For all the information you could ever want to know about how the new 300D/Digital Rebel compares to the other DSLRs that are out there just go check out photo.net. There is a full review of the body, plus lots of discussion about it in the forums.
Your second question, about whether or not to switch to digital, is not a question that we can answer for you, especially with the amount of information that you gave us. Both film and digital have their respective advantages. Both will continue to exist for quite some time. For a well thought out examination of film and digital photography, see Ken Rockwell's article on the subject.
What most people don't realize is that digital and film have been working together quite well for some time now and that the digital revolution has already made a huge impact in the printing phase. Lightjet and Chromira machines enable the highest quality prints and Fuji Frontier machines create good quality prints quickly. The quality of these prints is not just the resolution, but the color reproduction and tonality as well.
What it all comes down to, though, is not the equipment. You have to be in the right place when the light happens. Mastering light is far more important than having a certain kind of camera.
My experience with XDarwin has been fantastic. Apps under XDarwin seem so much faster and more responsive than they ever did under LinuxPPC, YDL or Debian on my 266mhz Powerbook. The Gimp runs great under XDarwin as well. My advice would be to install Fink and then use that to do all of the rest of your Unix apps. The fact that I have apt on my OS X system is just that much cooler. Stick with OS X and you won't be disappointed. This is advice coming from someone who faithfully used LinuxPPC for 3 years and has now switched wholeheartedly to OS X.
--neutrino
With the recent release of MacOS X what are your reactions to it and what plans do you have to compete with a truly user-friendly desktop OS combined with the stability of a UNIX backend?
--neutrino
There was a great article over at A List Apart a while ago that dealt with the problem of online advertising. The problem is far deeper than just targeting the audience better. The real problem is that banner ads, by their very nature, are not enticing to readers. What the online publications need to do is to make advertising an integral part of the site. One big problem is that banners are not constant. If I see an ad for Acme corporation and later want to learn more, I can't simply reload the page to find out about Acme. Instead, I will get another ad alltogether. How many times have you had that problem with a magazine? Also, banners usually have little or no actual information on them. This can't be good. You need information to be integrated with the ad. What needs to be done is more effort needs to be put into online advertising. Instead of regarding it as an additon to a site that pays the bills, it should be made an inseperable part of that site.
--neutrino
I am a huge supporter of Mozilla. It is my regular browser. I do have one wish for the more recent releases, though: Continue releasing binaries for alternate architectures. For the releases before 0.6 (all the Mxx releases), they pu up binaries for PPC, alpha and SPARC. They also released binaries for OS/2, HPUX and other more fringe oses. These weren't released at the same time as the Linux x86 and Windows binaries, but they were released. I know that I can compile it on my own machine (LinuxPPC), but their build host sits idle now instead of building other binaries. Just my thoughts, though.
--neutrino
The tree-based rendering pipeline is described in detail at the review page that I pointed to earlier. In it, they state:
What used to be a simple layer stack in GIMP 1.x, which is combined using layer modes (Normal, Combine, Difference,...) will, with the help of GEGL, become a rendering pipeline which can be thought of as a tree of layers which is viewed from its root. The nodes of the tree are operators with an arbitrary number of inputs and outputs. These inputs and outputs access rectangular regions of pixel-data, the edges of the tree. Each edge (comparable to the layers we have now) can hold its data internally as pixels, vectors, text or whatever and only needs to provide a well-defined interface so it can be plugged into the rendering pipeline. A similar approach will be used for the operators: Simple functions like color corrections or blur filters as well as affine transformations and more complex effects are possible.
So, the rendering pipeline becomes much harder to explain in simple terms. This new model, however, yields great flexibility and power. You can have an image that contains both pixel-based portions and vector scalable portions. These objects can then be modified at any time. So, unlike current GIMP, text could be added and then later changed. This is truly useful for speeding up many uses that people have for the GIMP, but it also makes for more new uses.
The changes that are planned for 2.0 were discussed at Gimpcon this summer. There is a review of the planned new features. For those who are too lazy to read the page, the following is a short rundown:
It will use the GEGL, or Generic Graphics Library for pixel manipulatio
The rendering pipepline will be tree-based instead of a simple stack
The program itself will become much more modular in design -- Model View Controller
Plugins and app will communicate via CORBA
Generally more flexible -- possible integration with vector-based tools and video editing capabilities
This is, as you can see, all backend work. Obviously, some changes will have to be made to the interface, but the result will be a vastly more powerful system. If they can successfully implement all of these ideas, then the GIMP would be, in many ways, the most technologically advanced image manipulation program in the world.
Also, you should see the code as it stands in the 1.x series. It is obvious that the program has evolved to its current state and was not designed to this point.
Theo,
I also am an avid rock climber and I was wondering what level you climb at and what you feelis your biggest climbing accomplishment. Do you do big wall or any mountaineering, or do you just do sport climbing and bouldering?
--neutrino
At the end of his atricle, he suggests using NS 4.x over version 6.0. I beg you not to. Why? Because then you are a part of the reason why today's web sucks. NS 6.0 is far more standards compliant than anything else on the market. NS 4.x is just the opposite. It doesn't just lack support for w3c standards, it implements them incorrectly. This causes web developers to have to
Code to the lowest common denominator
Make multiple versions of every page
or make their sites accessible only to certain browsers
And yet, so many/.ers cry afoul when they encounter a site that doesn't work in their browser, NS 4.x. The rendering engine in this decrepit old browser is 3+ years behind! If you don't like NS 6 or Mozilla, use something else. Opera or Konquerer or IE or anything but NS 4.x.
Additionally, the author of the article is incorrect when he states that NS 6 is less standards compliant than Opera. Test it for yourself. NS6/Moz is the most standards compliant browser out there.
The difference between the present and the days of the 3.0 browsers is that today the "standards" are innovative. CSS-2, DHTML, DOM and XML/XSL are technologies that empower developers to do far more than can be done today. No browser even supports a signifigant portion of CSS-2. CSS-2 and XML allow us to finally escape the shackles of presentation and content being intertwined. If you don't develope internet sites, this might seem trivial, but when you are down in the trenches trying to provide a site that looks good on a new and "innovative" browser, but at the same time works on a cell phone, you learn to appreciate the innovation that the standards have. Check out the amazing things that can be done with standards over at the W3C and then try and find a better innovation that is needed in some browser. --neutrino
I think that you should check out the newer specifications of which you speak more carefully. The introduction of CSS (Cascading Style Sheets) makes the web more accessible than ever. This is because the specs encourage the use of content based tags instead of physical tags. This means that an HTML page isn't littered with Bold and Italic tags, but instead has Emphasis and Strong tags. CSS2 goes even further by introducing tags that are designed for speech capable user agents. The end effect of CSS and hopefully the adoption of XML and XSL is that the web is more accessible to all users. These specifications also eliminate the need for hacks like using tables for layout and replace them with tools like are used in desktop publishing. Check out the W3C's UI section to learn more about this. Also, the new O'Reilly book, Cascading Style Sheets: The definitive Guide, is a great source for making accessible pages that look great too.
I agree with most of what you say, that most new technologies make pages more and more limited as far as who can view them, but this is not the case with CSS. A properly coded page that relies on CSS is actually better in Lynx, for instance, than an equivalence HTML page. Why? Well because you can avoid unnecessary usage of tags like in your document that you would use in straight HTML. Instead, you can create a new subclass of other tags which will give you the desired appearance but which will be treated appropriately by browsers that cannot produce the desired style. This is only with CSS 1. When CSS 2 support comes along, look out. It allows you to get rid of the biggest joke in HTML layout techniques, tables. Instead of dealing with putting every piece of content into a table cell, you define the layout in an intelligent way with a style sheet, just like professional publishing programs do. Similarly, when a browser comes along that cannot present it in the desired manner, the content is still readable, even if the browser doesn't understand CSS. Check it out, it is the way we'll be publishing on the web here real soon.
Actually, Linux does support PMU. Check out pmud. It's a daemon that can be used to spin down the hard drive and other PMU functions. The latest release, 0.5-1 is the most recent, but it is buggy on a Wallstreet G3 laptop (Won't wake up from sleep mode). Anyhow, check it out, though. The real problem, though is that Linux doesn't do so well with any spin down of the hard drive, as it wites to disk every few minutes, so much of the functionality is lost. Interesting nonetheless, though.
Of course you could also use a laptop that doesn't suck down the power. Why plop an x86 cpu into a laptop? Ever notice that most x86 desktops have enormous fans and heat sinks on them? That heat has to come from somewhere. Get an Apple Powerbook. PPC chips don't drain a battery like a PII does. Plus, then you could join us LinuxPPC users out here on the fringe! We don't need no stinking binaries!
There are a number of reasons why the route chosen was the MHz route. Firstly, consider Moore's Law. If the speed of processors doubles every 18 months, then is there a similar relation for # of Procs on a system? Let's say that back in '92 we started down the multiproc route instead of the MHz route. In order to be at our current speed, we would need 32 procs to keep the speed. Silicon ain't cheap, though. This shows us that the CPU cost alone is prohibitive.
Secondly, the support hardware is also more complex for multiproc setups. You're memory controller must make sure that there is cache coherency across all of the processors' caches. This makes the motherboard more costly. There are a number of other hardware issues that jack the price up.
Finally, and perhaps most importantly, there is the software aspect of multiple CPUs. Writing concurrent programs is far more difficult than their sequential relatives. You not only have to debug the program for proper behavior, which is difficult when there can be process interference, but you also have to debug speed issues that arise from the interaction of processes. When a sequential program is run on a faster CPU, if it is a CPU-bound problem, then it will result in faster execution. This is not necessarily the case when you add processors. Concurrent programming is a whole new bucket of monkeys.
There was one aspect of this scheme that I wonder about: The group's use of the Cray C916. I'm not sure what the policies at SARA are, but with the Crays that I work on, there are fairly severe penalties for using processor time for projects that aren't specifically approved. On one, in fact, I get the following message upon login:
WARNING U.S. GOVERNMENT COMPUTER If not authorized to access this system, disconnect now. YOU SHOULD HAVE NO EXPECTATION OF PRIVACY By continuing, you consent to your keystrokes and data content being monitored.
The rules differ between systems, but every one that I have used has some punishment for unapproved system usage. Since the article said that they did this in secret, I assume that their work falls into the unapproved category. Although, SARA's website seems to indicate a fairly open view of their system. Who knows. I certainly wouldn't risk doing it on US systems that I have access to.
I'm not sure that the important (ie limiting) factor here is the bandwidth nor even the amount of memory. The system that they quote as using, a Cray C916, is a computer that is no longer offered by SGI. It is a vector computing solution which has been succeeded by the T90 series. These systems only come in configurations of up to 32 processors, which is remarkable, but not compared to the real high end solutions (T3E and Origin 2000). I imagine that the statement about 2 Gigs of RAM was because that was the minimum space that the problem could reside in, but the true use of the Cray was the fact that for Polynomial searches, the paralleled vector processors are ideal for the job. The real possibility lies in the fact that a T932 could be used to do the same work in approximately 1/4 to 1/8 the time. If a group had access to a single T932 to do the Polynomial portion and a T3E for the rest, there is the possibility that they could do a similar factorization in under a week. If the reward was large enough (bank transactions) the cost of the hardware is easily covered. Hmm, large keys seem terribly desirable, no?
I really like the direction that SGI has taken lately(aside from the foolish name change). Their focus on open source has been great. Now, as a Cray T3e programmer, I have a request: Please update the MPI libraries to MPI2 compliance. For those of you who aren't familiar with parallel computing, MPI(Message Passing Interface) is the industry standad for parallel computing. It can be used on a single processor PC all the way up to a 2048 processor T3e with little porting overhead. The problem is that nearly every implementation is of the older 1.2 implementation. MPI2 allows for trmendous speedup. If both Cray and MPICH, a free implementation, go to MPI2 compliance, then Beowulf clusters could fit more easily into many research environments. As it currently stands, the code bases for Supercomputers and clusters remain seperat entities in many cases. MPI2 would solve this. Oh well. Just my obscure wish, though.
I'm curious why you don't like Mach. I think that it's a curious design. Admittedly, you take a speed hit when using it, but are there other reasons you don't like it?
The difference is much bigger than just sheer numbers. Lets give Dell the benefit of the doubt and say that they have twice as many browsers pointed at them. They serve up mostly static pages. The parts that are dynamic are not very dynamic. Now look at eBay. They have a site that is nearly 100% dynamic what with the very nature of their business. Add in the personalized things that users can set up and you have a system that is working several orders of magnitude more than a site with heavier traffic but static content. Now, I don't think that Dell has anywhere near the number of hits. eBay junkies (My parents are antique dealers) check the site almost constantly throughout the evening. Once you've seen a Dell laptop, you know what it comes with. eBay customers expect a constant change in the site and thus check every few minutes as auctions that they are interested in draw near a close. eBay is easily a more taxed site.
Yeah, they could crack it real quick. They could open up the safe that has the message written on it and read it, but that wouldn't be much fun, kind of like using their dedicated hardware wouldn't be much fun.
This list is accurate at what it is trying to measure. The list is for general-purpose supercomputers and so this eliminates a number of machines that are much faster for dedicated tasks. The NSA, for instance, has machines that are very good at the task of factoring. These machines, however, are not capable of performing general tasks, and thus are not candidates for this list. In addition, you may say that there are other general purpose supercomputers out there that are faster, but not on the list. This is doubtful, as the machines have to be made by someone with experience at the rather specialized task of making ultra-fast systems. All of the major producers of supercomputers recognize the importance of such benchmarks and thus work with people to do Linpack tests. The importance of these benchmarks is that the company that produces high results will get more sales. The only computers now left out of this list would be those produced in secret by governments or other such entities. These groups fall behind because they do not have the existing knowledge and resource base needed to produce such computers. The end result: This list is quite accurate with perhaps a few Ultra Secret machines left out.
Yeah, I have an account on the 13th fastest computer in the world. Nothing quite as fun as sitting down and allocating 512 Processors to work on the formation of a planetary system. The Unicos/MK OS on Cray's is a very interesting example of microkernel design that is well done. Of course, the optimizing compilers on that baby are even more astounding.
My main problem with BeOS is not the hardware. I needed to upgrade some components anyhow, but my problem, which I realize needs time to be solved, is that the BeOS is not very useful to me at the moment. By this, I mean applications that I use. BeOS is lacking horribly in high quality applications of most sorts. This can be remedied, and in fact, I am learning the Be API in hopes of aiding this process, but until the time when I can get something done in BeOS, I'm booting into Linux on my desktop.
I am a user of BeOS and I think that it is a brilliant piece of software. I bought it as soon as it was available for x86 hardware, but encountered problems immediately. It didn't like my video card or nic. I wasn't too distraught, though, and bought hardware that made Be happy. I now have a fully functional BeOS system that is astounding in it's responsiveness and general speed, but I never use it. There are so few applications for it that I can't use it. I have things that I need to do and I can't do them in BeOS. I stick with Linux most of the time, as it has nearly everything I need. Plus, Linux never needed me to buy hardware to make the system work. Best of luck to Be, though, I'd love to be able to make it my desktop OS.
Slashdot Mirror
For all the information you could ever want to know about how the new 300D/Digital Rebel compares to the other DSLRs that are out there just go check out photo.net. There is a full review of the body, plus lots of discussion about it in the forums.
Your second question, about whether or not to switch to digital, is not a question that we can answer for you, especially with the amount of information that you gave us. Both film and digital have their respective advantages. Both will continue to exist for quite some time. For a well thought out examination of film and digital photography, see Ken Rockwell's article on the subject.
What most people don't realize is that digital and film have been working together quite well for some time now and that the digital revolution has already made a huge impact in the printing phase. Lightjet and Chromira machines enable the highest quality prints and Fuji Frontier machines create good quality prints quickly. The quality of these prints is not just the resolution, but the color reproduction and tonality as well.
What it all comes down to, though, is not the equipment. You have to be in the right place when the light happens. Mastering light is far more important than having a certain kind of camera.
--joshMy experience with XDarwin has been fantastic. Apps under XDarwin seem so much faster and more responsive than they ever did under LinuxPPC, YDL or Debian on my 266mhz Powerbook. The Gimp runs great under XDarwin as well. My advice would be to install Fink and then use that to do all of the rest of your Unix apps. The fact that I have apt on my OS X system is just that much cooler. Stick with OS X and you won't be disappointed. This is advice coming from someone who faithfully used LinuxPPC for 3 years and has now switched wholeheartedly to OS X.
--neutrino
With the recent release of MacOS X what are your reactions to it and what plans do you have to compete with a truly user-friendly desktop OS combined with the stability of a UNIX backend?
--neutrino
There was a great article over at A List Apart a while ago that dealt with the problem of online advertising. The problem is far deeper than just targeting the audience better. The real problem is that banner ads, by their very nature, are not enticing to readers. What the online publications need to do is to make advertising an integral part of the site. One big problem is that banners are not constant. If I see an ad for Acme corporation and later want to learn more, I can't simply reload the page to find out about Acme. Instead, I will get another ad alltogether. How many times have you had that problem with a magazine? Also, banners usually have little or no actual information on them. This can't be good. You need information to be integrated with the ad. What needs to be done is more effort needs to be put into online advertising. Instead of regarding it as an additon to a site that pays the bills, it should be made an inseperable part of that site.
--neutrino
I am a huge supporter of Mozilla. It is my regular browser. I do have one wish for the more recent releases, though: Continue releasing binaries for alternate architectures. For the releases before 0.6 (all the Mxx releases), they pu up binaries for PPC, alpha and SPARC. They also released binaries for OS/2, HPUX and other more fringe oses. These weren't released at the same time as the Linux x86 and Windows binaries, but they were released. I know that I can compile it on my own machine (LinuxPPC), but their build host sits idle now instead of building other binaries. Just my thoughts, though.
--neutrino
What used to be a simple layer stack in GIMP 1.x, which is combined using layer modes (Normal, Combine, Difference, ...) will, with the help of GEGL, become a rendering pipeline which can be thought of as a tree of layers which is viewed from its root. The nodes of the tree are operators with an arbitrary number of inputs and outputs. These inputs and outputs access rectangular regions of pixel-data, the edges of the tree. Each edge (comparable to the layers we have now) can hold its data internally as pixels, vectors, text or whatever and only needs to provide a well-defined interface so it can be plugged into the rendering pipeline. A similar approach will be used for the operators: Simple functions like color corrections or blur filters as well as affine transformations and more complex effects are possible.
So, the rendering pipeline becomes much harder to explain in simple terms. This new model, however, yields great flexibility and power. You can have an image that contains both pixel-based portions and vector scalable portions. These objects can then be modified at any time. So, unlike current GIMP, text could be added and then later changed. This is truly useful for speeding up many uses that people have for the GIMP, but it also makes for more new uses.
--neutrino
- It will use the GEGL, or Generic Graphics Library for pixel manipulatio
- The rendering pipepline will be tree-based instead of a simple stack
- The program itself will become much more modular in design -- Model View Controller
- Plugins and app will communicate via CORBA
- Generally more flexible -- possible integration with vector-based tools and video editing capabilities
This is, as you can see, all backend work. Obviously, some changes will have to be made to the interface, but the result will be a vastly more powerful system. If they can successfully implement all of these ideas, then the GIMP would be, in many ways, the most technologically advanced image manipulation program in the world.Also, you should see the code as it stands in the 1.x series. It is obvious that the program has evolved to its current state and was not designed to this point.
--neutrino
Theo,
I also am an avid rock climber and I was wondering what level you climb at and what you feelis your biggest climbing accomplishment. Do you do big wall or any mountaineering, or do you just do sport climbing and bouldering?
--neutrino
- Code to the lowest common denominator
- Make multiple versions of every page
- or make their sites accessible only to certain browsers
And yet, so manyAdditionally, the author of the article is incorrect when he states that NS 6 is less standards compliant than Opera. Test it for yourself. NS6/Moz is the most standards compliant browser out there.
--neutrino
The difference between the present and the days of the 3.0 browsers is that today the "standards" are innovative. CSS-2, DHTML, DOM and XML/XSL are technologies that empower developers to do far more than can be done today. No browser even supports a signifigant portion of CSS-2. CSS-2 and XML allow us to finally escape the shackles of presentation and content being intertwined. If you don't develope internet sites, this might seem trivial, but when you are down in the trenches trying to provide a site that looks good on a new and "innovative" browser, but at the same time works on a cell phone, you learn to appreciate the innovation that the standards have. Check out the amazing things that can be done with standards over at the W3C and then try and find a better innovation that is needed in some browser.
--neutrino
neutrino
I agree with most of what you say, that most new technologies make pages more and more limited as far as who can view them, but this is not the case with CSS. A properly coded page that relies on CSS is actually better in Lynx, for instance, than an equivalence HTML page. Why? Well because you can avoid unnecessary usage of tags like in your document that you would use in straight HTML. Instead, you can create a new subclass of other tags which will give you the desired appearance but which will be treated appropriately by browsers that cannot produce the desired style. This is only with CSS 1. When CSS 2 support comes along, look out. It allows you to get rid of the biggest joke in HTML layout techniques, tables. Instead of dealing with putting every piece of content into a table cell, you define the layout in an intelligent way with a style sheet, just like professional publishing programs do. Similarly, when a browser comes along that cannot present it in the desired manner, the content is still readable, even if the browser doesn't understand CSS. Check it out, it is the way we'll be publishing on the web here real soon.
Posted on a Powerbook G3 running LinuxPPC2000
There are a number of reasons why the route chosen was the MHz route. Firstly, consider Moore's Law. If the speed of processors doubles every 18 months, then is there a similar relation for # of Procs on a system? Let's say that back in '92 we started down the multiproc route instead of the MHz route. In order to be at our current speed, we would need 32 procs to keep the speed. Silicon ain't cheap, though. This shows us that the CPU cost alone is prohibitive.
Secondly, the support hardware is also more complex for multiproc setups. You're memory controller must make sure that there is cache coherency across all of the processors' caches. This makes the motherboard more costly. There are a number of other hardware issues that jack the price up.
Finally, and perhaps most importantly, there is the software aspect of multiple CPUs. Writing concurrent programs is far more difficult than their sequential relatives. You not only have to debug the program for proper behavior, which is difficult when there can be process interference, but you also have to debug speed issues that arise from the interaction of processes. When a sequential program is run on a faster CPU, if it is a CPU-bound problem, then it will result in faster execution. This is not necessarily the case when you add processors. Concurrent programming is a whole new bucket of monkeys.
WARNING U.S. GOVERNMENT COMPUTER If not authorized to access this system, disconnect now. YOU SHOULD HAVE NO EXPECTATION OF PRIVACY By continuing, you consent to your keystrokes and data content being monitored.
The rules differ between systems, but every one that I have used has some punishment for unapproved system usage. Since the article said that they did this in secret, I assume that their work falls into the unapproved category. Although, SARA's website seems to indicate a fairly open view of their system. Who knows. I certainly wouldn't risk doing it on US systems that I have access to.
neutrino
I'm not sure that the important (ie limiting) factor here is the bandwidth nor even the amount of memory. The system that they quote as using, a Cray C916, is a computer that is no longer offered by SGI. It is a vector computing solution which has been succeeded by the T90 series. These systems only come in configurations of up to 32 processors, which is remarkable, but not compared to the real high end solutions (T3E and Origin 2000). I imagine that the statement about 2 Gigs of RAM was because that was the minimum space that the problem could reside in, but the true use of the Cray was the fact that for Polynomial searches, the paralleled vector processors are ideal for the job. The real possibility lies in the fact that a T932 could be used to do the same work in approximately 1/4 to 1/8 the time. If a group had access to a single T932 to do the Polynomial portion and a T3E for the rest, there is the possibility that they could do a similar factorization in under a week. If the reward was large enough (bank transactions) the cost of the hardware is easily covered. Hmm, large keys seem terribly desirable, no?
neutrino
I really like the direction that SGI has taken lately(aside from the foolish name change). Their focus on open source has been great. Now, as a Cray T3e programmer, I have a request: Please update the MPI libraries to MPI2 compliance. For those of you who aren't familiar with parallel computing, MPI(Message Passing Interface) is the industry standad for parallel computing. It can be used on a single processor PC all the way up to a 2048 processor T3e with little porting overhead. The problem is that nearly every implementation is of the older 1.2 implementation. MPI2 allows for trmendous speedup. If both Cray and MPICH, a free implementation, go to MPI2 compliance, then Beowulf clusters could fit more easily into many research environments. As it currently stands, the code bases for Supercomputers and clusters remain seperat entities in many cases. MPI2 would solve this. Oh well. Just my obscure wish, though.
neutrino
I'm curious why you don't like Mach. I think that it's a curious design. Admittedly, you take a speed hit when using it, but are there other reasons you don't like it?
neutrino
The difference is much bigger than just sheer numbers. Lets give Dell the benefit of the doubt and say that they have twice as many browsers pointed at them. They serve up mostly static pages. The parts that are dynamic are not very dynamic. Now look at eBay. They have a site that is nearly 100% dynamic what with the very nature of their business. Add in the personalized things that users can set up and you have a system that is working several orders of magnitude more than a site with heavier traffic but static content. Now, I don't think that Dell has anywhere near the number of hits. eBay junkies (My parents are antique dealers) check the site almost constantly throughout the evening. Once you've seen a Dell laptop, you know what it comes with. eBay customers expect a constant change in the site and thus check every few minutes as auctions that they are interested in draw near a close. eBay is easily a more taxed site.
neutrino
Yeah, they could crack it real quick. They could open up the safe that has the message written on it and read it, but that wouldn't be much fun, kind of like using their dedicated hardware wouldn't be much fun.
neutrino
This list is accurate at what it is trying to measure. The list is for general-purpose supercomputers and so this eliminates a number of machines that are much faster for dedicated tasks. The NSA, for instance, has machines that are very good at the task of factoring. These machines, however, are not capable of performing general tasks, and thus are not candidates for this list. In addition, you may say that there are other general purpose supercomputers out there that are faster, but not on the list. This is doubtful, as the machines have to be made by someone with experience at the rather specialized task of making ultra-fast systems. All of the major producers of supercomputers recognize the importance of such benchmarks and thus work with people to do Linpack tests. The importance of these benchmarks is that the company that produces high results will get more sales. The only computers now left out of this list would be those produced in secret by governments or other such entities. These groups fall behind because they do not have the existing knowledge and resource base needed to produce such computers. The end result: This list is quite accurate with perhaps a few Ultra Secret machines left out.
BTW, I work on two of the machines on this list.
neutrino
Yeah, I have an account on the 13th fastest computer in the world. Nothing quite as fun as sitting down and allocating 512 Processors to work on the formation of a planetary system. The Unicos/MK OS on Cray's is a very interesting example of microkernel design that is well done. Of course, the optimizing compilers on that baby are even more astounding.
neutrino
My main problem with BeOS is not the hardware. I needed to upgrade some components anyhow, but my problem, which I realize needs time to be solved, is that the BeOS is not very useful to me at the moment. By this, I mean applications that I use. BeOS is lacking horribly in high quality applications of most sorts. This can be remedied, and in fact, I am learning the Be API in hopes of aiding this process, but until the time when I can get something done in BeOS, I'm booting into Linux on my desktop.
Joshua Pearson
I am a user of BeOS and I think that it is a brilliant piece of software. I bought it as soon as it was available for x86 hardware, but encountered problems immediately. It didn't like my video card or nic. I wasn't too distraught, though, and bought hardware that made Be happy. I now have a fully functional BeOS system that is astounding in it's responsiveness and general speed, but I never use it. There are so few applications for it that I can't use it. I have things that I need to do and I can't do them in BeOS. I stick with Linux most of the time, as it has nearly everything I need. Plus, Linux never needed me to buy hardware to make the system work. Best of luck to Be, though, I'd love to be able to make it my desktop OS.
Joshua Pearson