Lars said:
>> IQ isn't perfectly normal because there is a lower limit at 0, but no upper limit. But those aren't even one-in-a-million cases.
More to the point, accident, environment and disesase can all decrease IQ, but rarely if ever enhance it (OK, syphillis will enhance IQ, but only temporarily:-). This cause a far larger skew to the low end than the excess spread to the high end.
Glove d'OJ said:
>>Actually, with increasing sample sizes, the mean and median tend to converge. This holds for most standard distributions including normal.
Bluntly, no. This is true for symmetrical distributions. There are plenty of standard distributions which are asymmetric (Poisson - sp? - can be and log normal always is, e.g.), and it is rarely true for an asymmetric distribution.
30 seconds... You could just store an image of the RAM of a booted Linux system and boot INSTANTLY...!!!
Hmm, I guess to you, POST only means a piece of wood stuck in the ground.
Restoring the system memory to a known state is NOT ENOUGH. The hardware has to be initialized to a known state as well. This can't be done by simply loading memory somewhere. Some devices require resetting (because they have write-only control registers and you can't know what state they are in by reading status), a sequence of commands/register writes with appropriate time delays and often a register needs to be programmed with several distinct commands, in the right order.
Simply having a RAM image just doesn't cut it, you need startup code as well. Also, the devices POSTs (Power-On Self Test) impose delays that the OS writer has no control over.
Your arguments, sir, are total rubbish. There is nothing to stop you purchasing a commercial copy of Kylix and releasing under any license you wish (so long as it confirms with the Borland "No Nonsense" license). You don't want to pay for this privilige? Then live without it. Your rights are in no way infringed, because to be quite frankly you have no right to avail yourself of another's labour without recompense (slavery is still illegal, after all). In this case the only recompense demanded is conformance with the GPL, hardly an onerous or unreasonable demand from someone who doesn't want to pay cash to do otherwise!
As for the "Pacman" argument (proposed by Mr. Mundy, IIRC, definitely NOT Mr. Gates), it has all the force of someone objecting to freely available municipal bicycles - as seen in Amsterdam, for instance - because he can't take them as his own and charge for their use. Basically, Microsoft is saying: "We can't steal it so it isn't fair!". Anyone is free to develop their own equivalent of any GPL'ed opus and inculde it in a proprietory product, they just can't use someone else's work in a way that the author has explicitly said he/she does not want without additional approval from that author. Something that Intellectual Property fans are supposed to approve of...
Actually, I think the LGPL would be quite an improvement over the BSD license if you want to disseminate technology widely. GPL is deliberately "viral", which definitely limits its spread (where the license is observed), while BSD is horribly vulnerable to MS-style "embrace, extend, extinguish" tactics.
A common useage of code is to describe "how to" do certain things. Look at all the reference implementations there are floating around, such as sample device drivers. Their purpose is definitely NOT to do anything in and of themselves, rather it is to show others how to write their own device driver (or their own texture mapper, or CRC generator...). This is most surely "expressive speach". Indeed, such sample code is frequently accompanied by warnings that it may not work and is for reference only.
Is there a risk of the "salaryman" syndrome? Sure, if the slowdown/recession (at the moment most of the non-computing/comms economy is still growing in the US) lasts for five-ten years. Might happen. Wouldn't bet anything I couldn't afford to lose on it, though. Most likely Intel will be back in hiring mode in 1-2 years (maybe less), so there won't be any chance for such a mentality to really set in.
Would such a situation limit innovation? It didn't in Japan. More to the point, in most large US corporations, innovation is limited by management, not available talent. Why? Innovation is by definition risky (in the sense that costs are not quantifiable in advance, not in the "it might kill us" sense). MBAs are systematically taught to reduce risk in their companys. Most other managers are either taught the same or pick it up as they go. Budgets are about the most innovation-lethal process immaginable (if you know how much a process costs, it is almost by definition not innovative). By the way, this isn't a knock against management, it's just an observation of the ways of the world.
Interpolation is an avaraging pixel values. When e.g. a texture in a 3-d model is enlarged (or shrunk) to fit a surface, the pixel values are calculated with interpolation.
Anti-aliasing is any technique that makes edges look sharper to the human eye. Note that this applies to object edges in 3-D models every bit as much as in text.
In practice, if you use interpolation to calculate the pixel values along the representation of a line, only thoes pixels that wholly or partially contain the line will be affected. With AA, pixels that do not contain any part of the line, but are "near" the line, may be drawn in a subtly different colour to fool the eye into seeing a smoother edge.
Take a screen capture of some AA text and blow it up in the gimp so that you can see the pixels, then take a look around the text edges - it's quite enlightening.
P.s., although I could easily give formulas for interpolation (it's simple linear interpolation), I don't offhand know what the calculations for AA are - but you can look them up with Google as easily as me, so its left as an exercise for the reader.
Plus manufacture and disposal of solar cells has a really nasty effect on the environment. Bacteria are much more environmentally friendly.
Have you even thought about the numbers? Photsynthesis is what, 5% efficient? Certainly less than 10%. The powerplant/grid system is at most 30% efficient (probably less than 20%, but let's let that one slide). Producing sugar requires MORE energy than is release when the coal is burnt (lb for lb of cardon). So for every Watt of power that is actually consumed, you need in excess of 30 Watts of solar energy. Probably 50-100 in reality. Yikes, that's a LOT of ground that has to be covered with mirrors. Anyone know the energy density of sunlight?. Remeber, you have to double the requirement to allow for nightime (on avarage, 50% of the time). Those mirrors will degrade and require maintainance/replacement. This is also environmentally damaging. Not as much on an acre-for-acre comparison, but there will be many times the acres required (which will ultimately cover grass or other flora, hence degrading the environment). Then what do you do with all the bacterial sludge produced? I doubt if it can be disposed of in an environmentally friendly manner, as this will release the cardon back into the atmosphere. It will require transportation an disposal which will both require additional energy, upping the entire bill.
I can't prove which approach is more environmentally sound, but there are serious reasons to doubt this one a priori.
Growing a biologically oil-rich bacterium and then using it as a fuel source might well be a much saner approach. Of course, it all comes down to actual experience and measurement, but being "natural" because it uses "bacteria" doesn't in any way automatically make a method "better".
It has been pointed out many times that IP is a compromise, where the state grants certain "ownership" rights in exchange for the eventual release of the knowledge into the public domain. Still you and many others seem to have lost track of the whole point of the exercise, to increase the amount of knowledge in the public domain. It was never intended as a "fair reward" for innovators and inventors, but rather a poor but necessary compromise.
The whole ediface is built on the assumption that without some legal protection, companies and individuals will not innovate. That may have been true two hundred years ago (indeed, some incentive was almost certainly needed to entice craftsmen to abandon their home for an uncertain future in a developing nation - which America was at the time - especially when the European nations, Britain in particular, of course, were actively trying to prevent the knowledge transfer). Today, we have vibrant, competitive markets in most of which the choices are innovate or die. Even if there were absolutely no legal protections, companies would have no choice but to continue innovating.
It is truly bizarre that in an age when there is no need for legal protections to induce the creation of new knowledge, IP is being extended beyond the wildest dreams (nightmares?) of the Founding Fathers of the US of A.
To sum up, I disagree. For all but the owners (which is the overwhelming majority), IP is a bad thing.
Never mind that US aid agencies send uncountable tons of food to needy countries, only to have the foodstuffs impounded by the local governments. It's obviously America's fault.
You might want to reconsider this position. Food donations have been found to be one of the worst ways to "help" people in a famin. Indeed, they often worsten and deepen the problem they are supposed to help. (If you find that hard to believe, consider this: Food donations are "free" in the monetary sense. Thus areas receiving them have their food prices artificially lowered, so farmers grow less food, seeing less oportunity for profit and, indeed, sometimes cannot get the startup loans they need becuase of the lowered expected return. This leads directly to another famine after the next harvest.) Famines where there simply isn't enough food in an entire region to feed its people are extremely rare, what's usually missing is the money to make it worthwhile transporting/selling the available stocks. Thus if America really wanted to help famine victims, they would send cash, not food, and would let the free market that they gush over take care of the problem. That, however, wouldn't help US farmers, so the hell with the locals, send them food. It appears to be doing something positive (which it isn't) and gets farmers votes without annoying the city dwellers. Perfect!
Perhaps we should collectively patent the "business model" of patenting something trivial and obvious then making money by suing anyone trying to do honest business using our "patented" technieque. Then each time one of these cases comes up, we sue the plaintif for violation of our patented business model.
Admittedly, Rambus has clear prior art, but that hasn't mattered a tinker's dam since the PO reforms...
You forgot to mention that unit tests are required to be automated and binary. If you aren't using automatic, binary regression testing, you aren't doing XP, period. Also, they must be test all the functionality currently being implemented and include aat least one additional test for each failure discovered during testing. This is one of the few core, required elements of XP.
If the tests are automated, binary (pass or fail, tertium non datur) and of sufficient scope, it takes real talent to keep a project in a constantly broken state.
Applying Occams Razor, there ain't no belivable way this project was doing proper regression testing.
Actually, no. There are situations where racism is not bad for a company - for instance, when there is a plentyful pool of suitable talent to draw from, so that the racist organisation is no worse off for its refusal to make use of certain qualified groups.
Also, sometimes it is conceivably better for a company to be racist/sexist, from a purely economic point of view. The depreciated cost of obtaining the alternative talent could excede the economic benefit. Culturally coherent groups communicate more effectively with each other, so decision making is easier, cheaper and arguably better.
Racism can be quite economically rational, as opposed to merely irrational bigotry. This is, to my mind, not an argument in favour of racism, but rather an argument against unthinking libertarianism - which will happily leave serious social problems alone in the mistaken view that they will simply sort themselves out.
(And yes, in that regard, Linux is better than NT - between chroot and capabilities, you can monkey up most of what you get from Plex86, but not nearly as slick as, say, Inferno.)
???? I'm missing something here - how do chroot and capabilities allow you to run other OSes (and their apps) natively??
Perhaps you were making some other point, but I am honestly missing what it could be. Enlightenment appreciated - thanks.
But then I haven't tried this (I keep a seperate machine for non-essential things like opening Office docs and playing Half-Life on.;-) ), so I could be talkin' out my ass.
Trust me, you are (:-)
First off, the hardware is virtualized, so someone has to find a way for host accesses to play nicely with guest accesses, if you're talking hardware acceleration. The guest OS driver has to co-operate with the host OS driver. Funny thing, this is implemented on ANY guest OS... You'ld think they know better.
Next, Direct3D does NOT default back to software if a hardware acceleration is missing - it simply doesn't support it. (The DirectX philosophy is that speed matters more than completeness, so Direct3D drivers report what they can do and the application is supposed to choose from this menu and not try anything that isn't supported).
Anyway, we're working on it (or at least thinking about working on it - just wait until I have my home firewall configured and the cable run to the DSL...)
As we have a large, captive audience here, I'd like to point out that a major need for the Plex86 project is documentation. Is there anyone out there who can lend a hand? Kevin keeps asking, but no-one has stepped up the the plate, yet. If you can help spiffy up existing documents and fill in what's missing, here's your chance to be a hero!
Maybe closer than you think. I am going to take a crack at "double-heading" video, so that a box with two video cards can dedicate one to the host and the other to the guest. If it pans out (when??? it pans out:-) that will allow native windows DirectX drivers to use the dedicated hardware. Emulation of video hardware, including DirectX, is a hot discussion topic as well. Expect it to be done. (If nothing else, I want to update my DirectX driver knowledge to DirectX8).
Bochs is an x86 emulator, whereas Plex86 is an x86 vitulizer. Plex86 can ONLY run on x86 hardware, so it will never replace Bochs on PPC, for instance.
There is no architectural limit preventing Plex86 from being ported to just about any x86 OS, indeed it is explicitly catered for in the design (via a OS-specific kernel code file with all the needed support functions in it). You may want to think about the implications of running a fundametally insecure OS (like windows) on top of a built-for-security OS like FreeBSD, however.
This may not be a good question to put to Kevin, as he basically feels (with what looks like good reason) that the founders of VmWare basically stole their idea from him, using Bochs as a base. He doesn't actually say so in so many words, but it's pretty clear. So he probably doesn't care and this is a rather special case, anyway.
Hydrogen fuel cells are not a viable solution because it requires more energy to produce the Hydrogen than can be produced from the hydrogen.
While I agree that the anti-nuclear sentiment is excessive and poorly reasoned (and the idea that fussion will be a clean alternative to fission is half-baked at best - most of the energy escapes from the reactor via the free neutron flux, after all), I don't think you're really thinking all that clearly. Of course Hydrogen fuel cells produce less energy than it costs to produce the hydrogen (third? law of thermodynamics) So what? They can be used as a cheap storage system to suck up excess electricity during low demand periods and release it during peak demands. The biggest problem that electricity suppliers face is the lack of efficient, low-loss, cheap storage that would allow them to run their generators continuously at peak efficiency. Plus, they could be used to capture alternate energy (wind, solar, whatever) and release it when the sun don't shine. The problem with hydrogen fuel cells is basically that there's no good way to store the hydrogen, but that's being worked on.
The day when rooftop solar pannels provide all the electricity you need (with fuel cells as the buffer for night-time and cloudy days) may not be all that far off. But then again, it might...
Don't be too sure of that. I've heard several studies report similar problems with geography. Then just last night, on "Who wants to be a millionair" (Ok, laugh if you must) a "fastest finger" qustion was arrange the Michiner novels in geographical order, east to west, "Chesapeake (sp?), Texas, Alaska, Hawaii" (Not that order, that's the answer). I wondered just how easy these questions could possibly get, when the results came on. Only 2 of this already select group got the right answer, and this is American geography.
Hey, what about the whole 8086 family? They designed the x86 just after DEC had moved from architecture to VAX. One of the highlights of that was abandoning segmentation for a uniform memory addressing scheme. Despite this along comes the segmented 8086! Intel tried to maintain compatiblity with the 8080 8-bit instruction set (shows very clearly in the register set of the 8086) and failed, leaving an abortion of a design to which they were committed, to our detriment. The competing micros at the time (Motorola 68x00, National Semi 32032 range etc) all had a flat memory architecture, as did just about every mini and mainframe. If IBM hadn't been a major Intel shareholder (because they'd bought in to protect - believe it or not - their RAM supplies), and therefore used it for their PC, who knows if Intel would even exist Today.
Then they topped that screw-up with the 80286, bringing us those lovely LDTs and GDTs! Segments became selectors, which were even harder for coders to deal with, just so that they could get more memory range without breaking every piece of code in existence... Didn't get it half-way right until the '386, which were stuck with to this day.
Then, of course, there was that quality crunch they had about a decade ago. There were rumours of Fabs with lines that never produced a single useable chip...
It just goes on and on... We should have a top ten list, we should have a top 100!
Slashdot Mirror
Lars said:
:-). This cause a far larger skew to the low end than the excess spread to the high end.
>> IQ isn't perfectly normal because there is a lower limit at 0, but no upper limit. But those aren't even one-in-a-million cases.
More to the point, accident, environment and disesase can all decrease IQ, but rarely if ever enhance it (OK, syphillis will enhance IQ, but only temporarily
Glove d'OJ said:
>>Actually, with increasing sample sizes, the mean and median tend to converge. This holds for most standard distributions including normal.
Bluntly, no. This is true for symmetrical distributions. There are plenty of standard distributions which are asymmetric (Poisson - sp? - can be and log normal always is, e.g.), and it is rarely true for an asymmetric distribution.
30 seconds... You could just store an image of the RAM of a booted Linux system and boot INSTANTLY...!!!
Hmm, I guess to you, POST only means a piece of wood stuck in the ground.
Restoring the system memory to a known state is NOT ENOUGH. The hardware has to be initialized to a known state as well. This can't be done by simply loading memory somewhere. Some devices require resetting (because they have write-only control registers and you can't know what state they are in by reading status), a sequence of commands/register writes with appropriate time delays and often a register needs to be programmed with several distinct commands, in the right order.
Simply having a RAM image just doesn't cut it, you need startup code as well. Also, the devices POSTs (Power-On Self Test) impose delays that the OS writer has no control over.
Colin
Troll feeding time:
Your arguments, sir, are total rubbish. There is nothing to stop you purchasing a commercial copy of Kylix and releasing under any license you wish (so long as it confirms with the Borland "No Nonsense" license). You don't want to pay for this privilige? Then live without it. Your rights are in no way infringed, because to be quite frankly you have no right to avail yourself of another's labour without recompense (slavery is still illegal, after all). In this case the only recompense demanded is conformance with the GPL, hardly an onerous or unreasonable demand from someone who doesn't want to pay cash to do otherwise!
As for the "Pacman" argument (proposed by Mr. Mundy, IIRC, definitely NOT Mr. Gates), it has all the force of someone objecting to freely available municipal bicycles - as seen in Amsterdam, for instance - because he can't take them as his own and charge for their use. Basically, Microsoft is saying: "We can't steal it so it isn't fair!". Anyone is free to develop their own equivalent of any GPL'ed opus and inculde it in a proprietory product, they just can't use someone else's work in a way that the author has explicitly said he/she does not want without additional approval from that author. Something that Intellectual Property fans are supposed to approve of...
Actually, I think the LGPL would be quite an improvement over the BSD license if you want to disseminate technology widely. GPL is deliberately "viral", which definitely limits its spread (where the license is observed), while BSD is horribly vulnerable to MS-style "embrace, extend, extinguish" tactics.
A common useage of code is to describe "how to" do certain things. Look at all the reference implementations there are floating around, such as sample device drivers. Their purpose is definitely NOT to do anything in and of themselves, rather it is to show others how to write their own device driver (or their own texture mapper, or CRC generator ...). This is most surely "expressive speach". Indeed, such sample code is frequently accompanied by warnings that it may not work and is for reference only.
Is there a risk of the "salaryman" syndrome? Sure, if the slowdown/recession (at the moment most of the non-computing/comms economy is still growing in the US) lasts for five-ten years. Might happen. Wouldn't bet anything I couldn't afford to lose on it, though. Most likely Intel will be back in hiring mode in 1-2 years (maybe less), so there won't be any chance for such a mentality to really set in.
Would such a situation limit innovation? It didn't in Japan. More to the point, in most large US corporations, innovation is limited by management, not available talent. Why? Innovation is by definition risky (in the sense that costs are not quantifiable in advance, not in the "it might kill us" sense). MBAs are systematically taught to reduce risk in their companys. Most other managers are either taught the same or pick it up as they go. Budgets are about the most innovation-lethal process immaginable (if you know how much a process costs, it is almost by definition not innovative). By the way, this isn't a knock against management, it's just an observation of the ways of the world.
Interpolation is an avaraging pixel values. When e.g. a texture in a 3-d model is enlarged (or shrunk) to fit a surface, the pixel values are calculated with interpolation.
Anti-aliasing is any technique that makes edges look sharper to the human eye. Note that this applies to object edges in 3-D models every bit as much as in text.
In practice, if you use interpolation to calculate the pixel values along the representation of a line, only thoes pixels that wholly or partially contain the line will be affected. With AA, pixels that do not contain any part of the line, but are "near" the line, may be drawn in a subtly different colour to fool the eye into seeing a smoother edge.
Take a screen capture of some AA text and blow it up in the gimp so that you can see the pixels, then take a look around the text edges - it's quite enlightening.
P.s., although I could easily give formulas for interpolation (it's simple linear interpolation), I don't offhand know what the calculations for AA are - but you can look them up with Google as easily as me, so its left as an exercise for the reader.
Plus manufacture and disposal of solar cells has a really nasty effect on the environment. Bacteria are much more environmentally friendly.
Have you even thought about the numbers? Photsynthesis is what, 5% efficient? Certainly less than 10%. The powerplant/grid system is at most 30% efficient (probably less than 20%, but let's let that one slide). Producing sugar requires MORE energy than is release when the coal is burnt (lb for lb of cardon). So for every Watt of power that is actually consumed, you need in excess of 30 Watts of solar energy. Probably 50-100 in reality. Yikes, that's a LOT of ground that has to be covered with mirrors. Anyone know the energy density of sunlight?. Remeber, you have to double the requirement to allow for nightime (on avarage, 50% of the time). Those mirrors will degrade and require maintainance/replacement. This is also environmentally damaging. Not as much on an acre-for-acre comparison, but there will be many times the acres required (which will ultimately cover grass or other flora, hence degrading the environment). Then what do you do with all the bacterial sludge produced? I doubt if it can be disposed of in an environmentally friendly manner, as this will release the cardon back into the atmosphere. It will require transportation an disposal which will both require additional energy, upping the entire bill.
I can't prove which approach is more environmentally sound, but there are serious reasons to doubt this one a priori.
Growing a biologically oil-rich bacterium and then using it as a fuel source might well be a much saner approach. Of course, it all comes down to actual experience and measurement, but being "natural" because it uses "bacteria" doesn't in any way automatically make a method "better".
It has been pointed out many times that IP is a compromise, where the state grants certain "ownership" rights in exchange for the eventual release of the knowledge into the public domain. Still you and many others seem to have lost track of the whole point of the exercise, to increase the amount of knowledge in the public domain. It was never intended as a "fair reward" for innovators and inventors, but rather a poor but necessary compromise.
The whole ediface is built on the assumption that without some legal protection, companies and individuals will not innovate. That may have been true two hundred years ago (indeed, some incentive was almost certainly needed to entice craftsmen to abandon their home for an uncertain future in a developing nation - which America was at the time - especially when the European nations, Britain in particular, of course, were actively trying to prevent the knowledge transfer). Today, we have vibrant, competitive markets in most of which the choices are innovate or die. Even if there were absolutely no legal protections, companies would have no choice but to continue innovating.
It is truly bizarre that in an age when there is no need for legal protections to induce the creation of new knowledge, IP is being extended beyond the wildest dreams (nightmares?) of the Founding Fathers of the US of A.
To sum up, I disagree. For all but the owners (which is the overwhelming majority), IP is a bad thing.
Never mind that US aid agencies send uncountable tons of food to needy countries, only to have the foodstuffs impounded by the local governments. It's obviously America's fault.
You might want to reconsider this position. Food donations have been found to be one of the worst ways to "help" people in a famin. Indeed, they often worsten and deepen the problem they are supposed to help. (If you find that hard to believe, consider this: Food donations are "free" in the monetary sense. Thus areas receiving them have their food prices artificially lowered, so farmers grow less food, seeing less oportunity for profit and, indeed, sometimes cannot get the startup loans they need becuase of the lowered expected return. This leads directly to another famine after the next harvest.) Famines where there simply isn't enough food in an entire region to feed its people are extremely rare, what's usually missing is the money to make it worthwhile transporting/selling the available stocks. Thus if America really wanted to help famine victims, they would send cash, not food, and would let the free market that they gush over take care of the problem. That, however, wouldn't help US farmers, so the hell with the locals, send them food. It appears to be doing something positive (which it isn't) and gets farmers votes without annoying the city dwellers. Perfect!
Perhaps we should collectively patent the "business model" of patenting something trivial and obvious then making money by suing anyone trying to do honest business using our "patented" technieque. Then each time one of these cases comes up, we sue the plaintif for violation of our patented business model.
Admittedly, Rambus has clear prior art, but that hasn't mattered a tinker's dam since the PO reforms...
You forgot to mention that unit tests are required to be automated and binary. If you aren't using automatic, binary regression testing, you aren't doing XP, period. Also, they must be test all the functionality currently being implemented and include aat least one additional test for each failure discovered during testing. This is one of the few core, required elements of XP.
If the tests are automated, binary (pass or fail, tertium non datur) and of sufficient scope, it takes real talent to keep a project in a constantly broken state.
Applying Occams Razor, there ain't no belivable way this project was doing proper regression testing.
Actually, no. There are situations where racism is not bad for a company - for instance, when there is a plentyful pool of suitable talent to draw from, so that the racist organisation is no worse off for its refusal to make use of certain qualified groups.
Also, sometimes it is conceivably better for a company to be racist/sexist, from a purely economic point of view. The depreciated cost of obtaining the alternative talent could excede the economic benefit. Culturally coherent groups communicate more effectively with each other, so decision making is easier, cheaper and arguably better.
Racism can be quite economically rational, as opposed to merely irrational bigotry. This is, to my mind, not an argument in favour of racism, but rather an argument against unthinking libertarianism - which will happily leave serious social problems alone in the mistaken view that they will simply sort themselves out.
(And yes, in that regard, Linux is better than NT - between chroot and capabilities, you can monkey up most of what you get from Plex86, but not nearly as slick as, say, Inferno.)
???? I'm missing something here - how do chroot and capabilities allow you to run other OSes (and their apps) natively??
Perhaps you were making some other point, but I am honestly missing what it could be. Enlightenment appreciated - thanks.
But then I haven't tried this (I keep a seperate machine for non-essential things like opening Office docs and playing Half-Life on. ;-) ), so I could be talkin' out my ass.
Trust me, you are (:-)
First off, the hardware is virtualized, so someone has to find a way for host accesses to play nicely with guest accesses, if you're talking hardware acceleration. The guest OS driver has to co-operate with the host OS driver. Funny thing, this is implemented on ANY guest OS... You'ld think they know better.
Next, Direct3D does NOT default back to software if a hardware acceleration is missing - it simply doesn't support it. (The DirectX philosophy is that speed matters more than completeness, so Direct3D drivers report what they can do and the application is supposed to choose from this menu and not try anything that isn't supported).
Anyway, we're working on it (or at least thinking about working on it - just wait until I have my home firewall configured and the cable run to the DSL...)
Hi Kevin,
/. cast of thousands help out?
What are the greatest needs in the Plex86 project? How can the
(Ok Kevin, here's your chance - you'll never get a better one to recruit new helpers!:-)
Hi Everyone,
As we have a large, captive audience here, I'd like to point out that a major need for the Plex86 project is documentation. Is there anyone out there who can lend a hand? Kevin keeps asking, but no-one has stepped up the the plate, yet. If you can help spiffy up existing documents and fill in what's missing, here's your chance to be a hero!
Maybe closer than you think. I am going to take a crack at "double-heading" video, so that a box with two video cards can dedicate one to the host and the other to the guest. If it pans out (when??? it pans out:-) that will allow native windows DirectX drivers to use the dedicated hardware. Emulation of video hardware, including DirectX, is a hot discussion topic as well. Expect it to be done. (If nothing else, I want to update my DirectX driver knowledge to DirectX8).
Bochs is an x86 emulator, whereas Plex86 is an x86 vitulizer. Plex86 can ONLY run on x86 hardware, so it will never replace Bochs on PPC, for instance.
There is no architectural limit preventing Plex86 from being ported to just about any x86 OS, indeed it is explicitly catered for in the design (via a OS-specific kernel code file with all the needed support functions in it). You may want to think about the implications of running a fundametally insecure OS (like windows) on top of a built-for-security OS like FreeBSD, however.
Hey Kevin,
When do you think you can get around to implementing LBA support, so that you don't have to keep on answering this question?
(Serious, though, as I understand it Plex86 already supports partitions, but it isn't very useful and the disk size currently has a 500 MB limit).
This may not be a good question to put to Kevin, as he basically feels (with what looks like good reason) that the founders of VmWare basically stole their idea from him, using Bochs as a base. He doesn't actually say so in so many words, but it's pretty clear. So he probably doesn't care and this is a rather special case, anyway.
Hydrogen fuel cells are not a viable solution because it requires more energy to produce the Hydrogen than can be produced from the hydrogen.
While I agree that the anti-nuclear sentiment is excessive and poorly reasoned (and the idea that fussion will be a clean alternative to fission is half-baked at best - most of the energy escapes from the reactor via the free neutron flux, after all), I don't think you're really thinking all that clearly. Of course Hydrogen fuel cells produce less energy than it costs to produce the hydrogen (third? law of thermodynamics) So what? They can be used as a cheap storage system to suck up excess electricity during low demand periods and release it during peak demands. The biggest problem that electricity suppliers face is the lack of efficient, low-loss, cheap storage that would allow them to run their generators continuously at peak efficiency. Plus, they could be used to capture alternate energy (wind, solar, whatever) and release it when the sun don't shine. The problem with hydrogen fuel cells is basically that there's no good way to store the hydrogen, but that's being worked on.
The day when rooftop solar pannels provide all the electricity you need (with fuel cells as the buffer for night-time and cloudy days) may not be all that far off. But then again, it might...
Don't be too sure of that. I've heard several studies report similar problems with geography. Then just last night, on "Who wants to be a millionair" (Ok, laugh if you must) a "fastest finger" qustion was arrange the Michiner novels in geographical order, east to west, "Chesapeake (sp?), Texas, Alaska, Hawaii" (Not that order, that's the answer). I wondered just how easy these questions could possibly get, when the results came on. Only 2 of this already select group got the right answer, and this is American geography.
Maybe you went to an unusually good high school.
Hey, what about the whole 8086 family? They designed the x86 just after DEC had moved from architecture to VAX. One of the highlights of that was abandoning segmentation for a uniform memory addressing scheme. Despite this along comes the segmented 8086! Intel tried to maintain compatiblity with the 8080 8-bit instruction set (shows very clearly in the register set of the 8086) and failed, leaving an abortion of a design to which they were committed, to our detriment. The competing micros at the time (Motorola 68x00, National Semi 32032 range etc) all had a flat memory architecture, as did just about every mini and mainframe. If IBM hadn't been a major Intel shareholder (because they'd bought in to protect - believe it or not - their RAM supplies), and therefore used it for their PC, who knows if Intel would even exist Today.
Then they topped that screw-up with the 80286, bringing us those lovely LDTs and GDTs! Segments became selectors, which were even harder for coders to deal with, just so that they could get more memory range without breaking every piece of code in existence... Didn't get it half-way right until the '386, which were stuck with to this day.
Then, of course, there was that quality crunch they had about a decade ago. There were rumours of Fabs with lines that never produced a single useable chip...
It just goes on and on... We should have a top ten list, we should have a top 100!