You're not considering the error rate on the fab. His idea is quite similar to how processors are differentiated. Your fab contains one design, slightly overrated for what the process can cope with. Some dies will work, some will fail. In some cases the failures will knock out a whole pipeline, and you can just disable that pipe in the driver and sell it as a lower rated part. In some cases the pipes work, but only at a lower speed, so drop the clock on the card. Again this can be done in software.
All that is required is an automated test that stamps each card with an id for the driver telling it what clock / pipes can be used. I think the overall cost saving of only tooling up one design wins against the cost of wasted die space. Of course it is still a tradeoff and needs the right design - fab match to be cost effective.
Whether or not it is the correct answer does not change the fact that the microprocessor business is not a free market. That is what I was disputing in the original post.
The transputer was the right answer at the wrong time. Pursuing a multithreaded path means more effort in software, even if it is cheaper in hardware costs. In the mid 80's ramping up the clock speed was the "cheapest" way to increase performance. You should read my post more carefully.
While the Crusoe was aimed at the low power market, the architecture could have thrashed Intel in the performance market. But only if Transmeta had the fab technology that Intel did. In that case their failure was precisely because a market with an entrenched encumbent player, and a high cost of entry, is not free.
> Translation: What they created, wasn't really great, not even good enough for competing in a free market.
Complete bollocks, and there are two examples that show that. The first is Transmeta - their product was a better processor. The reason that they couldn't compete is precisely because the microprocessor market is not free. There are two large barriers to entry that aid an entrenched player like Intel and screw over the new entrant.
1. Intertia - x86 is the dominant instruction set, and if you want to use it you need to license Intel's patents 2. Product Costs - Fab plants cost a lot to make, and a lot more to keep up with. If you get someone else to fab your chips then their profit is one of your overheads.
The second example is the Transputer back in the 80's. That was a product that the market really wasn't ready for. A multi-threaded chip twenty years before the market realised that it was the way to go. Sometimes you can be far enough ahead that it doesn't sell - even though it's better.
The Amiga did require very low level knowledge of the hardware. One of the bad things about the system was the way software would break on newer machines because the low-level stuff wasn't completely backwards compatible. Don't get me wrong - commodity is nice. I love being able to buy a graphics card off the shelf and know that it will work with every game I own. Those (in)elegant layers of software do serve their purpose well.
One thing that really killed the Amiga as a games machine was the introduction of 3d games. Those tighly coupled chips were designed to to do 2d very well, and suddenly they were specialised for the wrong market. The lack of a fat graphics mode until far too late really let the PC enter the games market. Then the 3d graphics boards and DirectX really shut the door forever.
Did you not manage to read the entire post? I then explain that while the performance of the Amiga was surpassed long ago, only recently has the elegance of the architecture been overtaken.
The real problem was the lack of progress. When the Amiga came out in 1984 it blew away *everything*. But IBM was already carving up the business world and so the Amiga never really made inroads. Apart from in Austraila oddly enough - Lassiters used to run on one. As a games machine it was amazing. But then what really changed over the years? The AA chipset was a minor speedbump. The AGA chipset finally made the difference but it was too little too late. Over the two decades that followed Intel blew Motorola out of the water, and the console world went through how many generations?
Now (and only now) we are finally reaching the stage where the Amiga has been replaced. Linux is sufficiently Amiga-like (because they both use the same unix design principles). We've finally reached a stage where custom chipsets are returning - although now mine hangs off an AGP slot rather than being directly soldered onto the motherboard. When the desktop gets offloaded to the graphics card again the commodity PC will finally have caught up with the Amiga in elegance, having eclipsed it in power long ago.
(And yes, the new Nvidia cards are marginally more powerful than a copper program...)
An interesting and thoughtful reply. Certainly a coherent and refreshing change from the other rants. I don't know a great deal about the history of gun control in the UK, so it wouldn't suprise me that firearm ownership was widespread until that recently. I do remember that the biggest changes occured in the '80s. Until then handguns, rifles and shotguns were legal as long as they were licensed. There were certain barriers to getting a gun license but they didn't apply to most of the population.
After a couple of high profile massacres (at the beginning of rolling media coverage) the government moved to restrict gun ownership and as a result (AFAIK) there is no longer any legal gun ownership. This was largely a populist move by a government that needed some easy votes. This is what made the OP seem so strange. From an American viewpoint it may seem quite unnatural to suppress gun ownership - but in the uk it was a popular decision.
You make a good point about the need to compare views. The way the current laws work here is that racist speech is not a crime. But an incitement to commit a crime (eg assault) made on racist grounds is a crime in itself. Obviously this cannot be defined well in law, and so it is largely a matter of interpretation by the courts. The article didn't quote the racist postings, but it did claim that one encouraged people to attack the victims family. I would imagine this is the claim that landed the guy in prison.
The next law that Blair & co tried to slip in would have expanded the offense to include religious speech. But this was largely seen as completely unworkable in any sense, and got dropped. Interestingly I think that the right to protest is much more important than free speech. The examples that you give of widespread social changes are things that popular protest could have brought around. Sadly it is the right to protest that has been wittled away the most in the uk, and infact largely removed in the past couple of years. Without the benefit of decades of hindsight it is always harder to see which side of the debate has the merits, and some protection is needed for the process of finding out.
I didn't make a comparison of how effective different gun control laws are. I pointed out to the OP that not everyone is a pro-gun nut that believes keeping guns illegal is some sort of crime against humanity. You certainly showed me how flawed my sterotypes of gun nuts are though, didn't you?
Do you know what the diagnosis rate for severe mental illnesses is? It's all very well heving NICS but does it catch the nuts that have just popped? Some of the people that shouldn't be allowed weapons don't even have a classifable mental illness, and yet we can still classify them as 'sick'. Much like the recent guy in the Amish community.
As for your comments that all political speech should be protected, what about the current waves of anti-american speech? American is not very popular at the moment while it has troops stationed away from home on its imperial adventures. So how free should speech be - what about those who call for the destruction of America (and mainly the rest of the world). Would you feel alright about recruitment campaigns for Jihad in America? What about those that call for more terrorist strikes on your country and try to rally support for their cause - should they have freedom of speech? What about those handing out information on how to start terrorist cells, targets to hit, and methods to use?
Lastly, America made quite a big deal out of the transfer of nuclear technology from Pakistan to other Muslim nations. Surely that is also, just political speech...
Hello American. As a brit I like living in a country where guns are illegal - and the majority of people here agree with me. I like living in a country where any nutter can't walk into a supermarket and buy a gun. Do you know how many gun massacres we've had since the gun laws came in vs 'the land of the free'. Is that free to die because any loony has a right to weapons?
The laws against inciting racial hatred are controversal, and complex. But why should racist speech be protected? You assume that authorities are stripping freedoms away from people, but you forget that not everybody wants to live in the same kind of land as you do. America is seen as a strange, nutty, violent backwater by the rest of the world. Maybe in time you will come to appreciate the same checks and balances that we have.
Yup, I got that one wrong, see my earlier replies. The non-trivial data-structure is something that I referred to as being hard to deal with on a GPU. I've done some research in GPGPU - specifically doing RSA on a graphics card so I've come across problems that fit the architecture badly.
Oddly, real-time rendering is one application that may be good on a CPU, but it could still be handled more efficiently by custom hardware. The point that I was trying to make is that the custom hardware to do it is not the rasterisation engine on a GPU. Although, that is quite good for compositing windows.... If only to keep the CPU free for other tasks.
When I read it the way that you've put it, it does sound plausible. But the Intel quote was a bit ambiguous - you could read it as 100m rays per image, which I still think is a more natural way of describing it. If you read it the other way as 100m rays per second then it would be a division there, making it about 350 cycles per ray. The actual math could be done that quickly, but it would be very dependent on how cache friendly the data is. Using 3m rays per frame is roughly 3 rays per pixel - beneath the threshold for removing aliasing. Conventional wisdom is about 16 rays per pixel to get nice antialiasing. This of course assumes that those 350 cycles are including subsequent bounces for each ray - not treating each bounce as a separate ray. With a memory latency of ~150 cycles this means that the computation of each ray needs to pipeline exceedingly well...
Erm, yes it is actually. You and the other replies that pointed out that it scales better with complexity are correct. Google confirms that my memory was a bit off on this one...
The nice effect of placing the geometry loop on the outside is that clipping becomes a coherent decision for large groups of pixels. Again this has nice effects in both control flow, which can be amortised over many pixels, and the relative depth of pixels. If you ignore intersecting geometry then you can optimise even more pixels out of the calculation.
The problem with raytracing researchers is that they are incredibly myopic. *Everybody* should use raytracing for *everything* because it is superior to raster in *every case*. Well, bullshit. Take a look at the raytracing results people have posted links to, and then watch the video of Crysis. The problem is not raytracing, but geometric complexity. Raytracing does not scale nicely with the amount of geometry - mainly because of the shadow rays that have to be scattered from each intersection. The 100mil figure assumes about 100 rays per pixel. Well, you need 64 of them just to get around aliasing, and that doesn't leave many for ambient and shadow bounces.
But the GPU is interesting for raytracing. As it moves closer towards a giant floating point vector machine the motivating application will become raytracing. So at the moment a 7800gtx can push 280Gflops. That is 2800 cycles per ray for a single frame. (BTW Intels figures in the article are bullshit. 100mil rays at 30fps = 3 billion rays per second. Roughly one ray per cycle on averge. They are counting a huge number of rays that have been optimised out of the scene, eg shadows or interpolated from pervious frames using a cache).
The raw horsepower is getting there on the card but at the moment the communication soaks up all of the time. Raytracing is the poster-child problem for parallelisation - assuming that you have random access (readable) global memory. If you need to partition the memory into the compute nodes it begins to get harder. In a GPU building datastructures to hold the information is the bottleneck, and it drops the speed by factors of 100s or 1000s. Nvidia and ATi have given the general-purpose community hints that they will improve performance in reading data-structures so this particular roadblock may disappear. A real scatter operation in the fragment shader would be nice, but you would have to gut the ROPs in order to do it. This may happen anyway as the local-area operations that the ROPs compute could fold into fragment operations. To increase the write bandwidth in the card the retirement logic needs to start retiring 'pages' of pixels anyway, over a much wider bus. Otherwise the number of feasible passes per pixel will always be capped by the speed that the ROPs can retire the data.
So given how hard it would be to *efficiently* raytrace on a GPU - why bother when you can throw so much more raw horsepower at faking it with cheap raster techniques?
Sure, I think you're right that there is a change in quantity of terrorists as you consider different environments. I was genuinely interested in what would turn someone raised in a comfortable western lifestyle into someone that we would see as a 'crazed suicide bomber'. The motivations are so completely different as to be beyond comprehension. Poverty and suffering are key issues that create terrorism, but I feel there must be another reason in the explanation over here.
Both you and the GP are assuming that terrorists are born in the type of environment that you describe. But the people arrested today, and the terrorists who blew up the tube on 7/7 were both born and bred in the UK. I think that your arguement is missing a couple of key factors.
Except these police were confronted by an innocent plumber wearing a teeshirt and an open denim jacket. He wasn't aggressive, he didn't portray any threat, or attempt to run. And so the police shot someone (who clearly wasn't wearing a bomb of any kind) in the head multiple times. You don't need to be a genius to see that they fucked up royally, and it was a paniced overreaction.
Why does that make it more serious? What do you think life expectancy was at the beginning of the industrial revolution - forty perhaps? What is now, eighty? And cancer rates have doubled you say... what a suprise. It's only a suprise that they haven't more than doubled given that cancer strikes older people more often than younger. Almost as if all that soot is corrolated to a relative reduction in cancer...
How much of this is down to Cedega? I've got a dual-boot so that I can play Civ4 on either windows, or in linux. Actually, that is literally why I have the machine... it's a very addictive game. Since the 1.52 patch came out it has been as stable on linux as on windows, and I've stopped rebooting to play it. I wouldn't go as far as to actually call it stable, but then it isn't on windows either. It tends to fall over after running for an hour or so. It sometimes can't reload games if they've gotten too complicated. And when it does decide to crash it can just fall flat with no warning at all.
But this is Firaxis's programming - this is the state of the game under windows. And sadly it has improved a hell of a lot since the shipped version...
As far as installation goes. This was a bitch at first, but most of those forum posts are about how to get the game installed *before* Cedega supported it. Now you just run the installer (selecting XP mode or whichever way round it is) and then hit properties afterwards to set the right windows version. Everything else is automatic.
Not true at all. Most games aren't playable under Cedega because it doesn't cover all of DirectX. The ones that are supported actually run faster than native on windows. Don't ask me why, as it is natural to assume that the translation would slow them down. It could just be that OpenGL/linux is faster than DirectX/windows. I've tried dual-booting to verify this and it does seem to be true for the games that I've tried.
The weirdest and most extreme is SimCity 4. For whatever reason the hardware accelerated rendering is broken and so you're forced to use the software renderer - but this runs *as fast* as the hardware version under windows. I've no idea how badly Maxis managed to mangle that code but it quite a weird result.
Slashdot Mirror
You're not considering the error rate on the fab. His idea is quite similar to how
processors are differentiated. Your fab contains one design, slightly overrated for
what the process can cope with. Some dies will work, some will fail. In some cases
the failures will knock out a whole pipeline, and you can just disable that pipe in
the driver and sell it as a lower rated part. In some cases the pipes work, but only
at a lower speed, so drop the clock on the card. Again this can be done in software.
All that is required is an automated test that stamps each card with an id for the
driver telling it what clock / pipes can be used. I think the overall cost saving of
only tooling up one design wins against the cost of wasted die space. Of course it is
still a tradeoff and needs the right design - fab match to be cost effective.
Whether or not it is the correct answer does not change the fact that the microprocessor business is not a free market. That is what I was disputing in the original post.
The transputer was the right answer at the wrong time. Pursuing a multithreaded path means more effort in software, even if it is cheaper in hardware costs. In the mid 80's ramping up the clock speed was the "cheapest" way to increase performance. You should read my post more carefully.
While the Crusoe was aimed at the low power market, the architecture could have thrashed Intel in the performance market. But only if Transmeta had the fab technology that Intel did. In that case their failure was precisely because a market with an entrenched encumbent player, and a high cost of entry, is not free.
> Translation: What they created, wasn't really great, not even good enough for competing in a free market.
Complete bollocks, and there are two examples that show that. The first is Transmeta - their product was
a better processor. The reason that they couldn't compete is precisely because the microprocessor market
is not free. There are two large barriers to entry that aid an entrenched player like Intel and screw over
the new entrant.
1. Intertia - x86 is the dominant instruction set, and if you want to use it you need to license Intel's patents
2. Product Costs - Fab plants cost a lot to make, and a lot more to keep up with. If you get someone else to
fab your chips then their profit is one of your overheads.
The second example is the Transputer back in the 80's. That was a product that the market really wasn't
ready for. A multi-threaded chip twenty years before the market realised that it was the way to go. Sometimes
you can be far enough ahead that it doesn't sell - even though it's better.
The Amiga did require very low level knowledge of the hardware. One of the bad things about the system was the way software would break on newer machines because the low-level stuff wasn't completely backwards compatible. Don't get me wrong - commodity is nice. I love being able to buy a graphics card off the shelf and know that it will work with every game I own. Those (in)elegant layers of software do serve their purpose well.
One thing that really killed the Amiga as a games machine was the introduction of 3d games. Those tighly coupled chips were designed to to do 2d very well, and suddenly they were specialised for the wrong market. The lack of a fat graphics mode until far too late really let the PC enter the games market. Then the 3d graphics boards and DirectX really shut the door forever.
Did you not manage to read the entire post? I then explain that while the performance of the Amiga was surpassed long ago, only recently has the elegance of the architecture been overtaken.
The real problem was the lack of progress. When the Amiga came out in 1984 it blew away *everything*. But IBM was already carving up the business world and so the Amiga never really made inroads. Apart from in Austraila oddly enough - Lassiters used to run on one. As a games machine it was amazing. But then what really changed over the years? The AA chipset was a minor speedbump. The AGA chipset finally made the difference but it was too little too late. Over the two decades that followed Intel blew Motorola out of the water, and the console world went through how many generations?
Now (and only now) we are finally reaching the stage where the Amiga has been replaced. Linux is sufficiently Amiga-like (because they both use the same unix design principles). We've finally reached a stage where custom chipsets are returning - although now mine hangs off an AGP slot rather than being directly soldered onto the motherboard. When the desktop gets offloaded to the graphics card again the commodity PC will finally have caught up with the Amiga in elegance, having eclipsed it in power long ago.
(And yes, the new Nvidia cards are marginally more powerful than a copper program...)
An interesting and thoughtful reply. Certainly a coherent and refreshing change from the other rants. I don't know a great deal about the history of gun control in the UK, so it wouldn't suprise me that firearm ownership was widespread until that recently. I do remember that the biggest changes occured in the '80s. Until then handguns, rifles and shotguns were legal as long as they were licensed. There were certain barriers to getting a gun license but they didn't apply to most of the population.
After a couple of high profile massacres (at the beginning of rolling media coverage) the government moved to restrict gun ownership and as a result (AFAIK) there is no longer any legal gun ownership. This was largely a populist move by a government that needed some easy votes. This is what made the OP seem so strange. From an American viewpoint it may seem quite unnatural to suppress gun ownership - but in the uk it was a popular decision.
You make a good point about the need to compare views. The way the current laws work here is that racist speech is not a crime. But an incitement to commit a crime (eg assault) made on racist grounds is a crime in itself. Obviously this cannot be defined well in law, and so it is largely a matter of interpretation by the courts. The article didn't quote the racist postings, but it did claim that one encouraged people to attack the victims family. I would imagine this is the claim that landed the guy in prison.
The next law that Blair & co tried to slip in would have expanded the offense to include religious speech. But this was largely seen as completely unworkable in any sense, and got dropped. Interestingly I think that the right to protest is much more important than free speech. The examples that you give of widespread social changes are things that popular protest could have brought around. Sadly it is the right to protest that has been wittled away the most in the uk, and infact largely removed in the past couple of years. Without the benefit of decades of hindsight it is always harder to see which side of the debate has the merits, and some protection is needed for the process of finding out.
I didn't make a comparison of how effective different gun control laws are. I pointed out to the OP that not everyone is a pro-gun nut that believes keeping guns illegal is some sort of crime against humanity. You certainly showed me how flawed my sterotypes of gun nuts are though, didn't you?
Do you know what the diagnosis rate for severe mental illnesses is? It's all very well heving NICS but does it catch the nuts that have just popped? Some of the people that shouldn't be allowed weapons don't even have a classifable mental illness, and yet we can still classify them as 'sick'. Much like the recent guy in the Amish community.
As for your comments that all political speech should be protected, what about the current waves of anti-american speech? American is not very popular at the moment while it has troops stationed away from home on its imperial adventures. So how free should speech be - what about those who call for the destruction of America (and mainly the rest of the world). Would you feel alright about recruitment campaigns for Jihad in America? What about those that call for more terrorist strikes on your country and try to rally support for their cause - should they have freedom of speech? What about those handing out information on how to start terrorist cells, targets to hit, and methods to use?
Lastly, America made quite a big deal out of the transfer of nuclear technology from Pakistan to other Muslim nations. Surely that is also, just political speech...
Dunblain and Hungerford spring to mind. There were a crop of others that prompted the current restrictions on hand guns.
Hello American. As a brit I like living in a country where guns are illegal - and the majority of people here agree with me. I like living in a country where any nutter can't walk into a supermarket and buy a gun. Do you know how many gun massacres we've had since the gun laws came in vs 'the land of the free'. Is that free to die because any loony has a right to weapons?
The laws against inciting racial hatred are controversal, and complex. But why should racist speech be protected? You assume that authorities are stripping freedoms away from people, but you forget that not everybody wants to live in the same kind of land as you do. America is seen as a strange, nutty, violent backwater by the rest of the world. Maybe in time you will come to appreciate the same checks and balances that we have.
> a) a vast majority of buyers are over 35?
Have you ever connected to any online server, anywhere, ever? 95% of people who play games online seem to be about 12...
What loss of privacy? All of these cameras are in public places, so what privacy are they eliminating?
Yup, I got that one wrong, see my earlier replies. The non-trivial data-structure is something that I referred to as being hard to deal with on a GPU. I've done some research in GPGPU - specifically doing RSA on a graphics card so I've come across problems that fit the architecture badly.
.... If only to keep the CPU free for other tasks.
Oddly, real-time rendering is one application that may be good on a CPU, but it could still be handled more efficiently by custom hardware. The point that I was trying to make is that the custom hardware to do it is not the rasterisation engine on a GPU. Although, that is quite good for compositing windows
When I read it the way that you've put it, it does sound plausible. But the Intel quote was a bit ambiguous - you could read it as 100m rays per image, which I still think is a more natural way of describing it. If you read it the other way as 100m rays per second then it would be a division there, making it about 350 cycles per ray. The actual math could be done that quickly, but it would be very dependent on how cache friendly the data is. Using 3m rays per frame is roughly 3 rays per pixel - beneath the threshold for removing aliasing. Conventional wisdom is about 16 rays per pixel to get nice antialiasing. This of course assumes that those 350 cycles are including subsequent bounces for each ray - not treating each bounce as a separate ray. With a memory latency of ~150 cycles this means that the computation of each ray needs to pipeline exceedingly well...
Erm, yes it is actually. You and the other replies that pointed out that it scales better with complexity are correct. Google confirms that my memory was a bit off on this one...
[damn should have used preview] : The *other* nice effect...
The nice effect of placing the geometry loop on the outside is that clipping becomes a coherent decision for large groups of pixels. Again this has nice effects in both control flow, which can be amortised over many pixels, and the relative depth of pixels. If you ignore intersecting geometry then you can optimise even more pixels out of the calculation.
The problem with raytracing researchers is that they are incredibly myopic. *Everybody* should use raytracing for *everything* because it is superior to raster in *every case*. Well, bullshit. Take a look at the raytracing results people have posted links to, and then watch the video of Crysis. The problem is not raytracing, but geometric complexity. Raytracing does not scale nicely with the amount of geometry - mainly because of the shadow rays that have to be scattered from each intersection. The 100mil figure assumes about 100 rays per pixel. Well, you need 64 of them just to get around aliasing, and that doesn't leave many for ambient and shadow bounces.
But the GPU is interesting for raytracing. As it moves closer towards a giant floating point vector machine the motivating application will become raytracing. So at the moment a 7800gtx can push 280Gflops. That is 2800 cycles per ray for a single frame. (BTW Intels figures in the article are bullshit. 100mil rays at 30fps = 3 billion rays per second. Roughly one ray per cycle on averge. They are counting a huge number of rays that have been optimised out of the scene, eg shadows or interpolated from pervious frames using a cache).
The raw horsepower is getting there on the card but at the moment the communication soaks up all of the time. Raytracing is the poster-child problem for parallelisation - assuming that you have random access (readable) global memory. If you need to partition the memory into the compute nodes it begins to get harder. In a GPU building datastructures to hold the information is the bottleneck, and it drops the speed by factors of 100s or 1000s. Nvidia and ATi have given the general-purpose community hints that they will improve performance in reading data-structures so this particular roadblock may disappear. A real scatter operation in the fragment shader would be nice, but you would have to gut the ROPs in order to do it. This may happen anyway as the local-area operations that the ROPs compute could fold into fragment operations. To increase the write bandwidth in the card the retirement logic needs to start retiring 'pages' of pixels anyway, over a much wider bus. Otherwise the number of feasible passes per pixel will always be capped by the speed that the ROPs can retire the data.
So given how hard it would be to *efficiently* raytrace on a GPU - why bother when you can throw so much more raw horsepower at faking it with cheap raster techniques?
So you think that Vista has succumbed to the smallfries Corallary:
Every project that exists long enough eventually needs more lines of code than can be written in two years...
Sure, I think you're right that there is a change in quantity of terrorists as you consider different environments. I was genuinely interested in what would turn someone raised in a comfortable western lifestyle into someone that we would see as a 'crazed suicide bomber'. The motivations are so completely different as to be beyond comprehension. Poverty and suffering are key issues that create terrorism, but I feel there must be another reason in the explanation over here.
Both you and the GP are assuming that terrorists are born in the type of environment that you describe. But the people arrested today, and the terrorists who blew up the tube on 7/7 were both born and bred in the UK. I think that your arguement is missing a couple of key factors.
Except these police were confronted by an innocent plumber wearing a teeshirt and an open denim jacket. He wasn't aggressive, he didn't portray any threat, or attempt to run. And so the police shot someone (who clearly wasn't wearing a bomb of any kind) in the head multiple times. You don't need to be a genius to see that they fucked up royally, and it was a paniced overreaction.
Why does that make it more serious? What do you think life expectancy was at the beginning of the industrial revolution - forty perhaps? What is now, eighty? And cancer rates have doubled you say ... what a suprise. It's only a suprise that they haven't more than doubled given that cancer strikes older people more often than younger. Almost as if all that soot is corrolated to a relative reduction in cancer ...
How much of this is down to Cedega? I've got a dual-boot so that I can play Civ4 on either windows, or in linux. Actually, that is literally why I have the machine... it's a very addictive game. Since the 1.52 patch came out it has been as stable on linux as on windows, and I've stopped rebooting to play it. I wouldn't go as far as to actually call it stable, but then it isn't on windows either. It tends to fall over after running for an hour or so. It sometimes can't reload games if they've gotten too complicated. And when it does decide to crash it can just fall flat with no warning at all.
But this is Firaxis's programming - this is the state of the game under windows. And sadly it has improved a hell of a lot since the shipped version...
As far as installation goes. This was a bitch at first, but most of those forum posts are about how to get the game installed *before* Cedega supported it. Now you just run the installer (selecting XP mode or whichever way round it is) and then hit properties afterwards to set the right windows version. Everything else is automatic.
Not true at all. Most games aren't playable under Cedega because it doesn't cover all of DirectX. The ones that are supported actually run faster than native on windows. Don't ask me why, as it is natural to assume that the translation would slow them down. It could just be that OpenGL/linux is faster than DirectX/windows. I've tried dual-booting to verify this and it does seem to be true for the games that I've tried.
The weirdest and most extreme is SimCity 4. For whatever reason the hardware accelerated rendering is broken and so you're forced to use the software renderer - but this runs *as fast* as the hardware version under windows. I've no idea how badly Maxis managed to mangle that code but it quite a weird result.