Some of the netbook Atoms didn't a year or two back; isn't that still the case today?
As far as I can tell, the N270 (Diamondville series) was the last Atom that didn't support 64-bit. A quick Google search indicates that Intel hasn't officially discontinued it, but it seems to be almost impossible to find any new products that contain one. Newegg doesn't have any netbooks using this, though they do sell a 10-pack of Intel Atom N270 motherboards. Since they don't sell individual units, I assume these may be surplus stock (and probably intended for use in embedded systems). Even if the Diamondville series is not officially discontinued, it's definitely on its way out.
When people say "x86" they usually include the 64-bit extensions in this category. Every x86 CPU made today, whether from Intel, AMD, or even Via, supports the AMD64 extensions.
We will continue to make chips for servers, and low end crap. We can't compete with Intel for the consumer market in the short to medium term, however we are still relevant in business circles.
Consumers prepared to be gouged by Intel as soon as they figure this out.
Intel really can't gouge customers too hard, or it will hasten the transition away from x86 that they fear. ARM will be a much more serious competitor once Windows 8 is released with support for it. Yes, it requires everything to be recompiled, but I don't think Intel wants to be locked into a "legacy only" niche, which is what would happen if they pushed too hard on pricing. I keep hearing that if AMD doesn't perform, Intel is going to gouge the hell out of us, but the truth is that Intel has been beating the pants off of AMD since the release of Core 2 Duo (5+ years) and the gouging still hasn't happened. The Sandy Bridge 2500K is still the best bang for the buck of any CPU anywhere. People act as if the Netburst era was somehow emblematic of Intel, when it was in fact a bizarre aberration.
ME: I can go to Fry's this afternoon and buy any number of hard drives. There isn't a shortage.
Actually, at the moment, there is. I was at Fry's just this weekend and they not only had a "1 hard drive per person" restriction, but had hiked prices by 2x-4x. A 2TB hard drive is now about $250, not $80.
I'd be interested to see evidence that consumer-grade drives are not good enough for data integrity when used in a RAID-Z2 (double parity) configuration under ZFS. Too many people in the IT world are hung up on crappy, antiquated hardware-based RAID solutions simply because that is what the biggest vendors (Dell, HP, etc.) provide.
Quite a bit of old scientific/industrial equipment must continue to run either DOS or Windows 9x because it has to talk to the hardware directly in real-time. In NT-based versions of Windows (anything from 2000 onward), this is not permitted except for kernel mode driver code. And it's a lot easier to write a DOS app that talks directly to the hardware than to write a driver.
According to company spokesman Masashiro Nagayasu, they cut a hole in the roof of the Rajana factory, sent divers into the toxin-laden waters to unbolt some heavy equipment, and lifted it onto waiting boats. Some of the equipment is now being used in Nidec factories in China and the Philippines.
Is it just me, or does this sound like the recipe for some serious QC problems coming up? I'd be very leery of trusting the output from machines that had been through all of that.
I hope, but have little faith in Intel when it comes to putting out cooler chips with lower TDP.
Ever since they came to their senses and shitcanned NetBurst, Intel has beaten AMD in performance per watt on both desktops and servers.
But I think it might actually be an attempt to beat Calxeda before it sells any servers. http://www.calxeda.com/products/energycore/ecx1000 1.5Watts per chip (5Watts at maximum power.) The point to notice is the Pentium VTd feature is missing, which means it's not meant for VM's
I think this may be a preemptive strike against all the talk of ARM servers from various companies. Calxeda is just one of many places which are planning something like this. A cheap, low-power x86 that supports server features like ECC will dissuade many users from switching to ARM servers if and when they materialize.
I'm sure that Microsoft is working on improving its command line interface, but that's very different than going "full CLI." I very much doubt the standard GUI admin tools for Windows Server are going away any time soon.
The Intel site indicates this new chip has a TDP of only 15 watts, and supports ECC RAM. This would make it great for a home server, such as a ZFS NAS, if the price is reasonable (and assuming it isn't OEM-only).
Yes there is. "Standard graphics card" means no graphics card at all, and the PC will fall back to your integrated graphics. if your PC has an Intel chipset, this will be the Intel GMA known for its underperformance. AMD CPUs, on the other hand, mean you get an NVIDIA or AMD chipset, which means at least an integrated GeForce or an integrated Radeon.
You're a bit out of date with this analysis. First of all, most CPUs now have graphics integrated in the CPU itself, not the chipset. Secondly, Intel has drastically improved their offerings, and the "HD Graphics" module used in Sandy Bridge (i.e. almost all CPUs they sell for the desktop market today) is reasonably competitive with low-end discrete cards like the HD 5450.
At this point, really, only gamers and people doing 3D modeling or some kind of GPGPU function actually need discrete graphics cards. For Aero, basic acceleration in browsers, HD video decoding, and 2D gaming, modern integrated graphics from either AMD or Intel is fine.
I've read this online book, which has some interesting tidbits but deteriorates into a self-serving rant in the second half. Peterson doesn't seem to realize how bad some of this stuff makes him look. Here's his philosophy towards employment (chapter 11):
WordPerfect Corporation was not a platform for personal achievement, a career ladder to other opportunities, or a challenging opportunity for personal improvement. The company did not put the needs of the individual ahead of its own. The company was not concerned about an employee's personal feelings, except as they related to the company's well-being.
WordPerfect Corporation was not intended to be a social club for the unproductive. While other companies might condone many personal or social activities at the office, ours did not. Things like celebrating birthdays, throwing baby showers, collecting for gifts, selling Tupperware or Avon, managing sports tournaments, running betting pools, calling home to keep a romance alive or hand out chores to the children, gossiping or flirting with co-workers, getting a haircut, going to a medical or dental appointment, running to the cafeteria for a snack, coming in a little late or leaving a little early, taking Friday afternoon off, and griping about working conditions were all inappropriate when done on company time. Even though these activities were condoned by many businesses across the country, we felt there was no time for them at WordPerfect Corporation.
Needless to say, no employees like to be micromanaged in this sort of manner. Management might be able to get away with this with low-skill workers (though even then it will lower retention and increase shrinkage) but they can't expect to get experienced coders willing to work under these conditions. Peterson made the same mistake Atari made a decade earlier: treating skilled programmers like interchangeable cogs.
For example, as it ages a bit, Win7 now crashes about as often as XP did before.
What in the world are you doing? Do you have some kind of bizarre hardware installed? I've been running Win7 on my main PC for nearly a year now and haven't had a single crash. Not one. It's a decent, but by no means awesome box (Athlon X2 245, 4GB RAM, HD 5670 video card) and I regularly leave it up and running until Windows Update says I have to reboot. I use hibernate to RAM whenever I'm away for an extended period, and it comes out OK every time. Even Firefox I can leave open for weeks without a crash. Again, are you using some kind of very cheap or obscure hardware with half-ass drivers? Sure your system doesn't have malware? My experience with Windows 7 is that it is very stable, and most other users seem to agree.
Fixing things requires the most obscure procedures that are not even logical after you have discovered them. Compared to Win7, a full-fledged modern Linux is simple, clear and easy to administrate.
This is a joke, right? When was the last time Windows made you edit a text configuration file to change stuff? When was the last time you couldn't find drivers for your hardware on Windows? (Again, this assumes you aren't using hardware that is 20 years old - on Linux, you often have problems if it's *not* 20 years old.)
Anything that requires end-users to edit text configuration files is not suitable for mainstream use. Doesn't matter what arguments you come up with in its favor; the vast majority of people cannot and will not do this.
Zlib, libpng, jpeg libraries are used in almost everything, yet these libraries are stuck as single-core libraries because they were all initially written before 2003.
Huffman compression doesn't work very well in parallel unless the underlying data structure was specifically designed for it. With JPEG, you can use SIMD instructions to boost iDCT performance, but you can't do much about the Huffman section. For that matter, LZ77 (used in zip and png) doesn't really parallellize well either.
That's the problem - you can't just replace legacy code, in many cases you have to replace the legacy algorithms as well, and that just isn't going to happen. Bzip2 parallellizes well because it first divides the data into specifically sized blocks (900k) and you can work on different blocks at the same time. The original LZMA (7zip) algorithm didn't support multithreading very well, but LZMA2 does. The thing is that for the forseeable future we are going to continue to use deflate (ZIP, gz, header compression, etc.), JPEG, PNG, and other legacy formats - and these will *never* be able to leverage multi-core systems to their full efficiency.
2) So someone needs to buy the current batch of AMD crap[1] to keep AMD alive till they come up with something better.
There's no need to buy the crap. Instead, buy the competent products that AMD still releases. Their graphics cards are competitive against nVidia's in most markets. And for non-enthusiasts who just need a basic PC (surfing, email, facebook, etc.) an AMD E-350 system is a good deal. (Only if you build a nettop yourself, though - the laptops are too close in price to faster Sandy Bridge Celerons and Pentiums to be competitive.)
That was a bit of a special case. It's not a testament of how fundamentally awesome low power processors are, and more of a illustration of *just* how bad NetBurst was. The Pentium M skipped NetBurst entirely because they *couldn't* make it work acceptably in a mobile device.
All true. The thing is that Bulldozer basically *is* NetBurst, or at least all of NetBurst's worst mistakes: long pipeline, which causes low IPC, which in turn leads to cranking up the clock speed to compensate, which causes thermal issues and high power consumption... it's just a mess all around. The money AMD spent developing it is gone. It's not coming back and spending more money on this bad architecture will only get them in a deeper hole. They need to bite the bullet, kill it now, and instead continue step-by-step improvements to their existing products which don't suck.
We tolerate 35,000 auto-related deaths a year because cars are a vital part of our transportation infrastructure. We're not going to tolerate them if the only excuse is "driving is fun."
Bulldozer can't consistently beat Phenom X6 in desktop workloads.
It can't consistently beat Magny-Cours in server workloads.
It doesn't seem to be any more power-efficient than AMD's last generation, despite being built on a smaller process node (32nm vs 45nm).
At what point does AMD simply admit Bulldozer is a failure, pull the plug, and write off the sunk costs? Putting good money after bad is a classic business mistake that has killed many companies.
AMD should continue improving their existing cores on the 32nm process (they already have some of the work done with Llano) and forget about their "revolutionary new" architecture which is basically this decade's Prescott.
Or, heck, see if it's possible to scale up the Bobcat cores for mainstream desktop use. Don't forget, Intel's very successful Core 2 Duo came from a previous design (Pentium M) that had been reserved to laptops. AMD will probably have more luck increasing performance (both raw clock and IPC) on Bobcat than trying to tame the heat, insane transistor count, and long pipeline of Bulldozer.
AMD's 64-bit extensions arguably weren't a better technology either; they just happened to be in the right place at the right time, with a solution that maintained backward compatibility with the existing x86 code base
AMD's 64-bit extensions were a better technology because they maintained backward compatibility. You can't just write off this very important real-world consideration as if it were meaningless.
On the desktop, good single-threaded performance is still more important than SMP. Furthermore, x86's legacy lock-in is more significant because users are less likely to have the source code (or a vendor willing to recompile on their behalf) than in the server space.
Slashdot Mirror
Some of the netbook Atoms didn't a year or two back; isn't that still the case today?
As far as I can tell, the N270 (Diamondville series) was the last Atom that didn't support 64-bit. A quick Google search indicates that Intel hasn't officially discontinued it, but it seems to be almost impossible to find any new products that contain one. Newegg doesn't have any netbooks using this, though they do sell a 10-pack of Intel Atom N270 motherboards. Since they don't sell individual units, I assume these may be surplus stock (and probably intended for use in embedded systems). Even if the Diamondville series is not officially discontinued, it's definitely on its way out.
Why not just buy a third-party heatsink that comes with its own backplate? You don't have to use the default mounting hardware.
When people say "x86" they usually include the 64-bit extensions in this category. Every x86 CPU made today, whether from Intel, AMD, or even Via, supports the AMD64 extensions.
We will continue to make chips for servers, and low end crap. We can't compete with Intel for the consumer market in the short to medium term, however we are still relevant in business circles.
Consumers prepared to be gouged by Intel as soon as they figure this out.
Intel really can't gouge customers too hard, or it will hasten the transition away from x86 that they fear. ARM will be a much more serious competitor once Windows 8 is released with support for it. Yes, it requires everything to be recompiled, but I don't think Intel wants to be locked into a "legacy only" niche, which is what would happen if they pushed too hard on pricing. I keep hearing that if AMD doesn't perform, Intel is going to gouge the hell out of us, but the truth is that Intel has been beating the pants off of AMD since the release of Core 2 Duo (5+ years) and the gouging still hasn't happened. The Sandy Bridge 2500K is still the best bang for the buck of any CPU anywhere. People act as if the Netburst era was somehow emblematic of Intel, when it was in fact a bizarre aberration.
ME: I can go to Fry's this afternoon and buy any number of hard drives. There isn't a shortage.
Actually, at the moment, there is. I was at Fry's just this weekend and they not only had a "1 hard drive per person" restriction, but had hiked prices by 2x-4x. A 2TB hard drive is now about $250, not $80.
I'd be interested to see evidence that consumer-grade drives are not good enough for data integrity when used in a RAID-Z2 (double parity) configuration under ZFS. Too many people in the IT world are hung up on crappy, antiquated hardware-based RAID solutions simply because that is what the biggest vendors (Dell, HP, etc.) provide.
Quite a bit of old scientific/industrial equipment must continue to run either DOS or Windows 9x because it has to talk to the hardware directly in real-time. In NT-based versions of Windows (anything from 2000 onward), this is not permitted except for kernel mode driver code. And it's a lot easier to write a DOS app that talks directly to the hardware than to write a driver.
According to company spokesman Masashiro Nagayasu, they cut a hole in the roof of the Rajana factory, sent divers into the toxin-laden waters to unbolt some heavy equipment, and lifted it onto waiting boats. Some of the equipment is now being used in Nidec factories in China and the Philippines.
Is it just me, or does this sound like the recipe for some serious QC problems coming up? I'd be very leery of trusting the output from machines that had been through all of that.
Actually I'm shocked nobody has tried cooking up a low cost low power server based on brazos.
Brazos doesn't support ECC.
I hope, but have little faith in Intel when it comes to putting out cooler chips with lower TDP.
Ever since they came to their senses and shitcanned NetBurst, Intel has beaten AMD in performance per watt on both desktops and servers.
But I think it might actually be an attempt to beat Calxeda before it sells any servers. http://www.calxeda.com/products/energycore/ecx1000 1.5Watts per chip (5Watts at maximum power.) The point to notice is the Pentium VTd feature is missing, which means it's not meant for VM's
I think this may be a preemptive strike against all the talk of ARM servers from various companies. Calxeda is just one of many places which are planning something like this. A cheap, low-power x86 that supports server features like ECC will dissuade many users from switching to ARM servers if and when they materialize.
Microsoft is going full CLI for servers soon.
Citation needed.
I'm sure that Microsoft is working on improving its command line interface, but that's very different than going "full CLI." I very much doubt the standard GUI admin tools for Windows Server are going away any time soon.
The Intel site indicates this new chip has a TDP of only 15 watts, and supports ECC RAM. This would make it great for a home server, such as a ZFS NAS, if the price is reasonable (and assuming it isn't OEM-only).
Yes there is. "Standard graphics card" means no graphics card at all, and the PC will fall back to your integrated graphics. if your PC has an Intel chipset, this will be the Intel GMA known for its underperformance. AMD CPUs, on the other hand, mean you get an NVIDIA or AMD chipset, which means at least an integrated GeForce or an integrated Radeon.
You're a bit out of date with this analysis. First of all, most CPUs now have graphics integrated in the CPU itself, not the chipset. Secondly, Intel has drastically improved their offerings, and the "HD Graphics" module used in Sandy Bridge (i.e. almost all CPUs they sell for the desktop market today) is reasonably competitive with low-end discrete cards like the HD 5450.
At this point, really, only gamers and people doing 3D modeling or some kind of GPGPU function actually need discrete graphics cards. For Aero, basic acceleration in browsers, HD video decoding, and 2D gaming, modern integrated graphics from either AMD or Intel is fine.
Presumably they meant the CPU's integrated graphics, which is ubiquitous on both Nehalem and Sandy Bridge.
Who is Dawn Stover and why should we be taking her opinions seriously?
I've read this online book, which has some interesting tidbits but deteriorates into a self-serving rant in the second half. Peterson doesn't seem to realize how bad some of this stuff makes him look. Here's his philosophy towards employment (chapter 11):
WordPerfect Corporation was not a platform for personal achievement, a career ladder to other opportunities, or a challenging opportunity for personal improvement. The company did not put the needs of the individual ahead of its own. The company was not concerned about an employee's personal feelings, except as they related to the company's well-being.
WordPerfect Corporation was not intended to be a social club for the unproductive. While other companies might condone many personal or social activities at the office, ours did not. Things like celebrating birthdays, throwing baby showers, collecting for gifts, selling Tupperware or Avon, managing sports tournaments, running betting pools, calling home to keep a romance alive or hand out chores to the children, gossiping or flirting with co-workers, getting a haircut, going to a medical or dental appointment, running to the cafeteria for a snack, coming in a little late or leaving a little early, taking Friday afternoon off, and griping about working conditions were all inappropriate when done on company time. Even though these activities were condoned by many businesses across the country, we felt there was no time for them at WordPerfect Corporation.
Needless to say, no employees like to be micromanaged in this sort of manner. Management might be able to get away with this with low-skill workers (though even then it will lower retention and increase shrinkage) but they can't expect to get experienced coders willing to work under these conditions. Peterson made the same mistake Atari made a decade earlier: treating skilled programmers like interchangeable cogs.
For example, as it ages a bit, Win7 now crashes about as often as XP did before.
What in the world are you doing? Do you have some kind of bizarre hardware installed? I've been running Win7 on my main PC for nearly a year now and haven't had a single crash. Not one. It's a decent, but by no means awesome box (Athlon X2 245, 4GB RAM, HD 5670 video card) and I regularly leave it up and running until Windows Update says I have to reboot. I use hibernate to RAM whenever I'm away for an extended period, and it comes out OK every time. Even Firefox I can leave open for weeks without a crash. Again, are you using some kind of very cheap or obscure hardware with half-ass drivers? Sure your system doesn't have malware? My experience with Windows 7 is that it is very stable, and most other users seem to agree.
Fixing things requires the most obscure procedures that are not even logical after you have discovered them. Compared to Win7, a full-fledged modern Linux is simple, clear and easy to administrate.
This is a joke, right? When was the last time Windows made you edit a text configuration file to change stuff? When was the last time you couldn't find drivers for your hardware on Windows? (Again, this assumes you aren't using hardware that is 20 years old - on Linux, you often have problems if it's *not* 20 years old.)
Anything that requires end-users to edit text configuration files is not suitable for mainstream use. Doesn't matter what arguments you come up with in its favor; the vast majority of people cannot and will not do this.
Zlib, libpng, jpeg libraries are used in almost everything, yet these libraries are stuck as single-core libraries because they were all initially written before 2003.
Huffman compression doesn't work very well in parallel unless the underlying data structure was specifically designed for it. With JPEG, you can use SIMD instructions to boost iDCT performance, but you can't do much about the Huffman section. For that matter, LZ77 (used in zip and png) doesn't really parallellize well either.
That's the problem - you can't just replace legacy code, in many cases you have to replace the legacy algorithms as well, and that just isn't going to happen. Bzip2 parallellizes well because it first divides the data into specifically sized blocks (900k) and you can work on different blocks at the same time. The original LZMA (7zip) algorithm didn't support multithreading very well, but LZMA2 does. The thing is that for the forseeable future we are going to continue to use deflate (ZIP, gz, header compression, etc.), JPEG, PNG, and other legacy formats - and these will *never* be able to leverage multi-core systems to their full efficiency.
2) So someone needs to buy the current batch of AMD crap[1] to keep AMD alive till they come up with something better.
There's no need to buy the crap. Instead, buy the competent products that AMD still releases. Their graphics cards are competitive against nVidia's in most markets. And for non-enthusiasts who just need a basic PC (surfing, email, facebook, etc.) an AMD E-350 system is a good deal. (Only if you build a nettop yourself, though - the laptops are too close in price to faster Sandy Bridge Celerons and Pentiums to be competitive.)
That was a bit of a special case. It's not a testament of how fundamentally awesome low power processors are, and more of a illustration of *just* how bad NetBurst was. The Pentium M skipped NetBurst entirely because they *couldn't* make it work acceptably in a mobile device.
All true. The thing is that Bulldozer basically *is* NetBurst, or at least all of NetBurst's worst mistakes: long pipeline, which causes low IPC, which in turn leads to cranking up the clock speed to compensate, which causes thermal issues and high power consumption... it's just a mess all around. The money AMD spent developing it is gone. It's not coming back and spending more money on this bad architecture will only get them in a deeper hole. They need to bite the bullet, kill it now, and instead continue step-by-step improvements to their existing products which don't suck.
We tolerate 35,000 auto-related deaths a year because cars are a vital part of our transportation infrastructure. We're not going to tolerate them if the only excuse is "driving is fun."
Bulldozer can't consistently beat Phenom X6 in desktop workloads.
It can't consistently beat Magny-Cours in server workloads.
It doesn't seem to be any more power-efficient than AMD's last generation, despite being built on a smaller process node (32nm vs 45nm).
At what point does AMD simply admit Bulldozer is a failure, pull the plug, and write off the sunk costs? Putting good money after bad is a classic business mistake that has killed many companies.
AMD should continue improving their existing cores on the 32nm process (they already have some of the work done with Llano) and forget about their "revolutionary new" architecture which is basically this decade's Prescott.
Or, heck, see if it's possible to scale up the Bobcat cores for mainstream desktop use. Don't forget, Intel's very successful Core 2 Duo came from a previous design (Pentium M) that had been reserved to laptops. AMD will probably have more luck increasing performance (both raw clock and IPC) on Bobcat than trying to tame the heat, insane transistor count, and long pipeline of Bulldozer.
AMD's 64-bit extensions arguably weren't a better technology either; they just happened to be in the right place at the right time, with a solution that maintained backward compatibility with the existing x86 code base
AMD's 64-bit extensions were a better technology because they maintained backward compatibility. You can't just write off this very important real-world consideration as if it were meaningless.
On the desktop, good single-threaded performance is still more important than SMP. Furthermore, x86's legacy lock-in is more significant because users are less likely to have the source code (or a vendor willing to recompile on their behalf) than in the server space.