Well, I was responding to Zuriel who considered one as replacement for his "too noisy in idle" 4870, but said he is happy with the performance.
So the 6850 would be suitable for him. Even the 5770 might work for reducing idle noise while keeping the performance level of the 4870 (but I'd prefer the 6850 over the 5770).
Think as far ahead with the motherboard/cpu socket/RAM slots and type/expansion slots as your budget allows. Second priority...the hard drive. %00 GB minimum nowadays. Fill in the blanks with lower to medium price components-these can be upgraded piecemeal as your finances allow. ALWAYS look at 'bang for the buck' for all of the above. Here YMMV, depending on your own definition of best 'bang for buck'. Different needs/desires/goals change the definition.
Specifically for CPUs, I'd like to add a that a medium price CPU ("medium" defined as half the price of the fastest on the market) often offers more than 80% of the performance of said fastest part.
In GPU cards, the performance seems a bit more in proportion to the price. But even there, you tend to get more "bang for the buck" if you go for one step below the maximum performance parts. The current AMD release (yes we get back on topic;-) is a bit special as AMD did not release the high end parts first, despite the 68xx version number. These are more like "upper midrange" parts that count as cheaper alternative to the 58xx cards of the previous generation.
Looking at the specs, it seems the 6870 might be about equal to the 5850 in performance. Also, power consumption under load is the same.
Looking at the prices at alternate.de, the 6870 is about 10-20% cheaper than the the 5850. So we have a 10-20% improvement in performance/price. Better than nothing, but no spectacular improvement.
According to Wikipedia, both use about 150W under full load. See http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units for a comparison table. But at the same time, the 6870 is of course faster. So if you don't need the extra performance, what about a 6850? Should still be an upgrade in performance, have at least the same power advantage as the 6870 at idle, and uses only 127W under load.
But Microsoft seems to be really flexible on those Very Important Customer deals. The examples I used are about -no DRM (I think product activation is a form of DRM) -giving access to the otherwise carefully guarded sources.
For sufficiently large customers or groups of customers, it has always been possible to get special offers from Microsoft. I think the city of New York qualifies for this category.
For comparison: In the early 90s, Microsoft created the XP Corporate edition that does not need activation, to appease large customers who were worried about losing the ability to install their copies of XP. Later, Microsoft shared source code with some universities and government agencies to counter the advantage of Open Source in being more accessible for security audits.
Those who are interested in fast cars will usually make sure to buy the biggest engine (GPU/CPU) they can afford. Your average ricer kids (gaming nerds) are also likely to obsess about technical details and be at least somewhat well informed. They might also decorate the car (PC) with lots of spoilers (LED-illuminated fans). And then they go drag racing (comparing benchmarks;-)
It might be a medium-performance GPU by today's standards, integrated with a medium-performance CPU. Right now, AMD has (for instance) the following sepatate parts: -the 5570 GPU, a midrange GPU with 128 bit data bus. Complete cards use max. 43Watt (according to alternate.de) -the Athlon X4 610e (e for "energy efficient"), a "Propus" quadcore with 45Watt TDP. Put both on the same chip, assume some more improvement in power consumption, and you might get something loke the Llano. Except maybe for the shared memory bandwidth, that chip would beat my current system. Which is perfectly sufficient for games that are a few years old.
Microsoft has a problem there - break too many old applications and people actually might as well go to Ubuntu. So they try to get away with a compromise design that tries to add security while keeping most old stuff running. Which is only halfway successful. Take the "virtual store" in Vista and Windows 7 for example: For old applications that write data into their installation directory, the user gets a personal copy of the files in question into his user profile. Works fine after first installation and for one user. But as soon as you use the software from different accounts or uninstall/reinstall, you will see differences to Windows XP (let alone Windows 98).
I tend to agree that at some point Windows will need to break a lot of the old "standards". But doing so will hurt Microsoft for a while. Just like the big switch to OS X was difficult for Apple.
Sometimes, you have commas in your data. That makes it desirable to have the option of using another delimiter. Numbers in German, for instance, use the comma as decimal separator. Insisting on CSV files using the comma as delimiter would cause problems a soon as you have some number strings in your data.
Fortunately, there is a way around this (version: Office 2000): Save to.txt, open in Excel. A "text conversion assistant" will pop up, and offer you a choice of how to import. Including different delimiters.
Good points, but there may be some cases where a file format is actually patented (despite exclusions in patent law).
In that particular situation, I think the EU should actually do what the BSA letter suggests, and refrain from using that particular technology for its public standards. In the probably most relevant field for government bureaucracies, formats for office documents, there is already a good enough open standard in the form of ODF.
But is it still the same kind of hardware? With graphics pipelines that can run increasingly complex programs, the GPUs of today are not quite the spcialized devices of several years ago. It seems to me that "running task X on the GPU" is an overly broad and poorly specified claim that should be rejected for lack of accuracy.
It must seriously fall apart at higher speed if you need to put in a bypass at 70+.
Probably the motor/generator combination is not capable of delivering the power that is needed at those speeds. The power demand grows exponentially with speed. The motor is wound with a wire of a certain AWG, and it can only consume (or deliver) so many Amperes before it goes up in flames. The semiconductors that feed power to it also have a maximum rating, and it is probably thermally limited as well.
That is a version of "falling apart". Looks like a design error to me (assuming your speculation is accurate): If the motor/generator combination of the range extender tends to overheat at prolonged maximum output, either the gasoline engine is overdimensioned or the generator too weak. Especially considering that 70+ mph might not cover all scenarios that apply: If you drive up a mountain pass at high speed, but below 70+, the mechanical connection would not engage, and we are back to the generator overheating at some point.
The Prius as currently sold has a much smaller battery than the Volt. Even if you could plug it in, it would be hardly worthwhile because the all-electric range is much smaller. BTW there is an aftermarket conversion system that adds a larger battery and a charger. With that, the Prius becomes a lot more comparable to the Volt.
Then one day they realize they need to upgrade 30+ system all at once for some new piece of software they want.
If that is the only problem, "they" may actually be doing it right. Because up to this point, the systems did all that was required. And this also puts the need to upgrade in the proper perspective: Do you want that one piece of software badly enough that you are ready to buy new hardware too?
Which is why an operating system should be written in a language which requires the developer to consciously open up pointer manipulation and disable bounds checking rather than placing the burden on the developer to carefully check bounds everywhere on their own. The latter is virtually guaranteed to leave security holes as developers will undoubtedly forget to check bounds in at least a few places.
Good point, and I'd like to add that much of it could be achieved in "traditionally unsafe" languages too with compilers that offer bounds and range checking. You just need to activate those. I know for sure that Borland Delphi offers compiler switches to turn bounds and range checking on and off, just as an example.
Harder to get in with what? Following the IT news, it is obvious that Apple is good at rejecting things like fart apps and porn. In other words, things that are fairly obvious.
A halfway intelligently written trojan, however, will come across as a well-behaved and useful application and try to hide its malicious nature. I have yet to read about a case where Aplle managed to identify one of these up front and reject it.
Migrating applications to a new Windows installation is tricky anyway. In my experience, the most common problem is that the app puts some essential config data into a non-obvious registry key. You don't know about that => you fail to get the migrated application to work.
Usually, just backing up the documents and installing the apps from scratch on the new OS is the way to go.
Oh, I won't bet against it. Until recently, software from the company I worked for used to dump configuration data (which may be changed during everyday operation) into the application directory. That is C:\Program Files\$INSTALLDIR or something similar, again depending on OS language.
So I pushed for changing that on the project I was responsible for, and at least the.ini-file moved to CSIDL_COMMON_APPDATA. I can only hope the rest of the project will follow.
There is one "special folder" that is meant for user-independent storage of application data. You can get its name with feeding the value CSIDL_COMMON_APPDATA into the proper API call. See http://www.mvps.org/access/api/api0054.htm for an introduction. The actual path varies with Windows version and language. On my German Win XP Pro, for instance, it is C:\Dokumente und Einstellungen\All Users\Anwendungsdaten Also note that every user can create files and directories in there, but per default they are only writable to the user who created them. So it may be necessary to change permissions during installation.
I mostly agree. After some hands-on experience with Win7, I found:
-Performance seems not much different from XP. -UAC is a good idea for the security-conscious, but if you are careful about what you install, you don't absolutely need it. -My impressions of the GUI are mixed: Some details work better (like ejecting USB sticks from Windows Explorer) or are more intuitive, but like in earlier versions, the trend of hiding the low level functions under several layers of wizards and menus has continued. For the newbie this may be helpful, for experienced users it is more of an annoyance.
Bottom line: Windows 7 is OK, but not so much better that going and buying it for an existing computer is worthwhile. I expect that most buyers of new computers will accept it without the resistance Vista got, but TFA is right in pointing out there is no rush towards Windows 7.
Lowering the bar is all this appeal to the Supreme Court is about. it would be unrealistic to expect more.
Getting rid of software patents (or all patents) is a matter for Congress. It might also require the USA to withdraw from some international agreements.
Actually, I think the "Chat is limited to a pre-defined dictionary list" will be the most obvious and annoying restriction. Because each typo means you have technically entered a word that is not in the dictionary and your message will be blocked. Maybe a few common typos will be allowed for convenience, but I don't think that will really solve the problem.
Slashdot Mirror
Well, I was responding to Zuriel who considered one as replacement for his "too noisy in idle" 4870, but said he is happy with the performance.
So the 6850 would be suitable for him. Even the 5770 might work for reducing idle noise while keeping the performance level of the 4870 (but I'd prefer the 6850 over the 5770).
Think as far ahead with the motherboard/cpu socket/RAM slots and type/expansion slots as your budget allows.
Second priority...the hard drive. %00 GB minimum nowadays.
Fill in the blanks with lower to medium price components-these can be upgraded piecemeal as your finances allow.
ALWAYS look at 'bang for the buck' for all of the above. Here YMMV, depending on your own definition of best 'bang for buck'. Different needs/desires/goals change the definition.
Specifically for CPUs, I'd like to add a that a medium price CPU ("medium" defined as half the price of the fastest on the market) often offers more than 80% of the performance of said fastest part.
In GPU cards, the performance seems a bit more in proportion to the price. But even there, you tend to get more "bang for the buck" if you go for one step below the maximum performance parts. The current AMD release (yes we get back on topic ;-) is a bit special as AMD did not release the high end parts first, despite the 68xx version number.
These are more like "upper midrange" parts that count as cheaper alternative to the 58xx cards of the previous generation.
Looking at the specs, it seems the 6870 might be about equal to the 5850 in performance. Also, power consumption under load is the same.
Looking at the prices at alternate.de, the 6870 is about 10-20% cheaper than the the 5850. So we have a 10-20% improvement in performance/price. Better than nothing, but no spectacular improvement.
According to Wikipedia, both use about 150W under full load. See http://en.wikipedia.org/wiki/Comparison_of_AMD_graphics_processing_units for a comparison table.
But at the same time, the 6870 is of course faster. So if you don't need the extra performance, what about a 6850?
Should still be an upgrade in performance, have at least the same power advantage as the 6870 at idle, and uses only 127W under load.
In the case of NYC, obviously yes.
But Microsoft seems to be really flexible on those Very Important Customer deals. The examples I used are about
-no DRM (I think product activation is a form of DRM)
-giving access to the otherwise carefully guarded sources.
For sufficiently large customers or groups of customers, it has always been possible to get special offers from Microsoft. I think the city of New York qualifies for this category.
For comparison:
In the early 90s, Microsoft created the XP Corporate edition that does not need activation, to appease large customers who were worried about losing the ability to install their copies of XP.
Later, Microsoft shared source code with some universities and government agencies to counter the advantage of Open Source in being more accessible for security audits.
Those who are interested in fast cars will usually make sure to buy the biggest engine (GPU/CPU) they can afford. ;-)
Your average ricer kids (gaming nerds) are also likely to obsess about technical details and be at least somewhat well informed.
They might also decorate the car (PC) with lots of spoilers (LED-illuminated fans).
And then they go drag racing (comparing benchmarks
It might be a medium-performance GPU by today's standards, integrated with a medium-performance CPU.
Right now, AMD has (for instance) the following sepatate parts:
-the 5570 GPU, a midrange GPU with 128 bit data bus. Complete cards use max. 43Watt (according to alternate.de)
-the Athlon X4 610e (e for "energy efficient"), a "Propus" quadcore with 45Watt TDP.
Put both on the same chip, assume some more improvement in power consumption, and you might get something loke the Llano. Except maybe for the shared memory bandwidth, that chip would beat my current system. Which is perfectly sufficient for games that are a few years old.
Microsoft has a problem there - break too many old applications and people actually might as well go to Ubuntu. So they try to get away with a compromise design that tries to add security while keeping most old stuff running. Which is only halfway successful. Take the "virtual store" in Vista and Windows 7 for example:
For old applications that write data into their installation directory, the user gets a personal copy of the files in question into his user profile. Works fine after first installation and for one user. But as soon as you use the software from different accounts or uninstall/reinstall, you will see differences to Windows XP (let alone Windows 98).
I tend to agree that at some point Windows will need to break a lot of the old "standards". But doing so will hurt Microsoft for a while. Just like the big switch to OS X was difficult for Apple.
Sometimes, you have commas in your data. That makes it desirable to have the option of using another delimiter. Numbers in German, for instance, use the comma as decimal separator. Insisting on CSV files using the comma as delimiter would cause problems a soon as you have some number strings in your data.
Fortunately, there is a way around this (version: Office 2000): .txt, open in Excel. A "text conversion assistant" will pop up, and offer you a choice of how to import. Including different delimiters.
Save to
Good points, but there may be some cases where a file format is actually patented (despite exclusions in patent law).
In that particular situation, I think the EU should actually do what the BSA letter suggests, and refrain from using that particular technology for its public standards. In the probably most relevant field for government bureaucracies, formats for office documents, there is already a good enough open standard in the form of ODF.
But is it still the same kind of hardware?
With graphics pipelines that can run increasingly complex programs, the GPUs of today are not quite the spcialized devices of several years ago. It seems to me that "running task X on the GPU" is an overly broad and poorly specified claim that should be rejected for lack of accuracy.
It must seriously fall apart at higher speed if you need to put in a bypass at 70+.
Probably the motor/generator combination is not capable of delivering the power that is needed at those speeds. The power demand grows exponentially with speed. The motor is wound with a wire of a certain AWG, and it can only consume (or deliver) so many Amperes before it goes up in flames. The semiconductors that feed power to it also have a maximum rating, and it is probably thermally limited as well.
That is a version of "falling apart". Looks like a design error to me (assuming your speculation is accurate):
If the motor/generator combination of the range extender tends to overheat at prolonged maximum output, either the gasoline engine is overdimensioned or the generator too weak. Especially considering that 70+ mph might not cover all scenarios that apply:
If you drive up a mountain pass at high speed, but below 70+, the mechanical connection would not engage, and we are back to the generator overheating at some point.
The Prius as currently sold has a much smaller battery than the Volt. Even if you could plug it in, it would be hardly worthwhile because the all-electric range is much smaller.
BTW there is an aftermarket conversion system that adds a larger battery and a charger. With that, the Prius becomes a lot more comparable to the Volt.
Then one day they realize they need to upgrade 30+ system all at once for some new piece of software they want.
If that is the only problem, "they" may actually be doing it right. Because up to this point, the systems did all that was required. And this also puts the need to upgrade in the proper perspective:
Do you want that one piece of software badly enough that you are ready to buy new hardware too?
Which is why an operating system should be written in a language which requires the developer to consciously open up pointer manipulation and disable bounds checking rather than placing the burden on the developer to carefully check bounds everywhere on their own. The latter is virtually guaranteed to leave security holes as developers will undoubtedly forget to check bounds in at least a few places.
Good point, and I'd like to add that much of it could be achieved in "traditionally unsafe" languages too with compilers that offer bounds and range checking. You just need to activate those. I know for sure that Borland Delphi offers compiler switches to turn bounds and range checking on and off, just as an example.
Google [...] could probably write a PDF viewer in less than 100 megabytes.
Try Foxit Reader. Download size 7.5 Mbyte. Works well too, I dumped the Acrobat Reader after discovering Foxit.
Harder to get in with what? Following the IT news, it is obvious that Apple is good at rejecting things like fart apps and porn. In other words, things that are fairly obvious.
A halfway intelligently written trojan, however, will come across as a well-behaved and useful application and try to hide its malicious nature. I have yet to read about a case where Aplle managed to identify one of these up front and reject it.
Migrating applications to a new Windows installation is tricky anyway. In my experience, the most common problem is that the app puts some essential config data into a non-obvious registry key. You don't know about that => you fail to get the migrated application to work.
Usually, just backing up the documents and installing the apps from scratch on the new OS is the way to go.
Oh, I won't bet against it. Until recently, software from the company I worked for used to dump configuration data (which may be changed during everyday operation) into the application directory. That is C:\Program Files\$INSTALLDIR or something similar, again depending on OS language.
But with Vista and Windows 7, the "Virtual Store" feature makes the behavior of writing to this directory pretty difficult to predict. See also http://www.itwriting.com/blog/141-peeking-into-vistas-virtual-store.html. The rules are somewhat complicated.
So I pushed for changing that on the project I was responsible for, and at least the .ini-file moved to CSIDL_COMMON_APPDATA. I can only hope the rest of the project will follow.
There is one "special folder" that is meant for user-independent storage of application data. You can get its name with feeding the value CSIDL_COMMON_APPDATA into the proper API call.
See http://www.mvps.org/access/api/api0054.htm for an introduction.
The actual path varies with Windows version and language. On my German Win XP Pro, for instance, it is
C:\Dokumente und Einstellungen\All Users\Anwendungsdaten
Also note that every user can create files and directories in there, but per default they are only writable to the user who created them. So it may be necessary to change permissions during installation.
I mostly agree. After some hands-on experience with Win7, I found:
-Performance seems not much different from XP.
-UAC is a good idea for the security-conscious, but if you are careful about what you install, you don't absolutely need it.
-My impressions of the GUI are mixed: Some details work better (like ejecting USB sticks from Windows Explorer) or are more intuitive, but like in earlier versions, the trend of hiding the low level functions under several layers of wizards and menus has continued. For the newbie this may be helpful, for experienced users it is more of an annoyance.
Bottom line:
Windows 7 is OK, but not so much better that going and buying it for an existing computer is worthwhile. I expect that most buyers of new computers will accept it without the resistance Vista got, but TFA is right in pointing out there is no rush towards Windows 7.
Lowering the bar is all this appeal to the Supreme Court is about. it would be unrealistic to expect more.
Getting rid of software patents (or all patents) is a matter for Congress. It might also require the USA to withdraw from some international agreements.
Actually, I think the "Chat is limited to a pre-defined dictionary list" will be the most obvious and annoying restriction. Because each typo means you have technically entered a word that is not in the dictionary and your message will be blocked. Maybe a few common typos will be allowed for convenience, but I don't think that will really solve the problem.
Sounds like the kind of hyper-censored environment that even a 7-year old will be bored in. Let alone older kids.
I won't play that game for certain.