We had a 61% turnout last year, but since the FPTP system is so retarded, our Canadian version of Dubya won a (narrow) majority despite getting less than 40% of the votes.
The way most of us read the results, it means over 60% did NOT want that guy to win. Either way, since getting his majority he's been ramming all these big brother bills down our throat, along with unprecedented military spending and all the other abusive stuff you neighbours have been subjected to for the last few terms. Shit's going downhill fast and riots are become more and more frequent. Amazing how easiy one sellout can ruin a country for millions of people.
Don't tell anyone, but yea, I did read the damned thing, to the very end. It's like being at a bar, and some skinny blue-collar trainwreck starts chatting your ear off. Your fist wants to shut him up, but your brain wants to continue laughing at this guy's self-righteous hyperbole. Basically applying the Dunning-Kruger effect in entertainment form.
London $ACT_OF_CANNIBALISM its own $IMPORTANT_UTILITY to $WEASEL_WORD it got $FASCIST_INTEREST
You can apply the same template everywhere. One day they're selling out to the I.O.C., the next they're hobbling their own civil freedoms to benefit the U.S. military-industrial complex. Seeing how they were a historic superpower, watching the government destroy its own economy and culture is, well frankly, fucking pathetic.
"Hey, our traffic sucks so bad, a bunch of greedy attention whores might not want to come spend our money. Quick, let's drop everything and suck their snooty cocks, rather than fixing the underlying urban problems."
1. Be an ignorant hipster microserf excitable attention whore 2. Write an ignorant article that makes you and your equally unenlightened followers giddy 3. Submit to slashdot and hope it's one of those new moronic editors who reviews it 4. Traffic 5. ??? (hint: cinnamon-chai lattés until your head implodes) 6. PROFIT!
This site's getting so bad, it's making Gizmodo look good.
The fundamental problem is that, for a lot of these systems, newer does not mean better.
That DOS record keeping system, is it good enough ? Would a costly rewrite and/or data migration result in useful improvements to justify the expense ?
For many businesses, it only becomes a problem when the old hardware breaks down, and the new hardware can't run the obsolete software. In practice, people would sooner run their archaic system via DosBox than have it redeveloped.
Today's jukeboxes work on a subscription model. You pay $X per month, you get a constantly rotating selection of popular music, old and new. I wouldn't be surprised if these things had their own 3G uplink, because the places I've seen them installed didn't have any WFi. My educated guess is they're installed by bar owners who can't be bothered to set up a conventional sound system. I strongly doubt they'd put in the effort to set up a network for smartphones to vote on tunes, nor would they pay for a contractor to do it.
Given that I've spent the majority of my life working with computers, I've come to accept reality as just another theory. Does the OS know it's inside a virtual machine ? (without the hypervisor intentionally making itself known) How can any person know, with absolute certainty, that they're not a brain in a jar, being fed simulated input ? How can we even know we're a brain at all ? For all I know, my entire existence could be a work of fiction, the Internet could be a fabrication of my mind, along with all its inhabitants.
The only thing we can reasonably assume, is that thought exists.
(and yes, I think the best psych/philosophy profs were the ones who dropped acid on a regular basis:)
Err... yes and no. Software has gotten absolutely awful in absolute terms, but since hardware has been enjoying tremendous speed increases, we don't notice it so much.
Just starting a PHP interpreter takes about a half-second on my bleeding-edge PC. To run the same thing on a 90's era machine would probably take 5 to 10 seconds. Writing that same code in C and compiling to the platform's native bytecode would result in near-instant execution on the order of a few hundredths of a second. Software today is bloated to hell, but it is much easiler to develop so what we lose in machine time, we gain in human time.
I don't know what kind of obsolete hardware you're using, but KDE's "noticeable lag" is in the quarter-second range, and Win8's "unusable" speed is faster than Win7 for me.
Maybe you should donate that K6-233 to the nearest dumpster.
Yes and no. I find that if you slipstream those updates into an install disc, the resulting machine is snappy. It's when you install patches into an existing system than everything goes to shit, because it keeps old versions of those files around "for compatibility". In other words, if you're running XP SP3, you have a bunch of Gold, SP1 and SP2 duplicate DLLs kicking around, occasionally being loaded in parallel to satisfy some applications that predate those releases. The result is increased disk and memory usage. Meanwhile, a fresh install of XP SP3 runs all the same apps just fine - go figure!
With 7, it stays snappy. My install has been going for nearly 3 years now, still perfectly fast and I feel no pressure to wipe it and start over.
I haven't touched KDE in about two years, in part because KDE 4 left a sour taste in my mouth, but I have to agree. Window managers on Linux have little or no impact on application performance.
That said, I do feel that several aspects of Linux operation are slower than they could be. Anything that ties into a handful of shared libraries tends to chug, as though the run-time linker might be spinning its wheels. I also find the kernel rather slow to load and boot. Not init, but the kernel itself. In fact, on my last few PCs, init finishes in about 2 seconds while the kernel needs 30-40 seconds between the bootloader and init starting. That seems a bit much, given the absurdly fast hardware at its disposal. X takes another 5 seconds to grok the graphics card and set up a root display.
Windows brings up the same hardware in about 8 seconds. It used to be the other way around, but even our beloved open-source OS has fallen prey to bloat and sloppy coding.
I have an old install of Windows 7. I installed it the day it was released, and have not reformatted since. It's full of crap and tweaked the way I like. By far, the longest wait for me is the BIOS, which takes its sweet time to POST. I'd guesstimate about 90 seconds for that POST, and then 7-8 seconds for Windows to hit the desktop and be responsive.
The bulk of that POST time is spent waiting for disks and RAID controllers to initialize. If I had simpler hardware, it would probably take a quarter of the time.
Windows 7 in 4gb is perfectly usable. And by usable, I mean snappy. And by snappy, I mean I'm an impatient coder who starts punching things when I see an hourglass or beachball.
On 8gb, I bet it runs like a dream. I have a very large memory workstation because I do lots of VM work, but according to the task manager, my memory is mostly sitting idle in the disk cache. Actual working memory seems to hover around 3gb or so, with email, web and a game running - the type of general usage I'd expect from a non-programmer.
I've been running swap-free for years, never had any issues and software generally doesn't bother you - except Windows itself. It warns you if you disable the swap file, but I've never encountered any adverse effects. Photoshop has its own virtual memory system (scratch memory), it does not care about Windows' memory management, it will create its own temp files as needed and you can configure where these go. I point mine at an old SSD.
That's not how read caches work. Yes, it will preload a bunch of stuff it thinks you might need, but the moment an actuall application requests that memory, it is immediately freed. It doesn't need to "swap out" because the cache is just a copy of files that are already on disk, and readily available to be cached again later.
This is true of all(?) executable files and DLLs, as well as any file marked as read-only. Why write out data to the swapfile, if there's a perfectly good copy already on disk ?
Windows 7 will run with 0 swap. This is how I run it on my primary desktop, where I do everything: work, gaming, media processing. Mind you, that machine has 48gb of Ram, but even on my previous desktop with only 8gb, it ran perfectly fine. On my laptop, I run Win7 in a VM, with 3.5gb of RAM and no swap. Works fine, I use it for Eclipse Blackberry development all day long (yes, I hate my life). Win7 is far more memory and processor efficient than Vista, and boots faster than XP. If it didn't, I'd have used an XP VM instead. Eclipse certainly doesn't care what it runs on.
You should look into NLite. I've crammed XP into 128mb and 256mb virtual machines that were fully functional, running IE7/8 whatever was current a few years ago. Its idle memory usage, as reported by the task manager, was about 85mb. Turn off themes, and any background services you don't need, and XP can fit into relatively tight environments with ease.
I believe you fail to understand RAM's purpose: to store data. Your disk check probably did not require 11 GB, but since it was available, it chose to use it to improve its performance.
People who are new to Linux tend to bicker about the same thing, because the disk cache will consume whatever memory is available and report it as "used", even though it is the first thing to be released when any other software requires memory. In a sense, it is opportunistically used, but not reserved. Free RAM is wasted RAM.
Conversely, I have a 48gb system running Windows 7. It usually hovers around 8gb free when idling at the desktop. 2 gb goes to the OS and background tasks, the remaining 38gb is borrowed by the disk cache, because I tend to thrash very large video files which linger in memory after I'm done. I don't use a swap file, and I can run a bazillion apps at once without any noticeable slowdown. Yes, even a disk check, virus scan and a dozen virtual machines simultaneously.
Well see, that's the thing. A GPU is better suited to some kinds of massively parallel tasks, like video encoding. After all, you're applying various matrix transforms to an image, with a bunch of funky floating point math to whittle all that transformed data down to its most significant/perceptible bits. GPUs are supposed to be really really good at this sort of thing.
My hunch is that the problems we're seeing are caused by two big issues:
1. lack of standardization across GPU processing technologies. CUDA vs OpenCL vs Quicksync, and a bunch of tag-alongs too. Each one was designed around a particular GPU architecture, so porting programs between them is non-trivial.
2. lack of expertise in GPU programming. Let's be fair here: GPUs are a drastically different architecture than any PC or embedded platform we're used to programming. While I could follow specs and write an MPEG or H.264 encoder in any high-level language in a fairly straight-forward manner, I can't even begin to envision how I would convert that linear code into a massively parallel algorithm running on hundreds of dumbed-down shader processors. It's not at all like a conventional cluster, because shaders have very limited instruction sets, little memory but extremely fast interconnects. We have a hard enough time making CPU encoders scale to 4 or 8 cores, this requires some serious out-of-the-box thinking to pull off.
Moving to a GPU virtually requires starting over from scratch. This is a set of constraints that are very foreign to the transcoding world, where the accepted trend was to use ever-increasing amounts of cheaply available CPU and memory, with extensively configurable code paths. The potential is there, but it will take time for the hardware, APIs and developer skills to converge. GPU transcoding should be seen as a novelty for now, just like CPU encoding was 15 years ago when ripping a DVD was extremely error-prone and time-consuming. If you want a quick, low quality transcode, the GPU is your friend. If you're expecting broadcast-quality encodes, you're gonna have to wait a few years for this niche to grow and mature.
Yes, this is a predatory "cell phone" style purchase plan. Yes, it's a bad investment. Yes, there are people dumb enough to fall for it. Yes, I think they deserve to be financially punished for sucking at math and common sense.
I've been an Xbox 360 owner for a couple of years. I bought mine second-hand, actually I bought two refurbs for about the same total cost as one brand new unit, then gave the extra one to a friend. I see at least 3-4 Xboxes posted every day on my city's "used crap for sale" RSS feed, and I live in a pretty catatonic Canadian city. I would expect most people within range of a Microsoft store also have a dozen inexpensive used consoles available within walking distance of their home. For $120 you'll even get a 20 to 60gb hard drive with the console, and I've seen the Kinect gadget go for $50 or so. It'll be an older model Xbox, and no they don't catch fire or RROD ten minutes after you start your game. Mine's already got well over two thousand hours of playtime, and it's crashed exactly once, due to a shitty game full of bugs - not the hardware itself. The defective ones were from 2 or 3 hardware revisions ago, and most of those units have already died (or been repaired).
Actually, these days I'm seeing a lot of people selling their old-style Xboxes because they bought the new slim model. I don't know why they do it, but that translates into more cheap second-hand consoles flooding the market. If someone's looking to get into the Xbox for little money, that's the way to go. No contract, no overpriced payment plan, and if three months in you decide you don't like being called a "fat gay nigger cunt" ten times a day by inbred little rugrats, you can resell the cheap console in a matter of minutes.
Normally I'd have preordered two of these already, but it's too rich for my blood right now. This card is for us nutjobs who want quad-SLI and panoramic "3D Surround", with our custom-built driving cockpits and 3 large monitors, or the equally obsessive flight sim crowd. In my case, these displays run at 2560x1440 and that requires a ton of memory bandwidth on each card, just to push all those bits around.
For almost everyone else, a single $300 GPU is enough to run just about any game at 1920x1080 with very respectable settings.
As for your suggested prices, you're just talking out of your ass. If you're going to lowball the latest and greatest GPU on the market, maybe you should set games aside for a while and look at your income. Even though I agree the price is a bit high, spending $1000 on a hobby is nothing. You save up for that shit, and it lasts a very long time. My current cards are over 3 years old, so it works out to just over a dollar a day for kickass gaming graphics. Even if I played for just a few hours a week, it's still cheaper than any other form of modern entertainment. Cheaper than renting a movie, cheaper than a single pint at the pub, cheaper than basic cable TV, cheaper than bus fare to get to and from a free goddamned concert. For what I get out of it, having the highest end gaming hardware ends up being a sweet deal.
If you're doing serious GPGPU stuff, you shouldn't be relying on fickle consumer boards in the first place. This is a gaming card marketed to extreme gamers. I've fooled around with CUDA stuff like raytracing and H.264 encoding, mostly as a curiosity, but the reason I bought this quad-SLI setup years ago was for games and real-time 3D rendering. I couldn't care less about FP performance, and neither does Nvidia's target market for this product line.
GPGPU on consumer cards is still a novelty at this point. We're getting close to the tipping point, but for most users, as long as it plays their game and can handle 1080p video, they're content. If and when that balance tips in favour of OpenCL and CUDA, both GPU manufacturers will adjust their performance targets accordingly. Their #1 priority is still 3D gaming for now.
Yes, except the GTX 6xx series is slower at CUDA processing than its predecessors. This is a gaming product. Nvidia did this on purpose, sacrificing some compute speed in favour of rendering performance and power efficiency. They also artificially limit double-precision FP speed on consumer boards, to steer professional users toward the Quadro.
As a result of this hobbling, GPU computing hobbyists tend to gravitate toward the Radeon, which has outperformed the GeForce in OpenCL for a few years now, in both performance-per-dollar and performance-per-watt.
Slashdot Mirror
We had a 61% turnout last year, but since the FPTP system is so retarded, our Canadian version of Dubya won a (narrow) majority despite getting less than 40% of the votes.
The way most of us read the results, it means over 60% did NOT want that guy to win. Either way, since getting his majority he's been ramming all these big brother bills down our throat, along with unprecedented military spending and all the other abusive stuff you neighbours have been subjected to for the last few terms. Shit's going downhill fast and riots are become more and more frequent. Amazing how easiy one sellout can ruin a country for millions of people.
Adblock is your friend.
So is not getting your primary information from sites that exist solely to farm clickthroughs from aggregators like our once glorious /.
Don't tell anyone, but yea, I did read the damned thing, to the very end. It's like being at a bar, and some skinny blue-collar trainwreck starts chatting your ear off. Your fist wants to shut him up, but your brain wants to continue laughing at this guy's self-righteous hyperbole. Basically applying the Dunning-Kruger effect in entertainment form.
That guy is Scott Hanselman.
Mmmm Vibram Fivefingers. If I were sportier, I'd be all up in that bidniss. Them things are perversely comfortable.
London $ACT_OF_CANNIBALISM its own $IMPORTANT_UTILITY to $WEASEL_WORD it got $FASCIST_INTEREST
You can apply the same template everywhere. One day they're selling out to the I.O.C., the next they're hobbling their own civil freedoms to benefit the U.S. military-industrial complex. Seeing how they were a historic superpower, watching the government destroy its own economy and culture is, well frankly, fucking pathetic.
"Hey, our traffic sucks so bad, a bunch of greedy attention whores might not want to come spend our money. Quick, let's drop everything and suck their snooty cocks, rather than fixing the underlying urban problems."
Business plan for making $3.50 online:
1. Be an ignorant hipster microserf excitable attention whore
2. Write an ignorant article that makes you and your equally unenlightened followers giddy
3. Submit to slashdot and hope it's one of those new moronic editors who reviews it
4. Traffic
5. ??? (hint: cinnamon-chai lattés until your head implodes)
6. PROFIT!
This site's getting so bad, it's making Gizmodo look good.
The fundamental problem is that, for a lot of these systems, newer does not mean better.
That DOS record keeping system, is it good enough ? Would a costly rewrite and/or data migration result in useful improvements to justify the expense ?
For many businesses, it only becomes a problem when the old hardware breaks down, and the new hardware can't run the obsolete software. In practice, people would sooner run their archaic system via DosBox than have it redeveloped.
Depends on the legal drinking age where you live.
Here in Canada, it's usually 18 or 19. I don't have hard numbers but I would guess a good 5 to 10% of high school seniors could be legal.
This.
Today's jukeboxes work on a subscription model. You pay $X per month, you get a constantly rotating selection of popular music, old and new. I wouldn't be surprised if these things had their own 3G uplink, because the places I've seen them installed didn't have any WFi. My educated guess is they're installed by bar owners who can't be bothered to set up a conventional sound system. I strongly doubt they'd put in the effort to set up a network for smartphones to vote on tunes, nor would they pay for a contractor to do it.
Given that I've spent the majority of my life working with computers, I've come to accept reality as just another theory. Does the OS know it's inside a virtual machine ? (without the hypervisor intentionally making itself known) How can any person know, with absolute certainty, that they're not a brain in a jar, being fed simulated input ? How can we even know we're a brain at all ? For all I know, my entire existence could be a work of fiction, the Internet could be a fabrication of my mind, along with all its inhabitants.
The only thing we can reasonably assume, is that thought exists.
(and yes, I think the best psych/philosophy profs were the ones who dropped acid on a regular basis :)
Err... yes and no. Software has gotten absolutely awful in absolute terms, but since hardware has been enjoying tremendous speed increases, we don't notice it so much.
Just starting a PHP interpreter takes about a half-second on my bleeding-edge PC. To run the same thing on a 90's era machine would probably take 5 to 10 seconds. Writing that same code in C and compiling to the platform's native bytecode would result in near-instant execution on the order of a few hundredths of a second. Software today is bloated to hell, but it is much easiler to develop so what we lose in machine time, we gain in human time.
I don't know what kind of obsolete hardware you're using, but KDE's "noticeable lag" is in the quarter-second range, and Win8's "unusable" speed is faster than Win7 for me.
Maybe you should donate that K6-233 to the nearest dumpster.
Yes and no. I find that if you slipstream those updates into an install disc, the resulting machine is snappy. It's when you install patches into an existing system than everything goes to shit, because it keeps old versions of those files around "for compatibility". In other words, if you're running XP SP3, you have a bunch of Gold, SP1 and SP2 duplicate DLLs kicking around, occasionally being loaded in parallel to satisfy some applications that predate those releases. The result is increased disk and memory usage. Meanwhile, a fresh install of XP SP3 runs all the same apps just fine - go figure!
With 7, it stays snappy. My install has been going for nearly 3 years now, still perfectly fast and I feel no pressure to wipe it and start over.
I haven't touched KDE in about two years, in part because KDE 4 left a sour taste in my mouth, but I have to agree. Window managers on Linux have little or no impact on application performance.
That said, I do feel that several aspects of Linux operation are slower than they could be. Anything that ties into a handful of shared libraries tends to chug, as though the run-time linker might be spinning its wheels. I also find the kernel rather slow to load and boot. Not init, but the kernel itself. In fact, on my last few PCs, init finishes in about 2 seconds while the kernel needs 30-40 seconds between the bootloader and init starting. That seems a bit much, given the absurdly fast hardware at its disposal. X takes another 5 seconds to grok the graphics card and set up a root display.
Windows brings up the same hardware in about 8 seconds. It used to be the other way around, but even our beloved open-source OS has fallen prey to bloat and sloppy coding.
Not all SSDs are created equal.
I have an old install of Windows 7. I installed it the day it was released, and have not reformatted since. It's full of crap and tweaked the way I like. By far, the longest wait for me is the BIOS, which takes its sweet time to POST. I'd guesstimate about 90 seconds for that POST, and then 7-8 seconds for Windows to hit the desktop and be responsive.
The bulk of that POST time is spent waiting for disks and RAID controllers to initialize. If I had simpler hardware, it would probably take a quarter of the time.
Windows 7 in 4gb is perfectly usable. And by usable, I mean snappy. And by snappy, I mean I'm an impatient coder who starts punching things when I see an hourglass or beachball.
On 8gb, I bet it runs like a dream. I have a very large memory workstation because I do lots of VM work, but according to the task manager, my memory is mostly sitting idle in the disk cache. Actual working memory seems to hover around 3gb or so, with email, web and a game running - the type of general usage I'd expect from a non-programmer.
I've been running swap-free for years, never had any issues and software generally doesn't bother you - except Windows itself. It warns you if you disable the swap file, but I've never encountered any adverse effects. Photoshop has its own virtual memory system (scratch memory), it does not care about Windows' memory management, it will create its own temp files as needed and you can configure where these go. I point mine at an old SSD.
That's not how read caches work. Yes, it will preload a bunch of stuff it thinks you might need, but the moment an actuall application requests that memory, it is immediately freed. It doesn't need to "swap out" because the cache is just a copy of files that are already on disk, and readily available to be cached again later.
This is true of all(?) executable files and DLLs, as well as any file marked as read-only. Why write out data to the swapfile, if there's a perfectly good copy already on disk ?
Windows 7 will run with 0 swap. This is how I run it on my primary desktop, where I do everything: work, gaming, media processing. Mind you, that machine has 48gb of Ram, but even on my previous desktop with only 8gb, it ran perfectly fine. On my laptop, I run Win7 in a VM, with 3.5gb of RAM and no swap. Works fine, I use it for Eclipse Blackberry development all day long (yes, I hate my life). Win7 is far more memory and processor efficient than Vista, and boots faster than XP. If it didn't, I'd have used an XP VM instead. Eclipse certainly doesn't care what it runs on.
You should look into NLite. I've crammed XP into 128mb and 256mb virtual machines that were fully functional, running IE7/8 whatever was current a few years ago. Its idle memory usage, as reported by the task manager, was about 85mb. Turn off themes, and any background services you don't need, and XP can fit into relatively tight environments with ease.
I believe you fail to understand RAM's purpose: to store data. Your disk check probably did not require 11 GB, but since it was available, it chose to use it to improve its performance.
People who are new to Linux tend to bicker about the same thing, because the disk cache will consume whatever memory is available and report it as "used", even though it is the first thing to be released when any other software requires memory. In a sense, it is opportunistically used, but not reserved. Free RAM is wasted RAM.
Conversely, I have a 48gb system running Windows 7. It usually hovers around 8gb free when idling at the desktop. 2 gb goes to the OS and background tasks, the remaining 38gb is borrowed by the disk cache, because I tend to thrash very large video files which linger in memory after I'm done. I don't use a swap file, and I can run a bazillion apps at once without any noticeable slowdown. Yes, even a disk check, virus scan and a dozen virtual machines simultaneously.
Well see, that's the thing. A GPU is better suited to some kinds of massively parallel tasks, like video encoding. After all, you're applying various matrix transforms to an image, with a bunch of funky floating point math to whittle all that transformed data down to its most significant/perceptible bits. GPUs are supposed to be really really good at this sort of thing.
My hunch is that the problems we're seeing are caused by two big issues:
1. lack of standardization across GPU processing technologies. CUDA vs OpenCL vs Quicksync, and a bunch of tag-alongs too. Each one was designed around a particular GPU architecture, so porting programs between them is non-trivial.
2. lack of expertise in GPU programming. Let's be fair here: GPUs are a drastically different architecture than any PC or embedded platform we're used to programming. While I could follow specs and write an MPEG or H.264 encoder in any high-level language in a fairly straight-forward manner, I can't even begin to envision how I would convert that linear code into a massively parallel algorithm running on hundreds of dumbed-down shader processors. It's not at all like a conventional cluster, because shaders have very limited instruction sets, little memory but extremely fast interconnects. We have a hard enough time making CPU encoders scale to 4 or 8 cores, this requires some serious out-of-the-box thinking to pull off.
Moving to a GPU virtually requires starting over from scratch. This is a set of constraints that are very foreign to the transcoding world, where the accepted trend was to use ever-increasing amounts of cheaply available CPU and memory, with extensively configurable code paths. The potential is there, but it will take time for the hardware, APIs and developer skills to converge. GPU transcoding should be seen as a novelty for now, just like CPU encoding was 15 years ago when ripping a DVD was extremely error-prone and time-consuming. If you want a quick, low quality transcode, the GPU is your friend. If you're expecting broadcast-quality encodes, you're gonna have to wait a few years for this niche to grow and mature.
Yes, this is a predatory "cell phone" style purchase plan. Yes, it's a bad investment. Yes, there are people dumb enough to fall for it. Yes, I think they deserve to be financially punished for sucking at math and common sense.
I've been an Xbox 360 owner for a couple of years. I bought mine second-hand, actually I bought two refurbs for about the same total cost as one brand new unit, then gave the extra one to a friend. I see at least 3-4 Xboxes posted every day on my city's "used crap for sale" RSS feed, and I live in a pretty catatonic Canadian city. I would expect most people within range of a Microsoft store also have a dozen inexpensive used consoles available within walking distance of their home. For $120 you'll even get a 20 to 60gb hard drive with the console, and I've seen the Kinect gadget go for $50 or so. It'll be an older model Xbox, and no they don't catch fire or RROD ten minutes after you start your game. Mine's already got well over two thousand hours of playtime, and it's crashed exactly once, due to a shitty game full of bugs - not the hardware itself. The defective ones were from 2 or 3 hardware revisions ago, and most of those units have already died (or been repaired).
Actually, these days I'm seeing a lot of people selling their old-style Xboxes because they bought the new slim model. I don't know why they do it, but that translates into more cheap second-hand consoles flooding the market. If someone's looking to get into the Xbox for little money, that's the way to go. No contract, no overpriced payment plan, and if three months in you decide you don't like being called a "fat gay nigger cunt" ten times a day by inbred little rugrats, you can resell the cheap console in a matter of minutes.
Normally I'd have preordered two of these already, but it's too rich for my blood right now. This card is for us nutjobs who want quad-SLI and panoramic "3D Surround", with our custom-built driving cockpits and 3 large monitors, or the equally obsessive flight sim crowd. In my case, these displays run at 2560x1440 and that requires a ton of memory bandwidth on each card, just to push all those bits around.
For almost everyone else, a single $300 GPU is enough to run just about any game at 1920x1080 with very respectable settings.
As for your suggested prices, you're just talking out of your ass. If you're going to lowball the latest and greatest GPU on the market, maybe you should set games aside for a while and look at your income. Even though I agree the price is a bit high, spending $1000 on a hobby is nothing. You save up for that shit, and it lasts a very long time. My current cards are over 3 years old, so it works out to just over a dollar a day for kickass gaming graphics. Even if I played for just a few hours a week, it's still cheaper than any other form of modern entertainment. Cheaper than renting a movie, cheaper than a single pint at the pub, cheaper than basic cable TV, cheaper than bus fare to get to and from a free goddamned concert. For what I get out of it, having the highest end gaming hardware ends up being a sweet deal.
If you're doing serious GPGPU stuff, you shouldn't be relying on fickle consumer boards in the first place. This is a gaming card marketed to extreme gamers. I've fooled around with CUDA stuff like raytracing and H.264 encoding, mostly as a curiosity, but the reason I bought this quad-SLI setup years ago was for games and real-time 3D rendering. I couldn't care less about FP performance, and neither does Nvidia's target market for this product line.
GPGPU on consumer cards is still a novelty at this point. We're getting close to the tipping point, but for most users, as long as it plays their game and can handle 1080p video, they're content. If and when that balance tips in favour of OpenCL and CUDA, both GPU manufacturers will adjust their performance targets accordingly. Their #1 priority is still 3D gaming for now.
Yes, except the GTX 6xx series is slower at CUDA processing than its predecessors. This is a gaming product. Nvidia did this on purpose, sacrificing some compute speed in favour of rendering performance and power efficiency. They also artificially limit double-precision FP speed on consumer boards, to steer professional users toward the Quadro.
As a result of this hobbling, GPU computing hobbyists tend to gravitate toward the Radeon, which has outperformed the GeForce in OpenCL for a few years now, in both performance-per-dollar and performance-per-watt.