There is a well known result in microeconomic theory that basically says that assuming there is a R&D level that is socially optimal, private companies spend less than this amount on R&D. Thus it makes sense for the government to spend taxpayer money on R&D, so that the total amount of R&D is at the optimal level.
While I don't know what would have happened to this particular cancer drug had the government not spent money on developing it, the lesson is that in general, if government would stop R&D spending the companies would not replace all of this money with their own and thus the total amount of R&D would decrease.
You clearly have no idea what Capitalism is. What you described is any post-hunter-gatherer society, from Ancient Egypt onwards. Capitalism is a particular arrangement that did not come into existance until 19th century, where the greed of individuals is supposed to be cleverly funnelled into progress for all.
I beg to differ. AFAIK, there is no universal agreed upon definition of capitalism. Some people argue that merely the existence of trade implies capitalism. As you sure can imagine, trade goes as far back as civilization itself.
Ceramic brakes have been used for a pretty long time on racecars. IIRC at least since the early 90'ies. During night races (e.g. Le Mans 24 h) you can see how the brake discs glow red when they brake into the corners.
Re:One good reason at least
on
China Goes Nuclear
·
· Score: 2, Interesting
I didn't say it was going to be another Chernobyl exactly but you are simply BS'ing everyone if you are trying to claim everything was within the parameters of the design and there was no danger.
That's why I'm not saying that. It is well documented that several parameters exceeded their design limitations during the accident. And yes, there was danger and the reactor could have been even more destroyed. But it wasn't danger of blowing skyhigh and then continuing to burn for days, a la Chernobyl, e.g. danger of spreading significant amount of radioactivity into the environment.
In particular you are leaving out the wild card which was a 1000 cubic feet 1000 PSI Hydrogen bubble that formed in the vessel from the breakdown of the superheated water.
No, I'm not leaving it out. While it was a cause of great concern at the time, it was later determined that there was not enough oxygen in the vessel which could have caused the hydrogen explosion (one reason for this is of course that the superheated water didn't simply "break down" as you imply, rather it is a reaction with the Zr cladding where the cladding is oxidized).
It had an explosive potential of 3 tons of TNT which would have been enough to breach the vessel and containment if it had exploded.
See the link above. There wasn't enough Zr in the reactor to produce enough pressure to break the containment building.
Additionally, Westinghouse (the manufacturer) did some calculations where they concluded that the pressure vessel and high pressure system itself would perhaps have been able to contain the estimated 3000-4000 PSI blast pressure from the hypothetical hydrogen explosion.
If so the core might well have done a China Syndrome and melted through the floor of the vessel and containment building.
Yes, it was certainly a very real risk that the core would have melted through the pressure vessel, but how did you imagine it would melt itself out from the containment building? Gravity pulls the core downward while it ought to go sidewards if it is to reach the containment walls.
There was enough water pooled at the bottom of the vessel there was also a significant chance of a steam explosion when the melting core hit it and that could have also breeched the reactor.
Perhaps. But again there's the containment building preventing further catastrophy. OTOH, as the core was partially submerged in that same water, there was little possibility of a sudden big clump of molten core dropping into it as the water constantly cooled the core. And if the water wouldn't have been there in which case the core would have melted more dramatically, well there wouldn't be water there either to cause the steam explosion, now would it?;-)
All in all you seem to be claiming certainty about a situation that was unprecedented and anything but certain.
I'm claiming that TMI couldn't have developed into a Chernobyl. They were radically different designs, so spreading FUD about light-water reactors on the basis of Chernobyl is totally ridiculous.
Sure, but the reaction is self-sustaining, meaning that without the coolant to moderate the reaction it can run away...go critical, and meltdown.
Uh oh, seems like you need a refresh course in nuclear engineering. Specifically, what a moderator is. And how the fission probability of U235 depends on the energy of the incoming neutron.
A key reason reprocessing has such a stigma attached to it is its historically and still is in some places used to harvest weapons grade plutonium.
If you by reprocessing mean reprocessing of civilian nuclear waste, then no, it's not used for produceing weapons grade Pu. While it is theoretically possible to extract weapons grade Pu from used civilian reactor fuel, it is a very complicated and expensive process. That's why none of the nuclear powers have ever produced weapons grade Pu by this method. There are much simpler and cheaper ways that don't require Pu isotope enrichment.
Re:One good reason at least
on
China Goes Nuclear
·
· Score: 2, Interesting
They just got lucky and stopped the meltdown before they had a reactor breach.
No, they didn't "just get lucky". In a light water reactor like TMI, when the moderator (=water) boils off, the nuclear reactions grind to a halt. This helps prevent Chernobyl style accidents, which happened in part because the RBMK reactor is graphite moderated and has a positive void coefficient, i.e. when the coolant boils off, the reaction rate increases. See the difference?
Coolant boils off, heat output of reactor increases = bad. This can't happen in a light-water reactor.
If they hadn't there would have been a massive radiation release.
Even if the core would have melted through the pressure vessel, the TMI reactor (and thus core) was still within a containment building. Again, as opposed to Chernobyl.
Just because the consequences were mild doesn't change the fact that it could have easily been a major disaster.
Umm, no. Physics makes it impossible for TMI to have become a Chernobyl. TMI is about a worst-case scenario for a light-water reactor. TMI shows that while a light-water reactor accident is a financial disaster for the company owning it, it won't kill thousands of people.
But the point is, why put it in long term storage? We might want the stuff in a hundred or two hundred years.
Yes, that is correct. Currently burial is the best option, since reprocessing technology hasn't advanced as much as it once was hoped, but one day it might. But if we're able to dig it down, we're certainly able to dig it back up again if needed. Certainly it's better to store the waste 500 m underground in storage designed to be permanent, than in cooling ponds, drums and other temporary storage where all kinds of natural or man-made distasters, terrorism etc. can happen comparatively easily.
They are probably two independant cores manufactured seperately, they are just combined in the die. So if any given core fails along the way, you aren't discarding a second perfectly good core. Unless they fail when manufacturing adds them to the die.
No, they are manufactured together. If one core happens to be defective, it is disabled and the package is sold as a 1 core processor. Same as is commonly done with cache today.
IBM also a few years ago sold a "HPC Regatta" p690 system with 1-core power4 processors instead of the normal 2-core ones (i.e. a maximum of 16 cores instead of 32 as on the normal p690). I guess they had defective second cores, instead of IBM designing a special 1-core version of the power4 just for the HPC Regatta.
I read somewhere a few weeks ago that Oracle, MS and some other big software companies with per-cpu licensing intend to count a hyperthreaded cpu as 1 cpu since it's basically virtual processors. Multi-core cpu:s, OTOH, will be regarded as multiple cpu:s, as there are, duh, multiple cores. They just happen to sit on the same die.
Gas mixed with air explodes in a closed chamber and creates pressure. Wouldn't it accelerate your train just a bit, provided that the force generated is greater than the rolling friction?
If you're talking about starting from a standstill, then practically speaking, you'd either break your tranny or bend the crankshaft. Sure, it's possible to make an engine or transmission strong enough, but they'd be big as a house.
The existance of steam locomotives seams to confirm it's possible.
Steam locomotives work slightly differently. They have a continous steam pressure pushing at the pistons in the cylinders. That produces a smooth torque even at zero rpm. Also, if the train is too heavy, the piston won't move but nothing will break. For an internal combustion engine the situation is different, either the axle turns or something breaks.
But if modern train use generator-motor coupling even once they start moving and get some RPM, either all engineers are incredibly stupid, or that is really more efficient than mechanical transmission.
Yes, modern diesel locomotives have an electrical drivetrain all the way, no mechanical coupling. See the answer psetzer gave for more details.
Alternator efficiently is generally around 70%. Electric motor efficiency is about the same. Making for a 50% (okay, 49%) efficient conversion. Maybe they have really good tech to push that up to 60% or 70%, but thats still a lot of wasted energy.
Perhaps for the typical alternator found on a car engine, which is designed for low cost only. Dedicated alternators easily reach 90 % efficiency, with more advanced models at about 95 % efficiency. Same goes for electrical motors too, so total efficiency is in the 80 - 90 % range.
Why would someone want to put up with that? What possible reason to use such a system? Wires are solid state. If you have a large powertrain, like a railcar driving 4, 8, or even 16 wheels, the power distribution axles, with the attendant gear chains to allow the wheels to spin at different speeds when going around curves in the track, can get to be quite complicated and difficult to maintain.
Yes. A mechanical powertrain for a modern locomotive, with lots of traction wheels, independently turning boggies etc. would be hideously complicated, expensive and heavy. Wouldn't surprise me if all this complexity would reduce the efficiency to levels below the one for the electrical drive, especially considering that one of the big benefits of the electrical drive is that the engine can always turn at the optimum rpm.
Also consider the problem of getting the train going. With a mechanical powertrain you'd need some seriously big clutch capable of slipping for extended periods of time without overheating.
Is that like "Downers and organ meat destined for ALPO cans"?
LOL. Pretty much, I guess.
"Enterprise-Grade" is just a BS term invented by people faced with the prospect of spending their entire careers doing the utterly boring, dredging work of "read stuff from database, show it to the user and write changes back to DB", that constitutes the vast majority of all programming.
Real Programmers, as we all know, write programs to simulate the end of the world as we know it (or perhaps control the systems designed to bring said apocalypse about). And they use FORTRAN.
Have you seen and felt the effects of a big nuclear accident? most of Europe did, they KNOW what they fear, a nuclear accident is not an unlikely theoretical possibility, it has becoem reality in a rather prominent way already.
Ah, you mean the 32 people that so far have died because of the Chernobyl accident? As opposed to the estimated 100000 people in Europe that die prematurely every year due to inhaling fossil fuel exhaust?
Yep, we really can put things into the proper perspective here in Europe, as opposed to those rednecks on the other side of the pond.
Therefore, problems which cannot be translated into 100% parallel algorithms, and therefore do not run efficiently on commodity clusters, are easily tackled on SGI supercomputers.
Umm, no. If a problem can't be parallelized, it won't run any better on that 10000 cpu system than on a 1 cpu system.
Where the SGI wins over a beowulf, is for problems that require lots of low latency communication between the nodes during the computation. In this case the shared memory architecture (programmed using threads) is a lot faster than message passing between the nodes (MPI, typically). However, for serial algorithms, neither the SGI nor the beowulf will be any faster than a comparable 1 cpu system.
One problem is that overweight people easily get injuries. Running around the woods in the dark is a good recipe to sprain your ankle or some such. Doubly so if you're carrying 20 kilos of lard in addition to your kit.
Another problem is that the fat people slow everyone else down, reducing the effectiveness of their training.
The Finnish army practices lying in the snow, waiting for enemies to take out with one shot of the rifle.
And other armies don't practice ambushes?
Personally, I'd rather lie in the snow than in some muddy jungle infested with all kinds of creepy things.
When I did my military service (in Finland) IMHO the worst weather was when it was around 0 degrees and raining/snowing, because then there was no way to keep dry. OTOH, when the temperature was below -10 there were no problems, as long as you had enough self-preservation instinct to brush the snow off your clothes (so it wouldn't melt and make you wet).
Though one irritating thing about really cold weather was that my nailbands often started to bleed. I guess the mittens we were issued weren't warm enough or something.
Round wafers are easier to manufacture. Keep in mind that the the Si chips are made of is monocrystalline. When growing the wafer it is grown from the center outwards. When it hits the wall of the reactor vessel, it will probably break the crystal structure, and whatever growth that continues after that is not usable as there is a grain boundary. With a circular wafer you hit the edge at the same time.
Also, the wafers used today are what, 300 mm in diameter while the chips are something like 10x10 mm, so there's not much material lost anyway. And the leftovers are simply sent back to the wafer factory to be remanufactured into new wafers, so there's no material lost. Not that it would matter anyway, since Si is among the most abundant materials on earth.
For low end cards I find ATI stuff to be better than Nvidia. I have a radeon 9200, which is fast enough for most stuff, has no fan which sound like a jet engine (and breaks). Nice thing is that it works with open source drivers, no need for that binary crap NVIDIA needs for 3d acceleration.
Of course, if you want a fast card, you have to use binary drivers for ATI too, so I guess ATI and NVIDIA are about equally bad there.
Slashdot Mirror
You have a "mark" written on your forhead.
And you, I am sad to see, have "Marx" written on your forehead.
There is a well known result in microeconomic theory that basically says that assuming there is a R&D level that is socially optimal, private companies spend less than this amount on R&D. Thus it makes sense for the government to spend taxpayer money on R&D, so that the total amount of R&D is at the optimal level.
While I don't know what would have happened to this particular cancer drug had the government not spent money on developing it, the lesson is that in general, if government would stop R&D spending the companies would not replace all of this money with their own and thus the total amount of R&D would decrease.
You clearly have no idea what Capitalism is. What you described is any post-hunter-gatherer society, from Ancient Egypt onwards. Capitalism is a particular arrangement that did not come into existance until 19th century, where the greed of individuals is supposed to be cleverly funnelled into progress for all.
I beg to differ. AFAIK, there is no universal agreed upon definition of capitalism. Some people argue that merely the existence of trade implies capitalism. As you sure can imagine, trade goes as far back as civilization itself.
For more information, see e.g. the wikipedia page on capitalism.
Ceramic brakes have been used for a pretty long time on racecars. IIRC at least since the early 90'ies. During night races (e.g. Le Mans 24 h) you can see how the brake discs glow red when they brake into the corners.
I didn't say it was going to be another Chernobyl exactly but you are simply BS'ing everyone if you are trying to claim everything was within the parameters of the design and there was no danger.
That's why I'm not saying that. It is well documented that several parameters exceeded their design limitations during the accident. And yes, there was danger and the reactor could have been even more destroyed. But it wasn't danger of blowing skyhigh and then continuing to burn for days, a la Chernobyl, e.g. danger of spreading significant amount of radioactivity into the environment.
In particular you are leaving out the wild card which was a 1000 cubic feet 1000 PSI Hydrogen bubble that formed in the vessel from the breakdown of the superheated water.
No, I'm not leaving it out. While it was a cause of great concern at the time, it was later determined that there was not enough oxygen in the vessel which could have caused the hydrogen explosion (one reason for this is of course that the superheated water didn't simply "break down" as you imply, rather it is a reaction with the Zr cladding where the cladding is oxidized).
For more information see e.g. this report sumamry.
It had an explosive potential of 3 tons of TNT which would have been enough to breach the vessel and containment if it had exploded.
See the link above. There wasn't enough Zr in the reactor to produce enough pressure to break the containment building.
Additionally, Westinghouse (the manufacturer) did some calculations where they concluded that the pressure vessel and high pressure system itself would perhaps have been able to contain the estimated 3000-4000 PSI blast pressure from the hypothetical hydrogen explosion.
If so the core might well have done a China Syndrome and melted through the floor of the vessel and containment building.
Yes, it was certainly a very real risk that the core would have melted through the pressure vessel, but how did you imagine it would melt itself out from the containment building? Gravity pulls the core downward while it ought to go sidewards if it is to reach the containment walls.
There was enough water pooled at the bottom of the vessel there was also a significant chance of a steam explosion when the melting core hit it and that could have also breeched the reactor.
Perhaps. But again there's the containment building preventing further catastrophy. OTOH, as the core was partially submerged in that same water, there was little possibility of a sudden big clump of molten core dropping into it as the water constantly cooled the core. And if the water wouldn't have been there in which case the core would have melted more dramatically, well there wouldn't be water there either to cause the steam explosion, now would it?
All in all you seem to be claiming certainty about a situation that was unprecedented and anything but certain.
I'm claiming that TMI couldn't have developed into a Chernobyl. They were radically different designs, so spreading FUD about light-water reactors on the basis of Chernobyl is totally ridiculous.
Sure, but the reaction is self-sustaining, meaning that without the coolant to moderate the reaction it can run away...go critical, and meltdown.
Uh oh, seems like you need a refresh course in nuclear engineering. Specifically, what a moderator is. And how the fission probability of U235 depends on the energy of the incoming neutron.
A key reason reprocessing has such a stigma attached to it is its historically and still is in some places used to harvest weapons grade plutonium.
If you by reprocessing mean reprocessing of civilian nuclear waste, then no, it's not used for produceing weapons grade Pu. While it is theoretically possible to extract weapons grade Pu from used civilian reactor fuel, it is a very complicated and expensive process. That's why none of the nuclear powers have ever produced weapons grade Pu by this method. There are much simpler and cheaper ways that don't require Pu isotope enrichment.
They just got lucky and stopped the meltdown before they had a reactor breach.
No, they didn't "just get lucky". In a light water reactor like TMI, when the moderator (=water) boils off, the nuclear reactions grind to a halt. This helps prevent Chernobyl style accidents, which happened in part because the RBMK reactor is graphite moderated and has a positive void coefficient, i.e. when the coolant boils off, the reaction rate increases. See the difference?
Coolant boils off, heat output of reactor increases = bad. This can't happen in a light-water reactor.
If they hadn't there would have been a massive radiation release.
Even if the core would have melted through the pressure vessel, the TMI reactor (and thus core) was still within a containment building. Again, as opposed to Chernobyl.
Just because the consequences were mild doesn't change the fact that it could have easily been a major disaster.
Umm, no. Physics makes it impossible for TMI to have become a Chernobyl. TMI is about a worst-case scenario for a light-water reactor. TMI shows that while a light-water reactor accident is a financial disaster for the company owning it, it won't kill thousands of people.
If you leave enriched Uranium/Plutonium in a pile don't you run the risk of a critical mass.
The thing is, spent reactor fuel isn't enriched enough for this to happen. Especially in absence of a moderator.
But the point is, why put it in long term storage? We might want the stuff in a hundred or two hundred years.
Yes, that is correct. Currently burial is the best option, since reprocessing technology hasn't advanced as much as it once was hoped, but one day it might. But if we're able to dig it down, we're certainly able to dig it back up again if needed. Certainly it's better to store the waste 500 m underground in storage designed to be permanent, than in cooling ponds, drums and other temporary storage where all kinds of natural or man-made distasters, terrorism etc. can happen comparatively easily.
They are probably two independant cores manufactured seperately, they are just combined in the die. So if any given core fails along the way, you aren't discarding a second perfectly good core. Unless they fail when manufacturing adds them to the die.
No, they are manufactured together. If one core happens to be defective, it is disabled and the package is sold as a 1 core processor. Same as is commonly done with cache today.
IBM also a few years ago sold a "HPC Regatta" p690 system with 1-core power4 processors instead of the normal 2-core ones (i.e. a maximum of 16 cores instead of 32 as on the normal p690). I guess they had defective second cores, instead of IBM designing a special 1-core version of the power4 just for the HPC Regatta.
I read somewhere a few weeks ago that Oracle, MS and some other big software companies with per-cpu licensing intend to count a hyperthreaded cpu as 1 cpu since it's basically virtual processors. Multi-core cpu:s, OTOH, will be regarded as multiple cpu:s, as there are, duh, multiple cores. They just happen to sit on the same die.
Gas mixed with air explodes in a closed chamber and creates pressure. Wouldn't it accelerate your train just a bit, provided that the force generated is greater than the rolling friction?
If you're talking about starting from a standstill, then practically speaking, you'd either break your tranny or bend the crankshaft. Sure, it's possible to make an engine or transmission strong enough, but they'd be big as a house.
The existance of steam locomotives seams to confirm it's possible.
Steam locomotives work slightly differently. They have a continous steam pressure pushing at the pistons in the cylinders. That produces a smooth torque even at zero rpm. Also, if the train is too heavy, the piston won't move but nothing will break. For an internal combustion engine the situation is different, either the axle turns or something breaks.
But if modern train use generator-motor coupling even once they start moving and get some RPM, either all engineers are incredibly stupid, or that is really more efficient than mechanical transmission.
Yes, modern diesel locomotives have an electrical drivetrain all the way, no mechanical coupling. See the answer psetzer gave for more details.
Alternator efficiently is generally around 70%. Electric motor efficiency is about the same. Making for a 50% (okay, 49%) efficient conversion. Maybe they have really good tech to push that up to 60% or 70%, but thats still a lot of wasted energy.
Perhaps for the typical alternator found on a car engine, which is designed for low cost only. Dedicated alternators easily reach 90 % efficiency, with more advanced models at about 95 % efficiency. Same goes for electrical motors too, so total efficiency is in the 80 - 90 % range.
Why would someone want to put up with that? What possible reason to use such a system? Wires are solid state. If you have a large powertrain, like a railcar driving 4, 8, or even 16 wheels, the power distribution axles, with the attendant gear chains to allow the wheels to spin at different speeds when going around curves in the track, can get to be quite complicated and difficult to maintain.
Yes. A mechanical powertrain for a modern locomotive, with lots of traction wheels, independently turning boggies etc. would be hideously complicated, expensive and heavy. Wouldn't surprise me if all this complexity would reduce the efficiency to levels below the one for the electrical drive, especially considering that one of the big benefits of the electrical drive is that the engine can always turn at the optimum rpm.
Also consider the problem of getting the train going. With a mechanical powertrain you'd need some seriously big clutch capable of slipping for extended periods of time without overheating.
"Enterprise-Grade"
Is that like "Downers and organ meat destined for ALPO cans"?
LOL. Pretty much, I guess.
"Enterprise-Grade" is just a BS term invented by people faced with the prospect of spending their entire careers doing the utterly boring, dredging work of "read stuff from database, show it to the user and write changes back to DB", that constitutes the vast majority of all programming.
Real Programmers, as we all know, write programs to simulate the end of the world as we know it (or perhaps control the systems designed to bring said apocalypse about). And they use FORTRAN.
... at least that's what I've heard. My admittedly limited experience with a few Agilent instruments confirm this. Solid quality.
Have you seen and felt the effects of a big nuclear accident? most of Europe did, they KNOW what they fear, a nuclear accident is not an unlikely theoretical possibility, it has becoem reality in a rather prominent way already.
Ah, you mean the 32 people that so far have died because of the Chernobyl accident? As opposed to the estimated 100000 people in Europe that die prematurely every year due to inhaling fossil fuel exhaust?
Yep, we really can put things into the proper perspective here in Europe, as opposed to those rednecks on the other side of the pond.
As a minor nitpick, the Red Storm isn't a single system image, it's more like a cluster with a really quick proprietary interconnect.
Therefore, problems which cannot be translated into 100% parallel algorithms, and therefore do not run efficiently on commodity clusters, are easily tackled on SGI supercomputers.
Umm, no. If a problem can't be parallelized, it won't run any better on that 10000 cpu system than on a 1 cpu system.
Where the SGI wins over a beowulf, is for problems that require lots of low latency communication between the nodes during the computation. In this case the shared memory architecture (programmed using threads) is a lot faster than message passing between the nodes (MPI, typically). However, for serial algorithms, neither the SGI nor the beowulf will be any faster than a comparable 1 cpu system.
One problem is that overweight people easily get injuries. Running around the woods in the dark is a good recipe to sprain your ankle or some such. Doubly so if you're carrying 20 kilos of lard in addition to your kit.
Another problem is that the fat people slow everyone else down, reducing the effectiveness of their training.
The Finnish army practices lying in the snow, waiting for enemies to take out with one shot of the rifle.
And other armies don't practice ambushes?
Personally, I'd rather lie in the snow than in some muddy jungle infested with all kinds of creepy things.
When I did my military service (in Finland) IMHO the worst weather was when it was around 0 degrees and raining/snowing, because then there was no way to keep dry. OTOH, when the temperature was below -10 there were no problems, as long as you had enough self-preservation instinct to brush the snow off your clothes (so it wouldn't melt and make you wet).
Though one irritating thing about really cold weather was that my nailbands often started to bleed. I guess the mittens we were issued weren't warm enough or something.
Round wafers are easier to manufacture. Keep in mind that the the Si chips are made of is monocrystalline. When growing the wafer it is grown from the center outwards. When it hits the wall of the reactor vessel, it will probably break the crystal structure, and whatever growth that continues after that is not usable as there is a grain boundary. With a circular wafer you hit the edge at the same time.
Also, the wafers used today are what, 300 mm in diameter while the chips are something like 10x10 mm, so there's not much material lost anyway. And the leftovers are simply sent back to the wafer factory to be remanufactured into new wafers, so there's no material lost. Not that it would matter anyway, since Si is among the most abundant materials on earth.
Not that Excel has any graphing capability worth shit either, for that matter.
Personally, I use mainly R, and sometimes Grace when I want a nice and simple to use GUI thingy.
Which brings up the point that these turbines are going to have to be variable speed if they're going to accommodate humans' varying oxygen needs.
Yup.. you have to remember to switch on the afterburner (reheat for you British folks) when you need some extra Oomph.
For low end cards I find ATI stuff to be better than Nvidia. I have a radeon 9200, which is fast enough for most stuff, has no fan which sound like a jet engine (and breaks). Nice thing is that it works with open source drivers, no need for that binary crap NVIDIA needs for 3d acceleration.
Of course, if you want a fast card, you have to use binary drivers for ATI too, so I guess ATI and NVIDIA are about equally bad there.