GCC may do most of the stuff MSVC does, give or take, but there is a difference between just doing it and actualy doing it well.
At the heart of the matter is the main codegen. Simply forgetting about all the advanced optimization features for a moment, which will produce better code out of the same source when no such advanced optimizations are possible/allowed?
The evidence swings strongly towards GCC having rather poor codegen in relation to MSVC, and that additionally MSVC has rather poor codegen in relation to ICC.
In terms of compilers, GNU
This is actualy well known to optimization fanatics. If you want the best quality machine code you go with ICC. If you want the best production environment you go with MSVC. If you want the most portable source code you go with GCC. GCC has its good qualities, but generating near optimal machine code isnt one of them... its not even close.
It happens because flash drives write on very large (512KB) "blocks", but they still pretend like they have average sized (4KB) "sectors".
Essentialy what the intel write-combining technology is doing is combining multiple small (4KB) writes into a single block, and letting the old block become fragmented (having a bunch of 4KB holes in it.)
The scenario in the nutshell:
You have a 1MB file and a program which modifies a single 4KB chunk of it. Intels technology marks the original 4KB chunk within its original "block" as erased, and then allocates a new block (using the wear leveling algorithm) to hold the new version of the 4KB chunk and additionally combines it with any other small writer operations that may have recently occured or will recently occur. Up to 128 such 4KB writes can be combined into a single block write.
After this is done many hundreds of thousands of times, however, the drive begins to be in a state where nearly every "block" is only partialy used. The write combiner itself is stuck with whatever the wear leveling algorithm handed it, which is now a partialy used block instead of a fully virgin block. It can no longer combine 128 small 4KB writes together, but maybe only has space to combine 10 of them, or in the worst case scenario.. 1 of them.
I find it suspicious for another reason. Certain "undesirable" nations have recently started space programs, and a debris field could very well shift the situation back to "prohibitively expensive" for them, but not for the Big 4 (United States, Russia, China, and the European Union)
By my calculation, 32nm only offers ~2.7 times as many cores over 45nm, all other things being equal (most importantly, equal chip size)
So basically 32nm will offer ~10 symmetric i7 cores.
Of course most software is single threaded and that means that just adding symmetric cores doesnt linearly add value, so its rather obvious that assymetric is the near future, targeting specifically the highly parallel market.. which means jumping directly from 4 cores to 4/16 solutions.
The only scientists who claim that McIntyre's work has been debunked are the people he has exposed, or their direct underlings. His attention to detail is an embarassement for the likes of Mann, Jones, and Wang, who each obstruct independent validations of their work.
For the record, Sir, both of these men you are talking about are on the IPCC expert review panel. Their work has been published in peer reviewed journals, and unlike some other folks out there, they actually support and promote the independent verification of their work.
They are good enough for the IPCC, but not good enough for you.. is that it?
I'd MOD him up, but he cited Mann as verification for Mann... which is presented on a site which actively sensors content contributions that are contrary to what Mann is pushing.
It is pretty black and white when you make official requests for their data and source code and do not recieve them, so to have to resort to Freedom of Information requests which themselves get ignored until lawyers get involved, and only after many many years might you recieve the data required to validate their work but quite often you are met with the situation that "the data is no longer available." Even large scientific bodies such as the NOAA drag their feet and obfuscate when these requests are made.
This goes right to the peer review process. This stuff is supposed to be validatable, but even years after publication which is supposed to be post-validation, you are fighting to get the data needed to validate. Often a requirement of a Journal (such as Science, or Nature) is also that the data is to be archived and available, but the standard when it comes to Climate researchers who are publishing is that its simply OK that the data is neither archived nor available... that nobody bothered to do any validation at all.
The peer review process is a complete failure in the climate sciences. It appears to truely be a clique of climate scientists blindly signing off on each others work while they rake in the government grants.
I am reminded of how some circuits such as sorting networks are designed with redundancy such that a certain amount of sub-component error is tolerable. They will still sort their inputs even if multiple sub-components get their output wrong.
It seems to me that we have already been designing things in a probabalistic manner.. all that these new guys are doing is allowing the error tolerance to be much more liberal.
..and VBA support is *HUGE* in the commercial world.
People think that C and C++ is popular, and if you go by how much commercial software is developed then you would be right. But if you include all corporate in-house development, VBA busts every other language down to "novelty" status. Big corporations have hundreds of in-house VBA developers, each churning out program after program, day in and day out, with no recognition from any of the "most popular language" rankings. The millions of custom VBA programs out there that corporations use are essential assets to their operations that cannot be replaced without considerable expense.
You cannot capture these corporations as office suite customers without having VBA support. It will NEVER, EVER, HAPPEN. VBA is the driving force, not the particular office suite.
If microsoft designed a package manager and then enforced its use, you all would cry about microsofts unfair monopoly advantage over competing products such as Inno Setup.
After a few months, the EU would file a lawsuit against Microsoft seeking punative monetary damages for its clear violation of EU anti-trust statutes.
Your post went on about how it is unacceptable for microsoft to interfere with how 3rd party programs work. If you did not mean that, when why didnt you say what you meant? Could it have been... HATE?
Installing the plugin allows Firefox to work with ClickOnce software. Thats all. It *IS* a compatability shim, and compatability with ClickOnce is a GOOD THING, since ClickOnce offers a much better (safer) alternative to downloading naked binary installers. This is something *EVERY* windows browser should support.
You realize that Microsoft interferes with a lot of 3rd party programs FOR YOUR BENEFIT, right?
I guess not.
They have a LOT of application compatability shims in their OS's so that older applications continue to work on an operating system they werent designed for. Vista is FILLED with this stuff, even going so far as to virtualize disk access completely rerouting read and write operations to new locations.
I didn't hear you complain about it before. I wonder why. Could it be... HATE?
Did you really want a migration back to IE because grandma's shit didnt work under firefox? You should be thanking microsoft for being so fucking magnanimous.
Did you seriously wonder what gives them the right to change the way a new version of their OS behaves?
And as for no increase in value..
You must be a youngster. The value of computers has drastically gone up. There was a time when you couldn't play video *AT ALL* on a desktop computer.
Heres a thought.. If you do not think a new version of the OS provides value and you also do not think that a new version of the software you run provides value, then why the hell did you upgrade either of them?
If you are responsible for this serious misallocation of funding in your company, then let me know what company that is so that I can get you fired for your complete and utter incompetance and mismanagement of the companies funds. Oh I get it.. you had a budget to blow through or risk of having the budget reduced next year...
The corporate world doesnt leave XP because Vista simply doesnt run all the crap they need to run while there remains no incentive to add in another layer of their tangled mess of abstractions. Thats the damned truth.
Microsoft screwed the pooch by not listening to its biggest clients. They did very well when they paid attention to the TOP10 customers, which are corporate giants with tens of thousands and even hundreds of thousands of installations, running thousands of different programs (many custom) with hundreds of driver sets for equipment that simply isnt manufactured anymore (the manufacturers are out of business! Some for over 2 decades!)
Basically, there are no alternatives other than compatability for these giants. If its not compatible, then its not acceptable, because compatability has become MISSION CRITICAL.
Even if these giants had a blank check book in which to "upgrade" everything so that a route towards the "future" existed, it would still take a DECADE to implement such a change simply because of the massiveness of it all. These guys have to run 2000/XP because they have a collection of thousands of custom VB/VBA programs and scripts, all written inhouse, that are MISSION CRITICAL and used by tens of thousands of people, yet will not work well on anything more modern. Linux doesnt offer any route to using all that mission critical code. OSX doesnt offer any route to using all that mission critical code. VISTA nags with UAC because when all this crap was written, the things they do was considered standard practice (even recommended practice by Microsoft themmselves in some cases.)
Now, some people might suggest that they should convert from Microsoft to something like Linux, regardless of the cost.. but that doesnt prevent the problem at all. Eventualy they will get locked into a specific distribution, and then to a specific version of that distribution, simply because something mission critical has to continue to work no matter what.
Its easy to say "just rewrite it" when its some top or bottom layer thing, but when what you are rewriting is within some tangled mess of middleware which not only needs to talk to that old VMS program (yes, VMS), it needs to do so through that DOS client because the only platform that can run the softweare which talks to some 20 year old device happens to require DOS, which is controlled by a hidden terminal sitting under a win16 client using the KERMIT protocal, and THAT client is now controlled by a WIN32 client, which itself also talks to both an oracle and sql database to keep the VMS data and the data used by other clients with just as much bullshit in synch.
You get the fucking picture? When you got one guy in one office using this software.. then maybe, just maybe, you could afford some downtime and the monetary costs while its retooled.. but when ten thousand people from offices around the globe are using this software.. your just fucking nuts if you think anything other than continued compatability with the past is acceptable.
We still run FORTRAN and COBOL code for christ sakes, and thats kinda trivial to convert to another language when compared to the mosterous job of retooling the tangled web of technology found in mega-corporations.
When these mega-corporations do jump to a new layer, some account somewhere is going to be recieving a wire transfer of 10 or 11 digits just for licensing costs.. they don't do these things lightly.. they will only add that layer when some new "killer" application requires it, and it doesnt matter what OS its for because they already know the solution to that problem: more middleware.
Raytracing touches far less memory than a rasterizer, so WTF are you talking about?
Oh I get it, you want to give rasterizers the benefit of multiple independent memory controllers, but not raytracers.
Oh, and then you stick the GPU on a pedestal.. saying how its the only thing with enough processors and memory bandwidth..
umm.. helllo?? Do you know what Intel has been working towards? Have you missed the past 5 years of research papers comming out of Intel about this very subject? Its called Larrabee and they are well on their way to making it a reality. They are doing to research so that they know what the requirements will be.
nVidia has put out more than one press release saying that you do not need Intel to do realtime raytracing, that their GPU's will be able to do it.
If GPU makers are working towards RTRT, and Intel is also working towards RTRT, what does that make you? Just ignorant?
You cannot avoid big-O with horseshit. Your artists take too long to create models? Heres an idea.. design better tools. Thats whats been happening for years and years. Better tools for more complex scenery.
The big-O is comming to shatter your delusions, and NOTHING can stop it.
I realize that my social skills arent exactly stellar here, but I'm sick and tired of people from OTHER FIELDS who think that they have a clue about big-o when really they are completely and obviously ignorant in the eyes of every single expert programmer. Its like non-physicists telling physicists all about their perpetual motion machines, while expecting to be treated with a gentle hand. Big-o is like the laws of thermodynamics. Undeniable.
Rasterization is a corner case, just like bubble sort outperforms mergesort in cases of small N. Intel has also done research in regards to apprxominately what N needs to be where raytracing will begin to blow the doors right off of rasterization. You can go to their research site and actualy read this stuff. Maybe you should.
The real problem with backing these suckers up is how long it takes. HotHardware has this thing listed at an average 88MB/sec transfer rate in benchmarks (noting that the firmware may not be final.) They do not have write speed listed but we can be generous and allow write speed to match read speed.
Math is fun. Lets go crunching!
It will take over 6.31 hours to read/write the entire contents of one of these, and another 6.31 hour to verify the copy. Thats 12.62 hours of constant spinning and head moving.
Want a daily backup? Then 52.6% of every single day just backing it up, and a lifespan of probably 6 months.
Maybe you want to back it up over a network connection to some place offsite. Lets say you got a fairly phat pipe, like a T3 which in fantasy ideal land peaks out at 44.736 Mbps. It would take 99 hours to transfer one of these drives over such a pipe. If this were a weekly backup process, 59.2% of the week would be taken up making the copy.
It would take 426 single layer DVD-R's if you wanted to burn it. I'll leave the calculation of how long THAT would take to someone else. I know that if I was making such a copy, it would never get completed.
The proper use for these mega-drives is for daily backups of smaller drives, or in raid5 with your fingers crossed.
First they said we had to do something 'soon', but we didn't let them heavily regulate industry.
Then they said we had to do something 'now', but again we didn't let them heavily regulate industry.
Now they say that we have to do something 'yesterday.' I hope that this new version of global warming doesnt convince us to let them heavily regulate industry.
The central theme has always been to heavily regulate industry. Granting new powers to established power holders.
Those IPCC reports the warmers always mention never discusses geoengineering, because geoengineering would not grant new powers to those same established power holders.
Just about any country can afford to cool the planet via geoengineering, and could have done so decades ago. Lets not talk about that tho.. we should focus on granting new powers to established power holders, just like those power holders have been saying.
You obviously don't know what you are talking about.
Firstly, in rasterization, 4xAA does not mean 4 samples per pixel. Not all pixels are subsampled or supersampled. Nobody does that shit anymore.
Secondly, in raytracing, 4xAA does not mean 4 samples per pixel. Not all pixels are subsampled or supersampled. Nobody does that shit anymore.
Thirdly, what the fuck do current video cards have to do with *anything* about this? This is called RESEARCH. Ever do any?
Fourthly, rasterization began as a software rendering technique.
Finally, you dont have a fucking clue about why raytracing actualy scales *well* and is obviously a superior rendering method for the future. It may not replace rasterization, but something sure as damn well will and if its not raytracing, whatever is next will scale at least as well as raytracing.
Rasterization scales very poorly to scene complexity, which is a far more important metric than scaling poorly to resolution. Resolution has doubled 3 times in 25 years, while scene complexity has doubled more than 20 times under the same time period. The chance that resolution will double again in the next 5 years is just about zero, while the chance that scene complexity will double several times in the same period is just about 100%.
If you are a decent well learned programmer, essentialy an expert in algorithmic complexity, then surely you understand the comparison O(n) vs O(log n) and why you cannot refute it with horseshit.
If you arent a decent programmer, then realize that you are an ignorant blowhard spouting about something that you do not understand (go read some Donald Knuth, spend a few years living it, then open your mouth)
What makes you think that a virus requires root access, or needs to modify system files?
You have drunk some fairly ignorant koolaid, it seems.
..and before you go on that a virus can't do much to harm your system if it doesnt have root..
The modern virus doesn't try to harm your system. Usualy they try to harm other peoples systems, or fill other peoples e-mail boxes and other such stuff, by using your system and network connection. They can do this using programs and services that your regular account has full rights to access and leverage, be it linux, windows, or os/x.
The idea that this security model is somehow preventative is completely ignorant. You get these viruses by being stupid, and they don't need root privlidges for that. The odds are that if you are stupid you are going to give 'em the keys to the kingdom anyways, not that they need it.
Slashdot Mirror
GCC may do most of the stuff MSVC does, give or take, but there is a difference between just doing it and actualy doing it well.
At the heart of the matter is the main codegen. Simply forgetting about all the advanced optimization features for a moment, which will produce better code out of the same source when no such advanced optimizations are possible/allowed?
The evidence swings strongly towards GCC having rather poor codegen in relation to MSVC, and that additionally MSVC has rather poor codegen in relation to ICC.
In terms of compilers, GNU
This is actualy well known to optimization fanatics. If you want the best quality machine code you go with ICC. If you want the best production environment you go with MSVC. If you want the most portable source code you go with GCC. GCC has its good qualities, but generating near optimal machine code isnt one of them... its not even close.
It happens because flash drives write on very large (512KB) "blocks", but they still pretend like they have average sized (4KB) "sectors".
Essentialy what the intel write-combining technology is doing is combining multiple small (4KB) writes into a single block, and letting the old block become fragmented (having a bunch of 4KB holes in it.)
The scenario in the nutshell:
You have a 1MB file and a program which modifies a single 4KB chunk of it. Intels technology marks the original 4KB chunk within its original "block" as erased, and then allocates a new block (using the wear leveling algorithm) to hold the new version of the 4KB chunk and additionally combines it with any other small writer operations that may have recently occured or will recently occur. Up to 128 such 4KB writes can be combined into a single block write.
After this is done many hundreds of thousands of times, however, the drive begins to be in a state where nearly every "block" is only partialy used. The write combiner itself is stuck with whatever the wear leveling algorithm handed it, which is now a partialy used block instead of a fully virgin block. It can no longer combine 128 small 4KB writes together, but maybe only has space to combine 10 of them, or in the worst case scenario.. 1 of them.
I find it suspicious for another reason. Certain "undesirable" nations have recently started space programs, and a debris field could very well shift the situation back to "prohibitively expensive" for them, but not for the Big 4 (United States, Russia, China, and the European Union)
And ffter using it, they quickly learn what it doesnt do...
...which is run most windows applications
By my calculation, 32nm only offers ~2.7 times as many cores over 45nm, all other things being equal (most importantly, equal chip size)
.. which means jumping directly from 4 cores to 4/16 solutions.
So basically 32nm will offer ~10 symmetric i7 cores.
Of course most software is single threaded and that means that just adding symmetric cores doesnt linearly add value, so its rather obvious that assymetric is the near future, targeting specifically the highly parallel market
The only scientists who claim that McIntyre's work has been debunked are the people he has exposed, or their direct underlings. His attention to detail is an embarassement for the likes of Mann, Jones, and Wang, who each obstruct independent validations of their work.
For the record, Sir, both of these men you are talking about are on the IPCC expert review panel. Their work has been published in peer reviewed journals, and unlike some other folks out there, they actually support and promote the independent verification of their work.
They are good enough for the IPCC, but not good enough for you.. is that it?
I'd MOD him up, but he cited Mann as verification for Mann... which is presented on a site which actively sensors content contributions that are contrary to what Mann is pushing.
It is pretty black and white when you make official requests for their data and source code and do not recieve them, so to have to resort to Freedom of Information requests which themselves get ignored until lawyers get involved, and only after many many years might you recieve the data required to validate their work but quite often you are met with the situation that "the data is no longer available." Even large scientific bodies such as the NOAA drag their feet and obfuscate when these requests are made.
This goes right to the peer review process. This stuff is supposed to be validatable, but even years after publication which is supposed to be post-validation, you are fighting to get the data needed to validate. Often a requirement of a Journal (such as Science, or Nature) is also that the data is to be archived and available, but the standard when it comes to Climate researchers who are publishing is that its simply OK that the data is neither archived nor available... that nobody bothered to do any validation at all.
The peer review process is a complete failure in the climate sciences. It appears to truely be a clique of climate scientists blindly signing off on each others work while they rake in the government grants.
I am reminded of how some circuits such as sorting networks are designed with redundancy such that a certain amount of sub-component error is tolerable. They will still sort their inputs even if multiple sub-components get their output wrong.
It seems to me that we have already been designing things in a probabalistic manner.. all that these new guys are doing is allowing the error tolerance to be much more liberal.
So you are comparing the mobile version of opera with the desktop version of firefox then?
OK, he's an idiot.
hmmm
Bill, let me know how that Lyme disease works out for you, K? Not every damn thing spread by mosquitos is combated by keeping current on your shots.
Lyme disease is spread by Tick's.
Moron.
hmmm
No features? ehehe.. thats funny... ever run it?
..and VBA support is *HUGE* in the commercial world.
People think that C and C++ is popular, and if you go by how much commercial software is developed then you would be right. But if you include all corporate in-house development, VBA busts every other language down to "novelty" status. Big corporations have hundreds of in-house VBA developers, each churning out program after program, day in and day out, with no recognition from any of the "most popular language" rankings. The millions of custom VBA programs out there that corporations use are essential assets to their operations that cannot be replaced without considerable expense.
You cannot capture these corporations as office suite customers without having VBA support. It will NEVER, EVER, HAPPEN. VBA is the driving force, not the particular office suite.
"I demand a free slurpee. It is a good day to die!"
If microsoft designed a package manager and then enforced its use, you all would cry about microsofts unfair monopoly advantage over competing products such as Inno Setup.
After a few months, the EU would file a lawsuit against Microsoft seeking punative monetary damages for its clear violation of EU anti-trust statutes.
Your post went on about how it is unacceptable for microsoft to interfere with how 3rd party programs work. If you did not mean that, when why didnt you say what you meant? Could it have been... HATE?
Installing the plugin allows Firefox to work with ClickOnce software. Thats all. It *IS* a compatability shim, and compatability with ClickOnce is a GOOD THING, since ClickOnce offers a much better (safer) alternative to downloading naked binary installers. This is something *EVERY* windows browser should support.
You realize that Microsoft interferes with a lot of 3rd party programs FOR YOUR BENEFIT, right?
I guess not.
They have a LOT of application compatability shims in their OS's so that older applications continue to work on an operating system they werent designed for. Vista is FILLED with this stuff, even going so far as to virtualize disk access completely rerouting read and write operations to new locations.
I didn't hear you complain about it before. I wonder why. Could it be... HATE?
Did you really want a migration back to IE because grandma's shit didnt work under firefox? You should be thanking microsoft for being so fucking magnanimous.
Did you seriously wonder what gives them the right to change the way a new version of their OS behaves?
And as for no increase in value..
You must be a youngster. The value of computers has drastically gone up. There was a time when you couldn't play video *AT ALL* on a desktop computer.
Heres a thought.. If you do not think a new version of the OS provides value and you also do not think that a new version of the software you run provides value, then why the hell did you upgrade either of them?
If you are responsible for this serious misallocation of funding in your company, then let me know what company that is so that I can get you fired for your complete and utter incompetance and mismanagement of the companies funds. Oh I get it.. you had a budget to blow through or risk of having the budget reduced next year...
Depends what you mean by "people."
The corporate world doesnt leave XP because Vista simply doesnt run all the crap they need to run while there remains no incentive to add in another layer of their tangled mess of abstractions. Thats the damned truth.
Microsoft screwed the pooch by not listening to its biggest clients. They did very well when they paid attention to the TOP10 customers, which are corporate giants with tens of thousands and even hundreds of thousands of installations, running thousands of different programs (many custom) with hundreds of driver sets for equipment that simply isnt manufactured anymore (the manufacturers are out of business! Some for over 2 decades!)
Basically, there are no alternatives other than compatability for these giants. If its not compatible, then its not acceptable, because compatability has become MISSION CRITICAL.
Even if these giants had a blank check book in which to "upgrade" everything so that a route towards the "future" existed, it would still take a DECADE to implement such a change simply because of the massiveness of it all. These guys have to run 2000/XP because they have a collection of thousands of custom VB/VBA programs and scripts, all written inhouse, that are MISSION CRITICAL and used by tens of thousands of people, yet will not work well on anything more modern. Linux doesnt offer any route to using all that mission critical code. OSX doesnt offer any route to using all that mission critical code. VISTA nags with UAC because when all this crap was written, the things they do was considered standard practice (even recommended practice by Microsoft themmselves in some cases.)
Now, some people might suggest that they should convert from Microsoft to something like Linux, regardless of the cost.. but that doesnt prevent the problem at all. Eventualy they will get locked into a specific distribution, and then to a specific version of that distribution, simply because something mission critical has to continue to work no matter what.
Its easy to say "just rewrite it" when its some top or bottom layer thing, but when what you are rewriting is within some tangled mess of middleware which not only needs to talk to that old VMS program (yes, VMS), it needs to do so through that DOS client because the only platform that can run the softweare which talks to some 20 year old device happens to require DOS, which is controlled by a hidden terminal sitting under a win16 client using the KERMIT protocal, and THAT client is now controlled by a WIN32 client, which itself also talks to both an oracle and sql database to keep the VMS data and the data used by other clients with just as much bullshit in synch.
You get the fucking picture? When you got one guy in one office using this software.. then maybe, just maybe, you could afford some downtime and the monetary costs while its retooled.. but when ten thousand people from offices around the globe are using this software.. your just fucking nuts if you think anything other than continued compatability with the past is acceptable.
We still run FORTRAN and COBOL code for christ sakes, and thats kinda trivial to convert to another language when compared to the mosterous job of retooling the tangled web of technology found in mega-corporations.
When these mega-corporations do jump to a new layer, some account somewhere is going to be recieving a wire transfer of 10 or 11 digits just for licensing costs.. they don't do these things lightly.. they will only add that layer when some new "killer" application requires it, and it doesnt matter what OS its for because they already know the solution to that problem: more middleware.
Raytracing touches far less memory than a rasterizer, so WTF are you talking about?
Oh I get it, you want to give rasterizers the benefit of multiple independent memory controllers, but not raytracers.
Oh, and then you stick the GPU on a pedestal.. saying how its the only thing with enough processors and memory bandwidth..
umm.. helllo?? Do you know what Intel has been working towards? Have you missed the past 5 years of research papers comming out of Intel about this very subject? Its called Larrabee and they are well on their way to making it a reality. They are doing to research so that they know what the requirements will be.
nVidia has put out more than one press release saying that you do not need Intel to do realtime raytracing, that their GPU's will be able to do it.
If GPU makers are working towards RTRT, and Intel is also working towards RTRT, what does that make you? Just ignorant?
You cannot avoid big-O with horseshit. Your artists take too long to create models? Heres an idea.. design better tools. Thats whats been happening for years and years. Better tools for more complex scenery.
The big-O is comming to shatter your delusions, and NOTHING can stop it.
I realize that my social skills arent exactly stellar here, but I'm sick and tired of people from OTHER FIELDS who think that they have a clue about big-o when really they are completely and obviously ignorant in the eyes of every single expert programmer. Its like non-physicists telling physicists all about their perpetual motion machines, while expecting to be treated with a gentle hand. Big-o is like the laws of thermodynamics. Undeniable.
Rasterization is a corner case, just like bubble sort outperforms mergesort in cases of small N. Intel has also done research in regards to apprxominately what N needs to be where raytracing will begin to blow the doors right off of rasterization. You can go to their research site and actualy read this stuff. Maybe you should.
So at most 2 prior images?
The real problem with backing these suckers up is how long it takes. HotHardware has this thing listed at an average 88MB/sec transfer rate in benchmarks (noting that the firmware may not be final.) They do not have write speed listed but we can be generous and allow write speed to match read speed.
Math is fun. Lets go crunching!
It will take over 6.31 hours to read/write the entire contents of one of these, and another 6.31 hour to verify the copy. Thats 12.62 hours of constant spinning and head moving.
Want a daily backup? Then 52.6% of every single day just backing it up, and a lifespan of probably 6 months.
Maybe you want to back it up over a network connection to some place offsite. Lets say you got a fairly phat pipe, like a T3 which in fantasy ideal land peaks out at 44.736 Mbps. It would take 99 hours to transfer one of these drives over such a pipe. If this were a weekly backup process, 59.2% of the week would be taken up making the copy.
It would take 426 single layer DVD-R's if you wanted to burn it. I'll leave the calculation of how long THAT would take to someone else. I know that if I was making such a copy, it would never get completed.
The proper use for these mega-drives is for daily backups of smaller drives, or in raid5 with your fingers crossed.
First they said we had to do something 'soon', but we didn't let them heavily regulate industry.
Then they said we had to do something 'now', but again we didn't let them heavily regulate industry.
Now they say that we have to do something 'yesterday.' I hope that this new version of global warming doesnt convince us to let them heavily regulate industry.
The central theme has always been to heavily regulate industry. Granting new powers to established power holders.
Those IPCC reports the warmers always mention never discusses geoengineering, because geoengineering would not grant new powers to those same established power holders.
Just about any country can afford to cool the planet via geoengineering, and could have done so decades ago. Lets not talk about that tho.. we should focus on granting new powers to established power holders, just like those power holders have been saying.
You obviously don't know what you are talking about.
Firstly, in rasterization, 4xAA does not mean 4 samples per pixel. Not all pixels are subsampled or supersampled. Nobody does that shit anymore.
Secondly, in raytracing, 4xAA does not mean 4 samples per pixel. Not all pixels are subsampled or supersampled. Nobody does that shit anymore.
Thirdly, what the fuck do current video cards have to do with *anything* about this? This is called RESEARCH. Ever do any?
Fourthly, rasterization began as a software rendering technique.
Finally, you dont have a fucking clue about why raytracing actualy scales *well* and is obviously a superior rendering method for the future. It may not replace rasterization, but something sure as damn well will and if its not raytracing, whatever is next will scale at least as well as raytracing.
Rasterization scales very poorly to scene complexity, which is a far more important metric than scaling poorly to resolution. Resolution has doubled 3 times in 25 years, while scene complexity has doubled more than 20 times under the same time period. The chance that resolution will double again in the next 5 years is just about zero, while the chance that scene complexity will double several times in the same period is just about 100%.
If you are a decent well learned programmer, essentialy an expert in algorithmic complexity, then surely you understand the comparison O(n) vs O(log n) and why you cannot refute it with horseshit.
If you arent a decent programmer, then realize that you are an ignorant blowhard spouting about something that you do not understand (go read some Donald Knuth, spend a few years living it, then open your mouth)
What makes you think that a virus requires root access, or needs to modify system files?
..and before you go on that a virus can't do much to harm your system if it doesnt have root..
You have drunk some fairly ignorant koolaid, it seems.
The modern virus doesn't try to harm your system. Usualy they try to harm other peoples systems, or fill other peoples e-mail boxes and other such stuff, by using your system and network connection. They can do this using programs and services that your regular account has full rights to access and leverage, be it linux, windows, or os/x.
The idea that this security model is somehow preventative is completely ignorant. You get these viruses by being stupid, and they don't need root privlidges for that. The odds are that if you are stupid you are going to give 'em the keys to the kingdom anyways, not that they need it.
Apple already offers SSD's on its notebooks (At least the MacBook Pro)