You can view it as a setback for Pickens and his billion dollar water/land-grab in Texas, but it's not really a set-back for energy.
The turbines he purchased put money in the coffers of the companies building the turbines, which means more turbines and commercial viability for those companies. Pickens isn't likely to take the full $2bln hit from these things, so he'll find somewhere to put them, even if it wasn't along the Pickens Pipeline that he hoped for, so that he can regain some of the lost money in the form of energy ROI. All-in-all, the industry should see it as a boon.
The funny part is, if he had planned it better, he might have gotten away without having to invest in the alternative energy aspect at all, but idiotically he bought the turbines before having it entirely planned out. Sucks to be him (okay, not really, he's still rolling in billions), but it's a real win for us.
You're thinking of Sunken D from Starcraft. There are probably even earlier tower defense games, but this was the really, really big one (it was the Killer Game for an already Killer Game; I've bought several copies of SC over the years just to play this game, those damned keys are hard to keep track of).
This was my first thought too: just find a building that's relatively secure (any will do, just needs to be tall for line-of-sight purposes), put a (very?) high-gain directional antenna on it and link it to a base station in Afghanistan or Saudi Arabia or Kuwait. You could do the satellite way, but it's likely to be a hell of a lot more expensive, and you can't break it down and hide it in the case of an emergency evacuation, etc.
Might not be broadband, but it's the best you're going to do on such a short notice. Besides, Twitter messages are only 140 bytes. It's not like you have to send them over the cellular network where that would cost approximately a trillion dollars...
This is pure conjecture, but to me it seemed as if when AMD and ATI became one team and Nvidia and Intel became the other, that it would make sense for each one to offer incentives (read: threats) so that their partner would not bend over for the competition. So its not like its completely up to Nvidia to start improving their standing with AMD because of pressure from Intel. If that made any sense, then I'll drink a couple more beers before posting next time. Out
nVidia has made it quite clear on many occasions that they are not team players. They don't care about anyone else except themselves, will constantly put fault on anyone else that it can, and only acts in interest of protecting its own 'precious' IP. Not that Intel is any better on that last part, though.
nVidia and Intel aren't a team. They're competitors. DAAMIT is only starting to piss of nVidia more because they can actually push a whole platform (CPU + Chipset + Graphics), something Intel's had since its birth, and something nVidia has desperately wanted but never been able to pull off (with attempt after failed attempt using ARM chips, Intel chips, AMD's CPUs, chipsets, etc).
If anything, Intel and AMD have all of the incentive in the world to simply stop caring about nVidia and move to push them out of the market entirely (see Larrabee and AMD's similar Compute-GPU projects). It's going to get harder for nVidia to push graphics chips if everyone in the market they're competing with has high-performance graphics on-die with their CPUs.
I'm thinking Cortex A8's, which have been out for over a year. Stuff like the OMAP 3530(present in the Beagleboard, upcoming Pandora Handheld, and Palm Pre) consumes remarkably small amounts of power.
The Pandora developers said their device consumes around or just over 1 watt. Most of that is from the LCD. They did experiments completely shutting off certain hardware, to measure power consumption, and concluded...
CPU - about 20-40mw
DSP - about 30-60mw
SGX GPU - about 30-60mw
(Hard to get exact measurements due to the nature of how components interact. Anything loading the CPU probably loads up the memory as well. Anything hitting the GPU will hit the CPU, and DSP load varies greatly depending on the codec and video being decoded.)
The entire SoC uses a ludicrously small amount of power; something like 0.2-0.4w. Then add another 0.6w for the LCD, and a bunch more for wireless.
Now, compare that to the current Atoms, with 6+ watts just for the CPU/chipset, another 2+ for the HDD/SSD, at least 6-15w for the LCD, etc...
If any company can drive down their power consumption, Intel can, but that doesn't mean it'll be easy to catch ARM!
I just can't wait for Cortex A9's. Quad-core ARM in the exact same power envelope!
To be fair, the Atom runs at 6 Watts max, where average TDP can down to as little as 0.4W.
The problem with Atom, as you say, is all of the other hardware to make it work. Its current chipset is incredibly power hungry, but they're working on that (integrating more and doing even deeper clock gating). Future Atoms will likely use even less power, with Intel already shipping chips with a max 2.4W threshold.
And yes, you are being unfair comparing a device which has a hard drive with hundreds of gigabytes of space and a WXSVGA screen to a handheld device with a couple of gigs of flash memory and a HVGA screen. Nobody's stopping you from making an Atom device with those components (though it will take more power right now, it'll be vastly faster than the Cortex A8, and you won't have to recompile or use highly specialized toolkits, which is a huge Intel advantage).
Small projects in the stage of trying things is roughly the definition of drafting a design.
And quite frankly, if you don't draft the design correctly from the beginning, it can cause huge problems later. Some projects can skip this step and do iterative design (most open source projects, software libraries, e.g.), but commercial products typically don't have the luxury of having a product delayed by a huge amount because someone's design was wrong and it is now a burden to remove and replace.
Not stopping to design for the future early on is the reason Xorg is a complete and absolute clusterfuck today, for example.
Code reviews shouldn't be a "full stop, let's go over this" event. Code review should be a part of the every day workings of the development team. Nothing should go into version control's master/trunk/HEAD until it has been reviewed.
Sometimes you'll find that stopping to review a single module is helpful, but most of the time it's actively harmful to the team, since it takes developer's concentration off of what they're currently working on to review things that they half-don't-remember, which then makes the code review take forever.
Review and document inline with your coding, and you'll find you'll never need a "Full Stop" review.
I don't think so. Specifically:
"It will be months before anyone has general utility of the API."
And I agree with that. Science and Engineering require that applications are correct, that APIs have been tested and retested, formally verified where possible, etc. As much as it'd be great to see these APIs picked up quickly by Mathematica and the lot, it will probably be at least a year before we see it.
The content production market can move faster because they don't require the same levels of precision and accuracy, and they generally have more coders actively working at any given time.
Shortlist:
AGG is a software renderer, and is a bit faster than Cairo-Pixman, though is less accurate than Cairo (Cairo is designed for correctness over performance, where possible, but does have an extensive regression and performance testing framework).
Cairo has multiple backends: Pixman which is software, X11 (through XCB or XLib), Windows, Mac OS X, BeOS, PDF/PS/SVG, DirectFB, SDL, OS/2, several OpenGL backends (Glitz though it's unmaintained, and at least two 'plain' OpenGL backends from different hackers), an OpenVG backend and an in-progress DRI2 backend (which directly uses the graphics hardware in Linux). In addition, it has very rich font support, supporting the default font system on the platforms above, and includes support for user-defined fonts. AGG's font support is rather mediocre, supporting only Freetype and Win32. Neither library does much extensive layout code, though Pango uses Cairo to produce (arguably) the best text layout experience in the (FOSS) world.
AGG has a richer set of vector-related features, such as Gaussian filtering (which has been a todo item for Cairo for quite some time), Type-6/7 meshes with Gouraud shading and image warping, and provides Gamma correction. Some work has been done on adding Gamma and color correction to Cairo, but I don't believe there's anything deliverable at this point.
Cairo has a richer set of bindings; being written in C, it has bindings to practically every commonly used programming language, and a few uncommonly/rarely used ones, such as OCaml and Keith Packard's Nickel language.
All-in-all, it's about what you're going to do with the two libraries. AGG is written more towards a graphics package, like Inkscape or the GIMP. Cairo is written more to be the backend of a display system, like Quartz2D.
PostScript, PDF are still the winners in terms of adoption and efficiency. PostScript is more burdensome code-wise, so most everyone today will recommend PDF (it's essentially "PS-lite") or even a subset of PDF (most commonly throwing out things like quadradic Bezier curves and higher-level filtering).
Most modern Windowing systems are heavily influenced by these as well; Cairo is essentially DisplayPDF (using slightly different compositing operators; closer to the original Porter-Duff ones), Quartz even more so than Cairo (implements filtering operators, etc). Even Windows has gotten off the bitmap and uses a similar model for their display system in XP/Vista with GDI+.
You didn't have resolution-independent graphics, either.
It's all fine and dandy to write code that runs in a terminal and lets you play solitare in codepage 437 graphics, but this is 2009 and people expect more from their thousand dollar computers, and application developers like to have their software scale seamlessly to devices like phones, PDAs, Netbooks, etc.
QT/Webkit once you've got all of the dependencies down is over 20 megs. Cairo is an order of magnitude smaller.
The fact is, vector art isn't cheap. Not computationally, not memory wise. Making a real go at it (not just faking a bunch of it, as many libraries do for speed) takes time, it takes work, it takes a bunch of libraries (XML parser, CSS parser, layout engine, etc). Putting any size constraint on it should at least take these things into account, which this developer quite clearly has not.
I'd suggest he do his homework and look into the libraries that currently exist.
OTOH, you could simply use librsvg to render the SVG graphics to PNG (or some other supported format) and display with the appropriate library calls.
You bash Cairo for being "too immature" (wtf, really? It's one of the most advanced 2D vector libraries out there), and then recommend a rendering library that is written exclusively to render to Cairo. Well played sir.
According to one website, which is visited by some fraction of the number of actual Linux users. By that same method, 20% of the visitors to Slashdot probably use Linux, does that mean Linux has a 20% marketshare? How about Microsoft.com and its probable near 0% share?
Conclusion: There is just no way to accurately measure with any kind of precision the number of active Linux machines and Linux users.
Nope, not even close. Apple *designs* or works with manufacturers to create custom *designed* boards and hardware but they build nothing. They are the same chips and chipsets as Dell, which actually does the same thing and custom *designs* their gear just like Apple.
So in the same way, all cars and buildings are exactly alike because they all use exactly the same materials, the design is just a bit different. Right?
Yeah, the CPU and chipset are the same. They even use copper and solder to make their boards, oh my! But there's are millions of ways to lay out the same components, and they do yield different results. Take cities: they all have similar needs (hospitals, police stations.. just play Sim City), but how you lay them out can be the entire difference between a traffic-congested, slow ass stinky, cheap and broken down city, and a beautiful, traffic-free, low crime city.
Of course, we're geeks here. We don't care, as long as it's got the bigger numbers next to its MHz and GBs, we're happy. Meanwhile, there are people who like having computers that last more than 6 months to a year, computers that don't crash whenever someone turns on the microwave down the hall or accidentally turns off the air conditioner, or who's components are literally melting off of the machine (as is the acse with my 2 year old Toshitba laptop).
There's really not as much to it as you think. They're just layers of etched copper in a fiberglass substrate. Intel makes a lot of reference board designs with the complete layout of the traces and layers in the copper and fiberglass, and Dell is notorious for just taking those and putting their names on it. Formerly they didn't even bother to go that far, but now they actually do change the silkscreens (and sometimes the shape a bit to make them fit in their case, which really isn't hard for even a first grader to do as long as they're not moving RAM or CPU traces). Dell (and Asus, Gigabyte.. pretty much anyone building generic motherboards for the ubercheap computer range) does this because it's incredibly cheap; Intel's licensing allows Dell access to the reference designs basically for free, which means they don't have to pay for prototypes and don't have to pay engineers hundreds of thousands a year to lay them out.
Apple has, more or less since the beginning of their company, designed their own boards from scratch, taking the chips they want, then laying out the boards themselves and sending those designs off to be built. This alone doesn't mean much; anyone who's been through an electronics course in college has probably done a board or two. However, building modern boards actually is a really damned hard (thousands of pins, some traces need to have even length, enough substrate between them to eliminate crosstalk, power/noise concerns, etc). The fact that Apple cares to do this, and makes boards that are very high quality is a testament to their dedication to build the machine from the bare metal up.
Companies like Foxconn and ASUS build Apple's hardware to Apple's specifications, as they do for Dell and just like they used to do for Packard Bell.
One more time and we'll get it right:
Companies like Foxconn and ASUS build Apple's hardware to Apple's specifications. Companies like Foxconn and ASUS also build hardware for Dell and Packard Bell, most of which was (reference) designed by engineers at Intel. Hell, for the longest time, Dell's boards had "Intel" silkscreened on them, since they were literally carbon copies of Intel's reference board and nobdoy bothered to change it.
Honestly, I have recommended it to everyone I know. It's quite honestly the best mouse I've ever used.
It's only flaw is that it doesn't have Bluetooth, but at the same time its battery life is about 4 times as long as my desktop's (also a Logitech) Bluetooth mouse.
If I start firefox up on a machine that has 64MB of RAM
I'm sorry, I can't even take you seriously past here. Most modern web pages require a great deal of RAM even to be considered renderable in reasonable time; as most of the page's members are not dynamic, they can be freed (or as firefox does, thrown into a freelist) after rendering. But the dynamic members have to be held around. This wasn't the case when DOM didn't exist and you could essentially just render the page to an image and be done with it.
Webpages these days require lots and lots of data. 64MB is *barely* enough to browse the Javascript-Web-2.0 web. This is why even phones have more RAM than this today, and their screens are not even a quarter of the size of the displays on modern systems.
The fact is, the data is forcing you to upgrade when you're that low on RAM. God forbid you try to open the 4 megapixel picture your kid sent you of a cat and realize it takes a quarter of the RAM in your system. Consider spending the $25 bucks to upgrade your system to 2GB of RAM. Or cry about the fact your 10+ year old machine can't watch YouTube videos, and strains flat on Slashdot's new layout...
And that is Ubuntu's fault? More likely that's ATi's fault for absolutely failing to write a sane driver, dropping support for almost all of the video cards not made in the past year, etc.
ATi has a very, very long history of sucking ass in the software department. That isn't going to change because you're using Ubuntu.
A quick look down my process list (Fedora 11) shows top bulky processes are:
* FireFox with a resident size of 184MB
* Xorg with a resident size of 125MB
* Lots of Gnome bits and pieces totalling maybe 100MB
* Nautilus with a resident size of 33MB
Two schools of thought: If you have memory, use it, else it's not doing anything, and don't ever use memory, just feel really damned sluggish requerying the same shit every single time you need data.
Today, most processes are I/O bound. The fact of the matter is, for all of the memory, CPU and bus speed we have, we're still trying to squeeze too much data through pipes that are too thin to hold it. File managers and web browsers are especially culprits here, as file management requires going to disk, which is the just about the slowest thing you can do except for going to the internet to fetch something.
We may call our Fiber and Cable lines "high-speed", but we're still talking in MB/s. Your CPUs are talking GB/s. Most of the time, your high-powered CPU is sitting around doing nothing at all, simply because there is no work that it can do; it's waiting on data, on some I/O port to interrupt and say "Hey, I'm finally done loading that PNG."
So, are we going to complain about having to go out to RAM and using a lot of it (which is nano-/microseconds away), or are we going to complain about going out to disk/the internet and using very little RAM and taking (milli)seconds?
Slashdot Mirror
You can view it as a setback for Pickens and his billion dollar water/land-grab in Texas, but it's not really a set-back for energy.
The turbines he purchased put money in the coffers of the companies building the turbines, which means more turbines and commercial viability for those companies. Pickens isn't likely to take the full $2bln hit from these things, so he'll find somewhere to put them, even if it wasn't along the Pickens Pipeline that he hoped for, so that he can regain some of the lost money in the form of energy ROI. All-in-all, the industry should see it as a boon.
The funny part is, if he had planned it better, he might have gotten away without having to invest in the alternative energy aspect at all, but idiotically he bought the turbines before having it entirely planned out. Sucks to be him (okay, not really, he's still rolling in billions), but it's a real win for us.
You're thinking of Sunken D from Starcraft. There are probably even earlier tower defense games, but this was the really, really big one (it was the Killer Game for an already Killer Game; I've bought several copies of SC over the years just to play this game, those damned keys are hard to keep track of).
This was my first thought too: just find a building that's relatively secure (any will do, just needs to be tall for line-of-sight purposes), put a (very?) high-gain directional antenna on it and link it to a base station in Afghanistan or Saudi Arabia or Kuwait. You could do the satellite way, but it's likely to be a hell of a lot more expensive, and you can't break it down and hide it in the case of an emergency evacuation, etc.
Might not be broadband, but it's the best you're going to do on such a short notice. Besides, Twitter messages are only 140 bytes. It's not like you have to send them over the cellular network where that would cost approximately a trillion dollars...
This is pure conjecture, but to me it seemed as if when AMD and ATI became one team and Nvidia and Intel became the other, that it would make sense for each one to offer incentives (read: threats) so that their partner would not bend over for the competition. So its not like its completely up to Nvidia to start improving their standing with AMD because of pressure from Intel. If that made any sense, then I'll drink a couple more beers before posting next time. Out
nVidia has made it quite clear on many occasions that they are not team players. They don't care about anyone else except themselves, will constantly put fault on anyone else that it can, and only acts in interest of protecting its own 'precious' IP. Not that Intel is any better on that last part, though.
nVidia and Intel aren't a team. They're competitors. DAAMIT is only starting to piss of nVidia more because they can actually push a whole platform (CPU + Chipset + Graphics), something Intel's had since its birth, and something nVidia has desperately wanted but never been able to pull off (with attempt after failed attempt using ARM chips, Intel chips, AMD's CPUs, chipsets, etc).
If anything, Intel and AMD have all of the incentive in the world to simply stop caring about nVidia and move to push them out of the market entirely (see Larrabee and AMD's similar Compute-GPU projects). It's going to get harder for nVidia to push graphics chips if everyone in the market they're competing with has high-performance graphics on-die with their CPUs.
good architecture
Don't you mean ludicrously good architecture?
I'm thinking Cortex A8's, which have been out for over a year. Stuff like the OMAP 3530(present in the Beagleboard, upcoming Pandora Handheld, and Palm Pre) consumes remarkably small amounts of power.
The Pandora developers said their device consumes around or just over 1 watt. Most of that is from the LCD. They did experiments completely shutting off certain hardware, to measure power consumption, and concluded...
CPU - about 20-40mw DSP - about 30-60mw SGX GPU - about 30-60mw
(Hard to get exact measurements due to the nature of how components interact. Anything loading the CPU probably loads up the memory as well. Anything hitting the GPU will hit the CPU, and DSP load varies greatly depending on the codec and video being decoded.)
The entire SoC uses a ludicrously small amount of power; something like 0.2-0.4w. Then add another 0.6w for the LCD, and a bunch more for wireless.
Now, compare that to the current Atoms, with 6+ watts just for the CPU/chipset, another 2+ for the HDD/SSD, at least 6-15w for the LCD, etc...
If any company can drive down their power consumption, Intel can, but that doesn't mean it'll be easy to catch ARM!
I just can't wait for Cortex A9's. Quad-core ARM in the exact same power envelope!
To be fair, the Atom runs at 6 Watts max, where average TDP can down to as little as 0.4W. The problem with Atom, as you say, is all of the other hardware to make it work. Its current chipset is incredibly power hungry, but they're working on that (integrating more and doing even deeper clock gating). Future Atoms will likely use even less power, with Intel already shipping chips with a max 2.4W threshold.
And yes, you are being unfair comparing a device which has a hard drive with hundreds of gigabytes of space and a WXSVGA screen to a handheld device with a couple of gigs of flash memory and a HVGA screen. Nobody's stopping you from making an Atom device with those components (though it will take more power right now, it'll be vastly faster than the Cortex A8, and you won't have to recompile or use highly specialized toolkits, which is a huge Intel advantage).
Small projects in the stage of trying things is roughly the definition of drafting a design.
And quite frankly, if you don't draft the design correctly from the beginning, it can cause huge problems later. Some projects can skip this step and do iterative design (most open source projects, software libraries, e.g.), but commercial products typically don't have the luxury of having a product delayed by a huge amount because someone's design was wrong and it is now a burden to remove and replace.
Not stopping to design for the future early on is the reason Xorg is a complete and absolute clusterfuck today, for example.
Nobody was talking to you, Skynet.
Ahh the quick typist tripped over the lazy dog. Missed that one. Sorry.
then you're doing it wrong. Plain and simple.
Code reviews shouldn't be a "full stop, let's go over this" event. Code review should be a part of the every day workings of the development team. Nothing should go into version control's master/trunk/HEAD until it has been reviewed.
Sometimes you'll find that stopping to review a single module is helpful, but most of the time it's actively harmful to the team, since it takes developer's concentration off of what they're currently working on to review things that they half-don't-remember, which then makes the code review take forever.
Review and document inline with your coding, and you'll find you'll never need a "Full Stop" review.
I don't think so. Specifically: "It will be months before anyone has general utility of the API."
And I agree with that. Science and Engineering require that applications are correct, that APIs have been tested and retested, formally verified where possible, etc. As much as it'd be great to see these APIs picked up quickly by Mathematica and the lot, it will probably be at least a year before we see it.
The content production market can move faster because they don't require the same levels of precision and accuracy, and they generally have more coders actively working at any given time.
Shortlist: AGG is a software renderer, and is a bit faster than Cairo-Pixman, though is less accurate than Cairo (Cairo is designed for correctness over performance, where possible, but does have an extensive regression and performance testing framework).
Cairo has multiple backends: Pixman which is software, X11 (through XCB or XLib), Windows, Mac OS X, BeOS, PDF/PS/SVG, DirectFB, SDL, OS/2, several OpenGL backends (Glitz though it's unmaintained, and at least two 'plain' OpenGL backends from different hackers), an OpenVG backend and an in-progress DRI2 backend (which directly uses the graphics hardware in Linux). In addition, it has very rich font support, supporting the default font system on the platforms above, and includes support for user-defined fonts. AGG's font support is rather mediocre, supporting only Freetype and Win32. Neither library does much extensive layout code, though Pango uses Cairo to produce (arguably) the best text layout experience in the (FOSS) world.
AGG has a richer set of vector-related features, such as Gaussian filtering (which has been a todo item for Cairo for quite some time), Type-6/7 meshes with Gouraud shading and image warping, and provides Gamma correction. Some work has been done on adding Gamma and color correction to Cairo, but I don't believe there's anything deliverable at this point.
Cairo has a richer set of bindings; being written in C, it has bindings to practically every commonly used programming language, and a few uncommonly/rarely used ones, such as OCaml and Keith Packard's Nickel language.
All-in-all, it's about what you're going to do with the two libraries. AGG is written more towards a graphics package, like Inkscape or the GIMP. Cairo is written more to be the backend of a display system, like Quartz2D.
PostScript, PDF are still the winners in terms of adoption and efficiency. PostScript is more burdensome code-wise, so most everyone today will recommend PDF (it's essentially "PS-lite") or even a subset of PDF (most commonly throwing out things like quadradic Bezier curves and higher-level filtering).
Most modern Windowing systems are heavily influenced by these as well; Cairo is essentially DisplayPDF (using slightly different compositing operators; closer to the original Porter-Duff ones), Quartz even more so than Cairo (implements filtering operators, etc). Even Windows has gotten off the bitmap and uses a similar model for their display system in XP/Vista with GDI+.
You didn't have resolution-independent graphics, either.
It's all fine and dandy to write code that runs in a terminal and lets you play solitare in codepage 437 graphics, but this is 2009 and people expect more from their thousand dollar computers, and application developers like to have their software scale seamlessly to devices like phones, PDAs, Netbooks, etc.
QT/Webkit once you've got all of the dependencies down is over 20 megs. Cairo is an order of magnitude smaller.
The fact is, vector art isn't cheap. Not computationally, not memory wise. Making a real go at it (not just faking a bunch of it, as many libraries do for speed) takes time, it takes work, it takes a bunch of libraries (XML parser, CSS parser, layout engine, etc). Putting any size constraint on it should at least take these things into account, which this developer quite clearly has not.
I'd suggest he do his homework and look into the libraries that currently exist.
OTOH, you could simply use librsvg to render the SVG graphics to PNG (or some other supported format) and display with the appropriate library calls.
You bash Cairo for being "too immature" (wtf, really? It's one of the most advanced 2D vector libraries out there), and then recommend a rendering library that is written exclusively to render to Cairo. Well played sir.
Seems pretty apt.
Re:Hmm, wait, it's 1.02%
.
According to one website, which is visited by some fraction of the number of actual Linux users. By that same method, 20% of the visitors to Slashdot probably use Linux, does that mean Linux has a 20% marketshare? How about Microsoft.com and its probable near 0% share?
Conclusion: There is just no way to accurately measure with any kind of precision the number of active Linux machines and Linux users.
I use it under Linux just fine, all buttons working, etc. There's no fancy configuration software, but that's more Xorg's fault than Logitech's.
Nope, not even close. Apple *designs* or works with manufacturers to create custom *designed* boards and hardware but they build nothing. They are the same chips and chipsets as Dell, which actually does the same thing and custom *designs* their gear just like Apple.
So in the same way, all cars and buildings are exactly alike because they all use exactly the same materials, the design is just a bit different. Right?
Yeah, the CPU and chipset are the same. They even use copper and solder to make their boards, oh my! But there's are millions of ways to lay out the same components, and they do yield different results. Take cities: they all have similar needs (hospitals, police stations.. just play Sim City), but how you lay them out can be the entire difference between a traffic-congested, slow ass stinky, cheap and broken down city, and a beautiful, traffic-free, low crime city.
Of course, we're geeks here. We don't care, as long as it's got the bigger numbers next to its MHz and GBs, we're happy. Meanwhile, there are people who like having computers that last more than 6 months to a year, computers that don't crash whenever someone turns on the microwave down the hall or accidentally turns off the air conditioner, or who's components are literally melting off of the machine (as is the acse with my 2 year old Toshitba laptop).
There's really not as much to it as you think. They're just layers of etched copper in a fiberglass substrate. Intel makes a lot of reference board designs with the complete layout of the traces and layers in the copper and fiberglass, and Dell is notorious for just taking those and putting their names on it. Formerly they didn't even bother to go that far, but now they actually do change the silkscreens (and sometimes the shape a bit to make them fit in their case, which really isn't hard for even a first grader to do as long as they're not moving RAM or CPU traces). Dell (and Asus, Gigabyte.. pretty much anyone building generic motherboards for the ubercheap computer range) does this because it's incredibly cheap; Intel's licensing allows Dell access to the reference designs basically for free, which means they don't have to pay for prototypes and don't have to pay engineers hundreds of thousands a year to lay them out.
Apple has, more or less since the beginning of their company, designed their own boards from scratch, taking the chips they want, then laying out the boards themselves and sending those designs off to be built. This alone doesn't mean much; anyone who's been through an electronics course in college has probably done a board or two. However, building modern boards actually is a really damned hard (thousands of pins, some traces need to have even length, enough substrate between them to eliminate crosstalk, power/noise concerns, etc). The fact that Apple cares to do this, and makes boards that are very high quality is a testament to their dedication to build the machine from the bare metal up.
Let me fix that for you:
Companies like Foxconn and ASUS build Apple's hardware to Apple's specifications, as they do for Dell and just like they used to do for Packard Bell.
One more time and we'll get it right: Companies like Foxconn and ASUS build Apple's hardware to Apple's specifications. Companies like Foxconn and ASUS also build hardware for Dell and Packard Bell, most of which was (reference) designed by engineers at Intel. Hell, for the longest time, Dell's boards had "Intel" silkscreened on them, since they were literally carbon copies of Intel's reference board and nobdoy bothered to change it.
Honestly, I have recommended it to everyone I know. It's quite honestly the best mouse I've ever used.
It's only flaw is that it doesn't have Bluetooth, but at the same time its battery life is about 4 times as long as my desktop's (also a Logitech) Bluetooth mouse.
If I start firefox up on a machine that has 64MB of RAM
I'm sorry, I can't even take you seriously past here. Most modern web pages require a great deal of RAM even to be considered renderable in reasonable time; as most of the page's members are not dynamic, they can be freed (or as firefox does, thrown into a freelist) after rendering. But the dynamic members have to be held around. This wasn't the case when DOM didn't exist and you could essentially just render the page to an image and be done with it.
Webpages these days require lots and lots of data. 64MB is *barely* enough to browse the Javascript-Web-2.0 web. This is why even phones have more RAM than this today, and their screens are not even a quarter of the size of the displays on modern systems.
The fact is, the data is forcing you to upgrade when you're that low on RAM. God forbid you try to open the 4 megapixel picture your kid sent you of a cat and realize it takes a quarter of the RAM in your system. Consider spending the $25 bucks to upgrade your system to 2GB of RAM. Or cry about the fact your 10+ year old machine can't watch YouTube videos, and strains flat on Slashdot's new layout...
And that is Ubuntu's fault? More likely that's ATi's fault for absolutely failing to write a sane driver, dropping support for almost all of the video cards not made in the past year, etc.
ATi has a very, very long history of sucking ass in the software department. That isn't going to change because you're using Ubuntu.
A quick look down my process list (Fedora 11) shows top bulky processes are: * FireFox with a resident size of 184MB * Xorg with a resident size of 125MB * Lots of Gnome bits and pieces totalling maybe 100MB * Nautilus with a resident size of 33MB
Two schools of thought: If you have memory, use it, else it's not doing anything, and don't ever use memory, just feel really damned sluggish requerying the same shit every single time you need data.
Today, most processes are I/O bound. The fact of the matter is, for all of the memory, CPU and bus speed we have, we're still trying to squeeze too much data through pipes that are too thin to hold it. File managers and web browsers are especially culprits here, as file management requires going to disk, which is the just about the slowest thing you can do except for going to the internet to fetch something.
We may call our Fiber and Cable lines "high-speed", but we're still talking in MB/s. Your CPUs are talking GB/s. Most of the time, your high-powered CPU is sitting around doing nothing at all , simply because there is no work that it can do; it's waiting on data, on some I/O port to interrupt and say "Hey, I'm finally done loading that PNG."
So, are we going to complain about having to go out to RAM and using a lot of it (which is nano-/microseconds away), or are we going to complain about going out to disk/the internet and using very little RAM and taking (milli)seconds?