
Can be powered off at any time and powers up in seconds
Hit the power button on the front, and the computer goes to sleep in about five seconds. Waking up under OS X takes about one second. By the time the screen lights up, the computer is ready for use.
Can talk TCP/IP over ethernet to my other machines
Yup. OS X is based on Darwin, which is based on a BSD-type core.
Is compact enough that I can leave it on a corner table
Your definition of compact may differ from mine, but I find my iMac to be pretty small, especially in light of the features and costs.
Runs free software (without a month of hacking on my part) so I know I'm not getting locked in to whatever the manufacturer wants to force on me in the future
If by "runs free software" you mean, "can run only free software," then install LinuxPPC or Yellow Dog or whatever. But if you're happy with OS X but want free apps, go right ahead. There's a big effort underway to port lots of free (gratis) software to Darwin-slash-OS X.
I work with big-ass computers all day-- IRIX, Linux, Windows, whatever. When I go home, I want to read my email and browse the web, play the occasional game or two, and sometimes log in to the office with XFree86 to work from home. I can do all those things on my iMac with no big BS. So it works for me.
Actually, I'm using 64-bit IRIX on a bunch of computers in my lab right now. Everything in the SGI MIPS product line from the Octane on up-- everything but the O2/O2+-- runs a 64-bit OS.
But the beauty of IRIX is that 32-bit code runs alongside 64-bit code without any of that emulation stuff the article talked about. In fact, big chunks of the OS are compiled in 32-bit mode. As many have already said, only apps that need to deal with more than 2 GB of virtual memory need to be compiled with 64-bit pointers.
Sure, calling it a "conviction" and using what sounds like legal jargon makes it sound pretty dramatic, but whatever words you choose, it is what it is: action taken against an individual without due process.
A fundamental assumption in the US is that no person shall be relieved of his property or liberty without due process of law (the fifth amendment to the US Constitution). Technically, this guarantee only applies to the federal government, but the principle is a strong one anyway.
Re:Making do without flippy floppy
on
Case Tweaking
·
· Score: 2
The old Mac Classic could even boot from ROM (they had a complete version of the system software built onto ROM) for complete diskless operation.
I'm not entirely sure that's correct. It's true that older Macs had the MacOS Toolbox (the OS runtime, basically) in ROM, but I don't believe any Mac had the ability to boot to ROM directly.
On the other hand, Macs have had NetBoot-- which is basically bootp for diskless network booting-- at least since the debut of the iMac. Never used it myself, but it's cool that its there, built right into the hardware.
I really hate this kind of comparison, but I just couldn't let this stand.
If your $999 config is like the one I put together at the Dell web site (who do you have to do to get to the Dell online store, anyway? sheesh.) then it lacks an Ethernet card and FireWire.
Sure, an Ethernet card can be had for a few bucks, and FireWire cards aren't much more, but the point is, I have to buy them separately and add them. That takes an investment of time and money that I don't think I should have to spend.
I look at buying a home computer (as opposed to my loaded-out workstation at the office, which is an entirely different kettle of fish) the same way I look at buying a car, or a TV. I plunk down my money, and it's done. I don't want to spend the rest of my weekend adding stuff, or configuring it, or installing drivers, or whatever. I don't consider that to be fun.
I've bought two iMacs now, and each time it was the same experience.
1. Plug computer in.
2. Turn computer on.
3. There is no step 3.
of bit rates and band widths
on
HDTV Over IP
·
· Score: 5, Interesting
Just a clarification.
When people say "HDTV," they can mean either the full-bandwidth uncompressed signal (on the order of 1.3 Gbps) or the broadcast-standard signal (MPEG-2 compressed to the order of 19.7 Mbit, or about 50-to-1, more or less).
Obviously this test didn't use uncompressed HDTV. Must have used MPEG-2.
When it comes to standard definition TV, the stuff you get over digital cable or DBS is typically between 4 and 6 Mbit. I think most people would consider 2 Mbit to be unacceptably noisy... but then again, I can ignore an awful lot of softness and artifacting from my TiVo, so maybe even 2 Mbit would be acceptable under the right circumstances.
Uncompressed standard-def TV, on the other hand, is carried over a 270 Mbit signal.
MPEG-2 compression seems to be totally acceptable up to 50-to-1, and marginally so up to about 100-to-1. DVCpro 25 (25 Mbit, or about 10-to-1) is widely considered to be crappy by broadcast standards, but looks a damn sight better than my TiVo on my home TV.
My rambling point (coffee, please) is that "HDTV" is a soft, fuzzy concept. Squeeze it down to 5 or 6 Mbit and it'll still be HDTV, with a thousand lines of resolution on-screen. But it might be so fuzzy or artifact-y that nobody would watch it.
I got Southwestern Bell DSL service about two years ago, through Southwestern Bell Internet. Cheap and fast and (believe it or not) they delivered on time.
About one year ago, they switched my service from DHCP to PPPoE without even telling me. One day my computer simply stopped getting an IP addres from the server, and no one could tell me why. It took over three months of calls to tech support before we figured out what the problem was.
(At this time, I will refrain from launching into a skreed about the state of SWB Internet tech support. Those poor pimply-faced high school kids have enough troubles without my criticism.)
I had no idea how terrible the PPPoE client for Macintosh was going to be. I fiddle with network configs all day long; I bought an iMac so I wouldn't have to do it at home, too.
Needless to say, I cancelled their service shortly thereafter and moved over to another ISP.
I can't say whether my experience is typical for
the SBC "family of companies," but it was enough to get me to sign up with a different service provider for almost twice as much per month.
And every month I send them their check with a big smile on my face.
I'd bet a E10K or S/390 could handle that kind of I/O. If you meant something you could actually afford - probably not.
Oh, you'd be surprised. You can do HDTV editing-- 1080i resolution-- with just 32 Seagate fibre channel disks on four FC loops. That's enough bandwidth to do RGB HDTV (186 MB/s, more or less) times two. (Gotta do two for real-time dissolves and such.)
To do more I/O is really just a matter of buying more disks.
What's that you say? You need your operating system to be able to efficiently balance multi-gigabit-per-second I/Os across several I/O channels on several busses? Oh, that's a software problem. I believe the question was originally about hardware.;-)
Actually, the not-uncommon 1080p resolution (1920x1080@60p) is an awfully good compromise for film work. Traditional film work is done at 2K resolution-- 2048 pixels across by however-many (depending on your aspect ratio). Some 4K work is done sometimes, but very, very rarely.
With high-definition telecines and recorders becoming more common, many post houses are doing their film finishing in 1920x1080, 24-frames-per-second progressive-scan. The data rate is a lot lower than 2K, and a HELL of a lot lower than 4K, and you can display your work on commercially available HDTV broadcast monitors-- the 32" model from Sony comes in well under $100,000.
I don't know where this Douglas Gray guy got his information, but he got a lot of it wrong, wrong.
As far as I can figure, the display he's talking about is IBM's Big Bertha. It was custom-built for Laurence Livermore national lab, and it runs at a native resolution of 3840x2400.
I saw a prototype of this display at Supercomputing 2000 in Dallas last year. It was running off of an IBM-brand Wintel system-- can't recall which one, an Intellistation, I guess-- with four 1920x1200 graphics cards. The monitor was stitching the four images together seamlessly.
According to rumor they hooked it up to their bigger iron from time to time, but when I saw it, it was running NT.
So I don't know *where* the author got his "it takes 16 CPUs and costs $200,000" stuff. Hell, LLNL only paid $80,000 for the prototype-- see this Federal Computing Week article. According to the IBM guy I talked to at SC2000-- although I can't seem to find a confirmation of this in writing anywhere-- when the monitor is commercialized sometime this year, they're expecting to sell it initially for about $20,000. One too many zeros, Doug.;-)
And the obligatory remark: yeah, it's an incredible display. Like reading a newspaper-- effectively about 200 ppi. But for any traditional computer application, it's not really practical. Once you get past the "wow" factor, this thing really lives up to its nickname: the IBM Squintron 2000.
The sole reason Photoshop is faster on Macs than PCs (recently), and the reason Steve always brings it out to show off, is altivec.
You've got your cart before your horse. Just like a Photoshop "benchmark" is useless to you, a gcc "benchmark" is useless to media professionals who care about things like Gaussian blurs and MPEG encoding. How many Macs do you think Apple would sell with an ad campaign centered around how fast they compile kernels?
When you hear Apple say that a Mac is twice as fast as an Intel system, just assume they're talking about the kinds of tasks their target market would care about.
Just a clarification: Bigelow says he expects to be able to launch the first module in thirty months. It's a long way from a single module to an inhabitable space station.
Robert Bigelow, who made his fortune as owner of Budget Suites of America, said he anticipates that his Bigelow Aerospace division will be able to launch a full-size space station module into orbit within 30 months.
It would take three such modules linked together to create a space station the size of the current International Space Station, he said.
No disrespect intended, but do you really know what you're talking about? Several of your comments sound like they came right out of 1994.
Ease of installation is hideous too. There's umpteen dependencies to (manually) resolve for doing the most trivial of things.
I'm not sure what you're comparing to. RPM? Please. Granted, IRIX is a very big OS-- six CDs for the runtime only; add more for the development stuff. Dependencies across discs are inevitable. But I don't think "Package eoe.sw.foo cannot be installed because of missing prerequisites: eoe.sw.bar (1270000000-1290000000). Please insert the IRIX 6.5 Foundation 1 CD." qualifies as hideous, exactly.
Nothing seems to come by default (including NFS)
NFS used to be optional-- the story goes that it had to do with SGI's licensing agreement with Sun-- but has been a bundled part of the OS since the first 6.5 release two years ago.
which compiler do I want - is it the "Ansi C compiler", the "C compiler (ANSI)" or the "C compiler".
Once again, your complaint may have been valid many years ago, but for as long as I've been using SGIs-- since about '96-- you have exactly two choices of C compiler: the MIPSpro C compiler from SGI, and GCC, also available compiled for IRIX from SGI. What's the problem?
I'm sure everybody who's read my comments knows I'm an SGI apologist. I won't try to tell you they're perfect, but with so much actual stuff to complain about, you've got even less excuse for complaining about problems that haven't existed for years!
O2 - aside from CPU upgrades, has remained unchanged since the fall of 1996
There's been a lot of talk about this. The short version is that the people who buy O2s-- like the Weather Channel, for instance-- don't want 'em changed. They're fine the way they are. Otherwise the product would have been gone a long time ago.
All of the above machines, while featuring expansion, only have U/W 40MB/sec onboard SCSI... getting a bit old for modern 10K and 15K RPM drives.
Who cares? If you want performance storage on one of these machines, use Fibre Channel externally. It's far more cost-effective when you compare dollars to gigabytes-per-second. That's what everybody does.
Yes, PCs are more popular than Macs because they had an economy of scale that the Mac didn't have. But do you think anybody would be using PCs-- or Macs, or whatever-- if their utility hadn't grown beyond the level of home-written chess programs and Trade Wars?
Love 'em or hate 'em, Microsoft created MS Office, and MS Office is why tens of umpteen millions of computers are sold every year.
Sure, if it hadn't been MS it would have been somebody else, but let's not forget what drove the mass production of relatively standardized computers like the ones we use to run our hobby projects on.
Lots of discussion about testing servers. If you only have one server, consider running a virtual machine on it for testing. You can evaluate VMware for, I think, a month, and that should give you enough time to figure out what you want to do. Since VMware is fairly inexpensive, you might think about keeping it running for just this sort of thing.
On my server, I have a VMware GSX license, and I keep a clean virtual-disk with RedHat 7.1 + XFS (my dist of choice) on line at all times. When I need to test something-- anything, an app, a new kernel, whatever-- I clone off the test virtual-disk, boot the VM, and go.
24-bit color is visually identical to 32-bit color; the extra byte is simply to speed up accesses by aligning pixels on 32-bit boundaries.
You're mostly right. The extra byte is often used for other things, like an 8-bit alpha channel, or four bits of alpha plus four bits of Z-buffer.
But you're main point is correct: 8-bit-per-channel RGB is what we're talking about.
Makes me wonder, though. The de facto standard in the cinema world is 12 bits per pixel per channel-- 36 bit color. I wonder why no vendor has sold a PC-class product that does 36-bit color?
Cool trivia: the smallest pixels you can use on the Onyx2 IR2 system I have in my lab are 128 bits deep. That's 12 bits per channel RGBA, times two (double buffered) plus 32 bits of Z-buffer. Yowzers.
Aside from the fact that their patients would glow in the dark in no time flat, why does he think that X-rays would go through some soft tissue (skin), but not others (heart)? X-rays are only reflected by bone, just like we've known about for over a century.
Actually, X-rays are absorbed by all tissues, but the ratio of transmission to absorption varies with the density of the medium.
Bone shows up on an X-ray negative as white; so does tumor tissue, usually, because it's also very dense. Soft tissues like organs are rendered in all shades of gray. Fluids show up too; X-rays are really useful for finding abnormal fluid collections, like a collection of blood around the heart (called a cardiac tampenade). A trained radiologist can identify subtle structures from a plain, ordinary chest radiograph; you just have to know what to look for.
What the author wrote, that "doctors place heart patients in front of a fluoroscopic screen, turn on the X-rays and then, with the aid of a photoelectric cell, examine every section of the heart," is a pretty accurate description of a cardiac angiogram. Radiologists have been doing film angiograms for years-- basically using a motion picture camera instead of a static camera when shooting X-rays of contrast material flowing through blood vessels. In some hospitals. CCD-based X-ray cameras are replacing film cameras, so the author's prediction is actually dead-on correct.
While I agree with the substance of your message, you've got a couple of facts wrong.
First, the maximum filesystem size that XFS can handle is 18 exabytes. Since exabyte is also the name of a brand of tape drive, it's more common to hear of people talking of 18 million terabytes.
(Aside: an intresting statistic found on this page says that as of 1995, 5 million terabytes was about enough data to store all words ever spoken by humans, ever. Cool.)
Also, XFS was never used on Cray systems. XFS made its first appearance (if I remember correctly) on IRIX 5.3 back in 1994.
Other than being off by four orders of magnitude in your sizing and by ten years in your dates, I think you're exactly right.;-)
Re:Finally, something resembling clustering for Li
on
Mosix 1.0 Released
·
· Score: 1
Beowulf (distributed parallel computing environment), while an interesting and useful technology, has nothing whatsoever to do with clustering!
Agreed. Beowulf is really more of an array computing model than stuff that is traditionally referred to as cluster-computing.
Back in olden times-- circa 1995, I think-- SGI had a set of software and an API for their Challenge servers that allowed customers to configure them as arrays. I can't remember the marketing-name of the product, but I do remember that parallel HIPPI was the preferred interconnect (still damn fast, six years later) and some customers had arrays consisting of a great many 36-processor nodes. Pretty cool product. Very similar in concept to a Beowulf cluster using gigabit-e or Myrinet as the interconnect.
Slashdot Mirror
What really puzzles me is that these intrustive ads clearly do anger readers, and don't seem to work very well...yet this arms race of distracting ads continues unabated.
The reason for this fact should be obvious. The only feedback available from an ad is positive feedback: if you click the link, the advertiser knows it. They don't know why you clicked it-- maybe it was for a product you liked, or maybe it was an accident on your part. But the advertiser knows you clicked it, so another tick mark is added to that ad's score column.
If you don't click on the ad... nothing happens. The advertiser has no way of knowing whether you didn't click because you're behind a filtering proxy, or because you were offended by the ad, or because your browser crashed. There's no negative feedback mechanism here at all.
Maybe if web ads were focus-grouped like TV commercials are, advertising companies might have a better idea of how the public at large is reacting to their ads.
On the other hand, if somebody could somehow demonstrate that pop-under (or whatever) ads actually have a measurable negative impact on company revenues, that'd be another story.
(Copied shamelessly from here in the hopes that some of you might read it before forming an opinion. Emphasis, where used, is mine.)
About our new ads
A note to readers
Sept. 24, 2001 | Today Salon introduces a new kind of advertisement -- a full-screen message that will show up in your browser when you click on a link, and will play briefly before moving you on to the page you requested. (The ad should only show up once per day per user, unless you have turned "cookies" off in your browser.)
As most of you know, this has been a difficult year for advertising-supported publications, online and off. Like many other companies we've responded by trying to innovate for our advertisers -- so we can remain financially healthy and continue to serve you. As with any innovation, we expect to learn from our experience over time, to keep what works and drop what doesn't.
We know that some Web users find this sort of ad intrusive. But before you send in that irate e-mail, we ask you to consider that the content you come to Salon for -- independent-minded, thought-provoking, unavailable elsewhere -- does not come free.
Today we have two ways to support our writers, editors and the rest of the staff that keeps Salon coming to you every day -- through advertising and through subscriptions. If sitting through one five-second ad before you can read an article is simply too much of a delay for you, we offer a Salon Premium subscription as a different way to support Salon -- you get access to exclusive content and the option to turn off most ads on the site. (For more information, click here.)
Our intention, as always, is to bring you the most intelligent, provocative, fearless coverage of news and culture available anywhere.
Scott Rosenberg
Managing editor
The only functional difference between n32 and 64 is the size of the pointer. Under n32, sizeof(char*) is 32 bits, and under 64 it's 64 bits. That means you can address more than 2 GB of memory with the 64 ABI.
Under both ABIs, you have access to the 128-bit "long double" type. Under n32, a "long" is an "int" (32 bits), while under 64 a "long" is a "long long" (64 bits).
What this means is this: if you're careless, and use the "long" type in your source code, the same program compiled for the n32 ABI will probably be faster than the 64 ABI version, because you'll be able to fit more "longs" into cache.
The moral of the story? Don't use the "long" type. If you need 32 bits, use an int, and if you need 64, use a long long.
(Bye-bye, karma.)
You know, an iMac with OS X meets most of those criteria.
Doesn't have loud cooling fans
No fans in an iMac.
Doesn't have noisy disks
Okay, the hard drive in my iMac can be annoying, but there are two options. Either I could replace it with a quieter one, or (my choice) just set the disk to spin down when not in use. Silence is golden.
Can be powered off at any time and powers up in seconds
Hit the power button on the front, and the computer goes to sleep in about five seconds. Waking up under OS X takes about one second. By the time the screen lights up, the computer is ready for use.
Can talk TCP/IP over ethernet to my other machines
Yup. OS X is based on Darwin, which is based on a BSD-type core.
Is compact enough that I can leave it on a corner table
Your definition of compact may differ from mine, but I find my iMac to be pretty small, especially in light of the features and costs.
Runs free software (without a month of hacking on my part) so I know I'm not getting locked in to whatever the manufacturer wants to force on me in the future
If by "runs free software" you mean, "can run only free software," then install LinuxPPC or Yellow Dog or whatever. But if you're happy with OS X but want free apps, go right ahead. There's a big effort underway to port lots of free (gratis) software to Darwin-slash-OS X.
I work with big-ass computers all day-- IRIX, Linux, Windows, whatever. When I go home, I want to read my email and browse the web, play the occasional game or two, and sometimes log in to the office with XFree86 to work from home. I can do all those things on my iMac with no big BS. So it works for me.
But the beauty of IRIX is that 32-bit code runs alongside 64-bit code without any of that emulation stuff the article talked about. In fact, big chunks of the OS are compiled in 32-bit mode. As many have already said, only apps that need to deal with more than 2 GB of virtual memory need to be compiled with 64-bit pointers.
A fundamental assumption in the US is that no person shall be relieved of his property or liberty without due process of law (the fifth amendment to the US Constitution). Technically, this guarantee only applies to the federal government, but the principle is a strong one anyway.
I'm not entirely sure that's correct. It's true that older Macs had the MacOS Toolbox (the OS runtime, basically) in ROM, but I don't believe any Mac had the ability to boot to ROM directly.
On the other hand, Macs have had NetBoot-- which is basically bootp for diskless network booting-- at least since the debut of the iMac. Never used it myself, but it's cool that its there, built right into the hardware.
If your $999 config is like the one I put together at the Dell web site (who do you have to do to get to the Dell online store, anyway? sheesh.) then it lacks an Ethernet card and FireWire.
Sure, an Ethernet card can be had for a few bucks, and FireWire cards aren't much more, but the point is, I have to buy them separately and add them. That takes an investment of time and money that I don't think I should have to spend.
I look at buying a home computer (as opposed to my loaded-out workstation at the office, which is an entirely different kettle of fish) the same way I look at buying a car, or a TV. I plunk down my money, and it's done. I don't want to spend the rest of my weekend adding stuff, or configuring it, or installing drivers, or whatever. I don't consider that to be fun.
I've bought two iMacs now, and each time it was the same experience.
1. Plug computer in.
2. Turn computer on.
3. There is no step 3.
When people say "HDTV," they can mean either the full-bandwidth uncompressed signal (on the order of 1.3 Gbps) or the broadcast-standard signal (MPEG-2 compressed to the order of 19.7 Mbit, or about 50-to-1, more or less).
Obviously this test didn't use uncompressed HDTV. Must have used MPEG-2.
When it comes to standard definition TV, the stuff you get over digital cable or DBS is typically between 4 and 6 Mbit. I think most people would consider 2 Mbit to be unacceptably noisy... but then again, I can ignore an awful lot of softness and artifacting from my TiVo, so maybe even 2 Mbit would be acceptable under the right circumstances.
Uncompressed standard-def TV, on the other hand, is carried over a 270 Mbit signal.
MPEG-2 compression seems to be totally acceptable up to 50-to-1, and marginally so up to about 100-to-1. DVCpro 25 (25 Mbit, or about 10-to-1) is widely considered to be crappy by broadcast standards, but looks a damn sight better than my TiVo on my home TV.
My rambling point (coffee, please) is that "HDTV" is a soft, fuzzy concept. Squeeze it down to 5 or 6 Mbit and it'll still be HDTV, with a thousand lines of resolution on-screen. But it might be so fuzzy or artifact-y that nobody would watch it.
About one year ago, they switched my service from DHCP to PPPoE without even telling me. One day my computer simply stopped getting an IP addres from the server, and no one could tell me why. It took over three months of calls to tech support before we figured out what the problem was.
(At this time, I will refrain from launching into a skreed about the state of SWB Internet tech support. Those poor pimply-faced high school kids have enough troubles without my criticism.)
I had no idea how terrible the PPPoE client for Macintosh was going to be. I fiddle with network configs all day long; I bought an iMac so I wouldn't have to do it at home, too.
Needless to say, I cancelled their service shortly thereafter and moved over to another ISP.
I can't say whether my experience is typical for the SBC "family of companies," but it was enough to get me to sign up with a different service provider for almost twice as much per month.
And every month I send them their check with a big smile on my face.
I would say RTFA, but the FA completely failed to link to any relevant info. I had to search myself; IBM's page is here.
Oh, you'd be surprised. You can do HDTV editing-- 1080i resolution-- with just 32 Seagate fibre channel disks on four FC loops. That's enough bandwidth to do RGB HDTV (186 MB/s, more or less) times two. (Gotta do two for real-time dissolves and such.)
To do more I/O is really just a matter of buying more disks.
What's that you say? You need your operating system to be able to efficiently balance multi-gigabit-per-second I/Os across several I/O channels on several busses? Oh, that's a software problem. I believe the question was originally about hardware. ;-)
Actually, the not-uncommon 1080p resolution (1920x1080@60p) is an awfully good compromise for film work. Traditional film work is done at 2K resolution-- 2048 pixels across by however-many (depending on your aspect ratio). Some 4K work is done sometimes, but very, very rarely.
With high-definition telecines and recorders becoming more common, many post houses are doing their film finishing in 1920x1080, 24-frames-per-second progressive-scan. The data rate is a lot lower than 2K, and a HELL of a lot lower than 4K, and you can display your work on commercially available HDTV broadcast monitors-- the 32" model from Sony comes in well under $100,000.
As far as I can figure, the display he's talking about is IBM's Big Bertha. It was custom-built for Laurence Livermore national lab, and it runs at a native resolution of 3840x2400.
I saw a prototype of this display at Supercomputing 2000 in Dallas last year. It was running off of an IBM-brand Wintel system-- can't recall which one, an Intellistation, I guess-- with four 1920x1200 graphics cards. The monitor was stitching the four images together seamlessly.
According to rumor they hooked it up to their bigger iron from time to time, but when I saw it, it was running NT.
So I don't know *where* the author got his "it takes 16 CPUs and costs $200,000" stuff. Hell, LLNL only paid $80,000 for the prototype-- see this Federal Computing Week article. According to the IBM guy I talked to at SC2000-- although I can't seem to find a confirmation of this in writing anywhere-- when the monitor is commercialized sometime this year, they're expecting to sell it initially for about $20,000. One too many zeros, Doug. ;-)
And the obligatory remark: yeah, it's an incredible display. Like reading a newspaper-- effectively about 200 ppi. But for any traditional computer application, it's not really practical. Once you get past the "wow" factor, this thing really lives up to its nickname: the IBM Squintron 2000.
You've got your cart before your horse. Just like a Photoshop "benchmark" is useless to you, a gcc "benchmark" is useless to media professionals who care about things like Gaussian blurs and MPEG encoding. How many Macs do you think Apple would sell with an ad campaign centered around how fast they compile kernels?
When you hear Apple say that a Mac is twice as fast as an Intel system, just assume they're talking about the kinds of tasks their target market would care about.
Robert Bigelow, who made his fortune as owner of Budget Suites of America, said he anticipates that his Bigelow Aerospace division will be able to launch a full-size space station module into orbit within 30 months.
It would take three such modules linked together to create a space station the size of the current International Space Station, he said.
Ease of installation is hideous too. There's umpteen dependencies to (manually) resolve for doing the most trivial of things.
I'm not sure what you're comparing to. RPM? Please. Granted, IRIX is a very big OS-- six CDs for the runtime only; add more for the development stuff. Dependencies across discs are inevitable. But I don't think "Package eoe.sw.foo cannot be installed because of missing prerequisites: eoe.sw.bar (1270000000-1290000000). Please insert the IRIX 6.5 Foundation 1 CD." qualifies as hideous, exactly.
Nothing seems to come by default (including NFS)
NFS used to be optional-- the story goes that it had to do with SGI's licensing agreement with Sun-- but has been a bundled part of the OS since the first 6.5 release two years ago.
which compiler do I want - is it the "Ansi C compiler", the "C compiler (ANSI)" or the "C compiler".
Once again, your complaint may have been valid many years ago, but for as long as I've been using SGIs-- since about '96-- you have exactly two choices of C compiler: the MIPSpro C compiler from SGI, and GCC, also available compiled for IRIX from SGI. What's the problem?
I'm sure everybody who's read my comments knows I'm an SGI apologist. I won't try to tell you they're perfect, but with so much actual stuff to complain about, you've got even less excuse for complaining about problems that haven't existed for years!
There's been a lot of talk about this. The short version is that the people who buy O2s-- like the Weather Channel, for instance-- don't want 'em changed. They're fine the way they are. Otherwise the product would have been gone a long time ago.
All of the above machines, while featuring expansion, only have U/W 40MB/sec onboard SCSI... getting a bit old for modern 10K and 15K RPM drives.
Who cares? If you want performance storage on one of these machines, use Fibre Channel externally. It's far more cost-effective when you compare dollars to gigabytes-per-second. That's what everybody does.
I believe the SGI 230, 330, and 550 are Acer-built.
<flamebait>
Yes, PCs are more popular than Macs because they had an economy of scale that the Mac didn't have. But do you think anybody would be using PCs-- or Macs, or whatever-- if their utility hadn't grown beyond the level of home-written chess programs and Trade Wars?
Love 'em or hate 'em, Microsoft created MS Office, and MS Office is why tens of umpteen millions of computers are sold every year.
Sure, if it hadn't been MS it would have been somebody else, but let's not forget what drove the mass production of relatively standardized computers like the ones we use to run our hobby projects on.
</flamebait>
On my server, I have a VMware GSX license, and I keep a clean virtual-disk with RedHat 7.1 + XFS (my dist of choice) on line at all times. When I need to test something-- anything, an app, a new kernel, whatever-- I clone off the test virtual-disk, boot the VM, and go.
The eval download page for VMware GSX is here.
You're mostly right. The extra byte is often used for other things, like an 8-bit alpha channel, or four bits of alpha plus four bits of Z-buffer.
But you're main point is correct: 8-bit-per-channel RGB is what we're talking about.
Makes me wonder, though. The de facto standard in the cinema world is 12 bits per pixel per channel-- 36 bit color. I wonder why no vendor has sold a PC-class product that does 36-bit color?
Cool trivia: the smallest pixels you can use on the Onyx2 IR2 system I have in my lab are 128 bits deep. That's 12 bits per channel RGBA, times two (double buffered) plus 32 bits of Z-buffer. Yowzers.
Actually, X-rays are absorbed by all tissues, but the ratio of transmission to absorption varies with the density of the medium.
Bone shows up on an X-ray negative as white; so does tumor tissue, usually, because it's also very dense. Soft tissues like organs are rendered in all shades of gray. Fluids show up too; X-rays are really useful for finding abnormal fluid collections, like a collection of blood around the heart (called a cardiac tampenade). A trained radiologist can identify subtle structures from a plain, ordinary chest radiograph; you just have to know what to look for.
What the author wrote, that "doctors place heart patients in front of a fluoroscopic screen, turn on the X-rays and then, with the aid of a photoelectric cell, examine every section of the heart," is a pretty accurate description of a cardiac angiogram. Radiologists have been doing film angiograms for years-- basically using a motion picture camera instead of a static camera when shooting X-rays of contrast material flowing through blood vessels. In some hospitals. CCD-based X-ray cameras are replacing film cameras, so the author's prediction is actually dead-on correct.
Credit where it's due.
First, the maximum filesystem size that XFS can handle is 18 exabytes. Since exabyte is also the name of a brand of tape drive, it's more common to hear of people talking of 18 million terabytes.
(Aside: an intresting statistic found on this page says that as of 1995, 5 million terabytes was about enough data to store all words ever spoken by humans, ever. Cool.)
Also, XFS was never used on Cray systems. XFS made its first appearance (if I remember correctly) on IRIX 5.3 back in 1994. Other than being off by four orders of magnitude in your sizing and by ten years in your dates, I think you're exactly right. ;-)
Agreed. Beowulf is really more of an array computing model than stuff that is traditionally referred to as cluster-computing.
Back in olden times-- circa 1995, I think-- SGI had a set of software and an API for their Challenge servers that allowed customers to configure them as arrays. I can't remember the marketing-name of the product, but I do remember that parallel HIPPI was the preferred interconnect (still damn fast, six years later) and some customers had arrays consisting of a great many 36-processor nodes. Pretty cool product. Very similar in concept to a Beowulf cluster using gigabit-e or Myrinet as the interconnect.