It's not so much that Square isn't good with making characters seem human, it seems to me more like they aren't really good with the relatively short film format. Square is used to a somewhat interactive approach to plot/character development, with several hours of room to develop a world and exercise certain plot elements and devices that a movie cannot have (Though I would argue that FFX especially is almost as linear as a movie, a shame really...). Some elements of Final Fantasy games include a relatively rich world history to learn, elements of mystery that are hinted at repeatedly but take hours to fully reveal themselves, and, in general, a relatively complex plot to explore over the course of days, rather than hours as is the case of a movie. Of course, the FF movie was FF mostly in name only (no magic, no FF archetypes except Sid (chocobos, 'weapons', etc...), well, except for the graphics quality and the FF7 like view of the planet as living... Square can build beautiful worlds with great detail that can be explored in depth along with a complex plot when they have an audience willing to play for about 40 hours for a game. They are not so good at presenting a canned package that delivers everything in 90 minutes...
Actually, in RAID-4 (maybe 3, don't remember) there is a parity disk. The reason why parity info is distributed in RADI-5 is for performance consideration. Having a parity drive is in no more a single point of failre than distributed parity. So what if you lose the parity disk? Read and write operations wil continue to work (in fact, degraded mode in this circumstance would actually improve write performance, as you now have a RAID-0). Stick in your spare (or switch to a hot-spare), and the arrary reconstructs the parity disk just like any other.
Of course, the securityfocus site needs to actually be thought about before concluding things. In the total bugs chart, Linux does lead. However, those are "aggregate" vulnerabilities among all distirbutions, which means a *lot* of duplicates. By iteself, Windows styill leads..
Of course, GRUB still requires to be installed on the hard drive (otherwise it wouldn't boot), but what is different is that it is divided into two stages, Stage1 is like LILO, except more minimal, and Stage2 does the magic (fancy commandline, VFAT, ext2, reiserfs support) from a separate partition, probably a linux partition for most readers. Because it has intelligent FS drivers, it doesn't need to be re-installed on every kernel compile, and can on the fly load any linux kernel on the disk and boot from it, in addition to other neat stuff. Don't forget that it of course offers nice menus that are likely pre-configured by your distro so the "arcane" syntax can be ignored by many users, lilo-like functionality requires little additional knowledge. And yes, it supports graphics in a sense, can go into a framebuffer mode and have a static, 16 color background.
Interesting path with interesting implications. Especially if students are given a choice between NT and linux on the fly, and come into the college knowing Windows, then they would probably instinctively boot out of "that weird linux thing" and into Windows. Over time, some labs may do away with their linux option entirely. At my college I see something similar. A few years ago, it was almost completely Sun workstations. Then the NT labs started popping up, and at first were massively unpopular as the current students were familiar with the power of Solaris... What did happen was that freshmen. almost exclusively used the NTs because they wouldn't be bothered with anythin non-MS. So next year, Freshmen and Sophomores, and now, basically, not many people use them anymore and stick to NT. Linux, too, has made inroads (so much so that Library loaned laptops can dual-boot into Linux, but I have nbever seen anyone bother), but only in cases where they do not make NT available do I see students using Linux. Unfortunately, Linux is attractive not only for price, but also for easy path to MS apps..
Multithreading applications are a far cry from cluster-capable. In Multi-Threading, you just need to have two seprate computationally indepedent tasks that can be done in parallel, but still have a common memory space (yes, each processor has cache, but the architecture manages it so the cache can be ignored and will do a good job automatically. With Clustered applciations, each node not only has at least one processor, but also has an indepedent memory space (in theory, you *could* pool memory space, but having a network connection as a bottleneck to memory makes things worse rather than better). In addition to having computational indepedence, you must also be able to easily parse out data to the nodes that need it, and ensure consistency between machines is not trashed. The OS doesn't really have a chance at knowing how to divy up the data to the nodes without explicit hints from the application. That is why you need hand-crafted applications to take advantage of clustering. A batch Photoshop process, or time independent movie edits are very conceptually easy to cluster, but the OS has very little knowledge about the data that is being allocated and when it will be accessed and if that memory access could cause inconsistancy if not duplicated across nodes..
Not to nitpick, but the parity computations only impact writes, not reads, on reads RAID-5 is essentially a RAID-0. a RAID-1 of RAID-0 does provide stellar read performance, even compared with RAID-5, and, admittedly, the write performance would also be better (same amount of data being written, no computation or additional reads required, as is the case with RAID 5 where you have to both read and calcualte before the write can even begin). Still, on my 100-base network connection, the system could do a lot and I would never notice...
Of course, it would be cheaper to buy a TV card with video out and a little and a little Y-adapter to go from your soundcard out to RCA than to dish out the 200 dollars for a PS2 devel kit....
Well, the reliability is a point taken, one device fails on an IDE chain and the whole chain may collapse depending on the problem.
The performance is questionable. IDE is behind SCSI, but not nearly as badly anymore. And how are you accessing this storage space typically? Through a network whose speed is likely not to exceed 100 MBit, so network is essentially the bottleneck for throughput, unless of course you have some data processing application or something like oracle, in which case your point may be valid, but still, SCSI is still overpriced for what it gives... If only it had prevailed more, then it would be cheaper and this debate would be moot...
What do you mean "if you don't need redundancy", the only RAID level that doesn't offer redundancy is RAID-0. RAID-5 can tolerate single disk failures, and if you do multiple levels of RAID-5 you can tolerate more failures (depending on how you configure it). The common configuration of RAID-5 with available hot-spares is quite sufficient in all but the most critical configurations, especially if it is a system that is closely monitored. Sure, you can build RAID-1 arrays of N drives where you can tolerate up to N-1 drive failures without problems, but for one space is used a lot less efficiently and for another write performance decreases for every extra level of redundancy you add, but that is overkill for most situations, the chances that multiple drives will fail simultaneously (or within a few hours of each other) is significantly remote compared to single drive failure probability.
And now Mac uses IDE... I use IDE too, and it is pretty clearcut to me that while SCSI drives offer better performance, IDE drive offer better bang for the buck. Recent IDE drives are nearly indistinguishable (to me) from SCSI drives in terms of desktop application performance. Now when you are building a raid array with, say, a high-load database on top of it, then yes, SCSI will be worth it. On the other hand, if it is a single user workstation, I don't really see much of a reason to go to SCSI. In fact, I'm currently building a network file server with raid with no more than 3 or 4 users over a 100-MBit connection concurrently, and I see no reason to use SCSI raid over software RAID on IDE, since SCSI increases the cost dramatically and offers little perceivable benefit...
And as far as this story is concerned, the array I plan to build happens to cost about 1100 dollars and provides 480 gigs of storage (160 more is redundancy, so if redundancy wasn't an issue, it would be 640 gigs, and as we all know, "640 Gigs ought to be enough for anyone"), so a $5k terabyte seems a bit steep when you think about it.
http://homepages.tesco.net/~J.deBoynePollard/FGA/c srss-backspace-bug.html
It is kinda interesting, a long standing bug in CSRSS causes backspaces in high-level output to be handled so badly that it can bring down csrss and the whole system with it.
I think your point is valid, though some of the things he does is a bit different. A lot of these things are exploiting bugs in Outlook Express to show the low level of QA that has gone into the product. What MSN did was simply check the string and block out any non-IE browser, not even letting the browsers *try* to render the page, and when that check was bypassed, it was found there was no functional reason for the check.
Of course, one issue I take with his methods (from what little I can tell from them), is that half of the little tricks he pulls are not highlighting bugs in Outlook Express, but using features of Outlook Express in an annoying way. For example, the set text and bgcolor both to black looks more like using a WebTV specific thing to do something you could do to anyone with HTML mail anyway. Also, the reply-by field to make his messages red is along these lines. Highlighting messages that request immediate attention or should have been replied to by then should be highlighted.
The one thing that he does different from feature abuse is the use of the 'begin' bug. That truly highlights a bug in Outlook and that is a valid point. Just like the bug where you can crash anything from NT to XP (except the 9x series) by creating a weird file and doing a type on it in a cmd window....
But then you have nightmares in support and distribution. Not everyone has a fast, convenient net connection through which they can run an update nightly. Additionally, of those that do, not all *want* to update. Releases allow for a checkpoint to be available in which the user is assured some amount of testing has gone into that exact configuration. When you have a system where you just update packages at will, you end up with strange configurations that few have tested and will likely break. For a seasoned user, this may be workable, but especially for a new person, this would really be disappointing.
Imagine providing support where the number of revisions in common use is comparable to the days that have passed with updates to the codebase.
LVM and Software RAID are beautiful. Unfortunately, the benefit of LVM is completely lost on a great deal of the community. The strategy seems to no longer be to partition nicely, but rather, put all your space in / and don't worry about it. These are the people Mandrake is catering to, and it doesn't seem like it is worth their effort to support a type of configuration that has no benefit to their target users...
Speaking as someone who still occasionally runs DOS/Win3.11 on a 400 MHz machine... damn fast. On My 400 MHz machine, win311/DOS starts up in a negligible amount of time, really cool to do so too..
Heh, back in the days of Doom, you could tell I was a big fan of the chaingun, the 4 key broke off.... I still use that keyboard.. just a bit annoying.
How is RDRAM a benefit without pulling fancy tricks? RDRAM has both crappy latency and narrow bus (16-bit). Sure, it can clock high, but that means nothing when the bus is 4 times as narrow as SDRAM.
Straigh from the article:
"They began developing their own 64-bit extensions to the Pentium line, making sure the code was compatible with AMD's design."
Basicly, for once Intel is trying to make their processors follow a standard defined by another company. My, how the tables have turned. It's really surprising that intel would be this scared, AMD seems very popular among homebrew and budget systems, but in expensive home and business servers, Intel still really outsells AMD.... I guess their Itanium strategy could easily have been blindsided by AMD's better legacy support and they realize that now...
Of course, when the courts told Intel it couldn't copyright a model number, that is how the mess started. With AMD, although they make that model number prominent, they typically have the clock speed not far.... For technical users, it's still there, for non-technical users, they need something more accurate than MHz to indicate performance.
Re:Loki didn't work, but other things might:
on
Last Word on Loki
·
· Score: 2
As I said, the widely varying hardware makes the CD issue rather difficult, but closer to possiblity. As to the space requirements, that is not really the case. My X11R6 (with a *lot* of unneeded cruft) is about 98 megs, and you could cut out so much (most of/usr/X11R6/bin, for example). bin and lib could drastically be cut down (would it need a full blown bash, ls, etc?). You can build small, embedded linux systems now, in the same amount of space a directx install typically takes up on a Windows Game CD, the difficulty is on-the-fly configuration. You say you need to find swap and storage space, making it not play nice with other O/Ses. This simply isn't true (Well, saving does pose a problem, support of removeable media (i.e. floppy, zip) doesn't sound too appealing.
Admittedly, the market of this I could see is small (why bother when I can just install it), but the concept itself is possible.
As to transgaming, you can talk all you want about how they are helping perpetuate Windows-centric gaming, but lets face facts, no matter what linux game players do, it wouldn't be enough of a dent to make any game publishers think twice about what they are doing. We are too few to cut into their bottom line enough to make porting worthwhile (or even avoiding DirectX worthwhile). If we want to play the games, sadly, we are stuck with the quickest and cheapest path to market, which is the Transgaming bastardization of doing things.
Loki didn't work, but other things might:
on
Last Word on Loki
·
· Score: 2
For one, transgaming. Full porting takes too long, while you port, the market buys up the windows version. By 6 months, the market was saturated. With transgaming, they have a chance to get to the market before it *completely* dries up.
Another, Linux-based dedicated game Distros. You pop in the CD and boot, start the game, no install required. Make the PC into kinda a game console. The wide variety of PC hardware makes this not work that well however...
Finally, a really good game maker with exclusive Linux releases. A manufacturer willing to essentially shoot themselves in the foot in the name of Linux popularity. Not happening, but wouldn't it be neat to see, say, Squaresoft Linux exclusive titles?
You could even, in theory add Windows to that list. If you design the application around SDL and OpenGL, it should be minimal work to get it to work under Windows as well as other platforms. One huge problem with the approach you suggest is that the game would have to cater to the lowest common denominator for all platforms. Running a linux layer on top of PS1 would not leave much room for game performance, and thus every version would have to look essentially as crappy as the PS1 version. In any case, for all but the most popular games, one platform is enough and further platforms aren't really worth degrading your flagship platform for. So in essence, it is possible, but I don't think you would have any takers with good games...
Well, judging from my new research, at least the 2D and TV functionality work fine, but 3D support is uncertain.....
The Radeon All-In-Wonder 7500 will probably have full support in short order following its release, so I may just go with that instead. ATI doesn't seem to be too supportive of dri so much, but they do seem to be supporting the GATOS efforts thoroughly. If 8500DV will not have 3D support in the near future, there is no point in me getting it, but the 7500 might fit my purposes....
Slashdot Mirror
It's not so much that Square isn't good with making characters seem human, it seems to me more like they aren't really good with the relatively short film format. Square is used to a somewhat interactive approach to plot/character development, with several hours of room to develop a world and exercise certain plot elements and devices that a movie cannot have (Though I would argue that FFX especially is almost as linear as a movie, a shame really...). Some elements of Final Fantasy games include a relatively rich world history to learn, elements of mystery that are hinted at repeatedly but take hours to fully reveal themselves, and, in general, a relatively complex plot to explore over the course of days, rather than hours as is the case of a movie. Of course, the FF movie was FF mostly in name only (no magic, no FF archetypes except Sid (chocobos, 'weapons', etc...), well, except for the graphics quality and the FF7 like view of the planet as living... Square can build beautiful worlds with great detail that can be explored in depth along with a complex plot when they have an audience willing to play for about 40 hours for a game. They are not so good at presenting a canned package that delivers everything in 90 minutes...
Actually, in RAID-4 (maybe 3, don't remember) there is a parity disk. The reason why parity info is distributed in RADI-5 is for performance consideration. Having a parity drive is in no more a single point of failre than distributed parity. So what if you lose the parity disk? Read and write operations wil continue to work (in fact, degraded mode in this circumstance would actually improve write performance, as you now have a RAID-0). Stick in your spare (or switch to a hot-spare), and the arrary reconstructs the parity disk just like any other.
Of course, the securityfocus site needs to actually be thought about before concluding things. In the total bugs chart, Linux does lead. However, those are "aggregate" vulnerabilities among all distirbutions, which means a *lot* of duplicates. By iteself, Windows styill leads..
Of course, GRUB still requires to be installed on the hard drive (otherwise it wouldn't boot), but what is different is that it is divided into two stages, Stage1 is like LILO, except more minimal, and Stage2 does the magic (fancy commandline, VFAT, ext2, reiserfs support) from a separate partition, probably a linux partition for most readers. Because it has intelligent FS drivers, it doesn't need to be re-installed on every kernel compile, and can on the fly load any linux kernel on the disk and boot from it, in addition to other neat stuff. Don't forget that it of course offers nice menus that are likely pre-configured by your distro so the "arcane" syntax can be ignored by many users, lilo-like functionality requires little additional knowledge. And yes, it supports graphics in a sense, can go into a framebuffer mode and have a static, 16 color background.
Interesting path with interesting implications. Especially if students are given a choice between NT and linux on the fly, and come into the college knowing Windows, then they would probably instinctively boot out of "that weird linux thing" and into Windows. Over time, some labs may do away with their linux option entirely. At my college I see something similar. A few years ago, it was almost completely Sun workstations. Then the NT labs started popping up, and at first were massively unpopular as the current students were familiar with the power of Solaris... What did happen was that freshmen. almost exclusively used the NTs because they wouldn't be bothered with anythin non-MS. So next year, Freshmen and Sophomores, and now, basically, not many people use them anymore and stick to NT. Linux, too, has made inroads (so much so that Library loaned laptops can dual-boot into Linux, but I have nbever seen anyone bother), but only in cases where they do not make NT available do I see students using Linux. Unfortunately, Linux is attractive not only for price, but also for easy path to MS apps..
Multithreading applications are a far cry from cluster-capable. In Multi-Threading, you just need to have two seprate computationally indepedent tasks that can be done in parallel, but still have a common memory space (yes, each processor has cache, but the architecture manages it so the cache can be ignored and will do a good job automatically. With Clustered applciations, each node not only has at least one processor, but also has an indepedent memory space (in theory, you *could* pool memory space, but having a network connection as a bottleneck to memory makes things worse rather than better). In addition to having computational indepedence, you must also be able to easily parse out data to the nodes that need it, and ensure consistency between machines is not trashed. The OS doesn't really have a chance at knowing how to divy up the data to the nodes without explicit hints from the application. That is why you need hand-crafted applications to take advantage of clustering. A batch Photoshop process, or time independent movie edits are very conceptually easy to cluster, but the OS has very little knowledge about the data that is being allocated and when it will be accessed and if that memory access could cause inconsistancy if not duplicated across nodes..
Not to nitpick, but the parity computations only impact writes, not reads, on reads RAID-5 is essentially a RAID-0. a RAID-1 of RAID-0 does provide stellar read performance, even compared with RAID-5, and, admittedly, the write performance would also be better (same amount of data being written, no computation or additional reads required, as is the case with RAID 5 where you have to both read and calcualte before the write can even begin). Still, on my 100-base network connection, the system could do a lot and I would never notice...
Of course, it would be cheaper to buy a TV card with video out and a little and a little Y-adapter to go from your soundcard out to RCA than to dish out the 200 dollars for a PS2 devel kit....
Well, the reliability is a point taken, one device fails on an IDE chain and the whole chain may collapse depending on the problem.
The performance is questionable. IDE is behind SCSI, but not nearly as badly anymore. And how are you accessing this storage space typically? Through a network whose speed is likely not to exceed 100 MBit, so network is essentially the bottleneck for throughput, unless of course you have some data processing application or something like oracle, in which case your point may be valid, but still, SCSI is still overpriced for what it gives... If only it had prevailed more, then it would be cheaper and this debate would be moot...
What do you mean "if you don't need redundancy", the only RAID level that doesn't offer redundancy is RAID-0. RAID-5 can tolerate single disk failures, and if you do multiple levels of RAID-5 you can tolerate more failures (depending on how you configure it). The common configuration of RAID-5 with available hot-spares is quite sufficient in all but the most critical configurations, especially if it is a system that is closely monitored. Sure, you can build RAID-1 arrays of N drives where you can tolerate up to N-1 drive failures without problems, but for one space is used a lot less efficiently and for another write performance decreases for every extra level of redundancy you add, but that is overkill for most situations, the chances that multiple drives will fail simultaneously (or within a few hours of each other) is significantly remote compared to single drive failure probability.
And now Mac uses IDE... I use IDE too, and it is pretty clearcut to me that while SCSI drives offer better performance, IDE drive offer better bang for the buck. Recent IDE drives are nearly indistinguishable (to me) from SCSI drives in terms of desktop application performance. Now when you are building a raid array with, say, a high-load database on top of it, then yes, SCSI will be worth it. On the other hand, if it is a single user workstation, I don't really see much of a reason to go to SCSI. In fact, I'm currently building a network file server with raid with no more than 3 or 4 users over a 100-MBit connection concurrently, and I see no reason to use SCSI raid over software RAID on IDE, since SCSI increases the cost dramatically and offers little perceivable benefit...
And as far as this story is concerned, the array I plan to build happens to cost about 1100 dollars and provides 480 gigs of storage (160 more is redundancy, so if redundancy wasn't an issue, it would be 640 gigs, and as we all know, "640 Gigs ought to be enough for anyone"), so a $5k terabyte seems a bit steep when you think about it.
http://homepages.tesco.net/~J.deBoynePollard/FGA/c srss-backspace-bug.html
It is kinda interesting, a long standing bug in CSRSS causes backspaces in high-level output to be handled so badly that it can bring down csrss and the whole system with it.
I think your point is valid, though some of the things he does is a bit different. A lot of these things are exploiting bugs in Outlook Express to show the low level of QA that has gone into the product. What MSN did was simply check the string and block out any non-IE browser, not even letting the browsers *try* to render the page, and when that check was bypassed, it was found there was no functional reason for the check.
Of course, one issue I take with his methods (from what little I can tell from them), is that half of the little tricks he pulls are not highlighting bugs in Outlook Express, but using features of Outlook Express in an annoying way. For example, the set text and bgcolor both to black looks more like using a WebTV specific thing to do something you could do to anyone with HTML mail anyway. Also, the reply-by field to make his messages red is along these lines. Highlighting messages that request immediate attention or should have been replied to by then should be highlighted.
The one thing that he does different from feature abuse is the use of the 'begin' bug. That truly highlights a bug in Outlook and that is a valid point. Just like the bug where you can crash anything from NT to XP (except the 9x series) by creating a weird file and doing a type on it in a cmd window....
But then you have nightmares in support and distribution. Not everyone has a fast, convenient net connection through which they can run an update nightly. Additionally, of those that do, not all *want* to update. Releases allow for a checkpoint to be available in which the user is assured some amount of testing has gone into that exact configuration. When you have a system where you just update packages at will, you end up with strange configurations that few have tested and will likely break. For a seasoned user, this may be workable, but especially for a new person, this would really be disappointing.
Imagine providing support where the number of revisions in common use is comparable to the days that have passed with updates to the codebase.
LVM and Software RAID are beautiful. Unfortunately, the benefit of LVM is completely lost on a great deal of the community. The strategy seems to no longer be to partition nicely, but rather, put all your space in / and don't worry about it. These are the people Mandrake is catering to, and it doesn't seem like it is worth their effort to support a type of configuration that has no benefit to their target users...
Speaking as someone who still occasionally runs DOS/Win3.11 on a 400 MHz machine... damn fast. On My 400 MHz machine, win311/DOS starts up in a negligible amount of time, really cool to do so too..
Heh, back in the days of Doom, you could tell I was a big fan of the chaingun, the 4 key broke off.... I still use that keyboard.. just a bit annoying.
I mean, my wife gets e-mails telling her to enlarge her penis and I get e-mail telling me to enlarge my breasts....
How is RDRAM a benefit without pulling fancy tricks? RDRAM has both crappy latency and narrow bus (16-bit). Sure, it can clock high, but that means nothing when the bus is 4 times as narrow as SDRAM.
Straigh from the article:
"They began developing their own 64-bit extensions to the Pentium line, making sure the code was compatible with AMD's design."
Basicly, for once Intel is trying to make their processors follow a standard defined by another company. My, how the tables have turned. It's really surprising that intel would be this scared, AMD seems very popular among homebrew and budget systems, but in expensive home and business servers, Intel still really outsells AMD.... I guess their Itanium strategy could easily have been blindsided by AMD's better legacy support and they realize that now...
Of course, when the courts told Intel it couldn't copyright a model number, that is how the mess started. With AMD, although they make that model number prominent, they typically have the clock speed not far.... For technical users, it's still there, for non-technical users, they need something more accurate than MHz to indicate performance.
As I said, the widely varying hardware makes the CD issue rather difficult, but closer to possiblity. As to the space requirements, that is not really the case. My X11R6 (with a *lot* of unneeded cruft) is about 98 megs, and you could cut out so much (most of /usr/X11R6/bin, for example). bin and lib could drastically be cut down (would it need a full blown bash, ls, etc?). You can build small, embedded linux systems now, in the same amount of space a directx install typically takes up on a Windows Game CD, the difficulty is on-the-fly configuration. You say you need to find swap and storage space, making it not play nice with other O/Ses. This simply isn't true (Well, saving does pose a problem, support of removeable media (i.e. floppy, zip) doesn't sound too appealing.
Admittedly, the market of this I could see is small (why bother when I can just install it), but the concept itself is possible.
As to transgaming, you can talk all you want about how they are helping perpetuate Windows-centric gaming, but lets face facts, no matter what linux game players do, it wouldn't be enough of a dent to make any game publishers think twice about what they are doing. We are too few to cut into their bottom line enough to make porting worthwhile (or even avoiding DirectX worthwhile). If we want to play the games, sadly, we are stuck with the quickest and cheapest path to market, which is the Transgaming bastardization of doing things.
For one, transgaming. Full porting takes too long, while you port, the market buys up the windows version. By 6 months, the market was saturated. With transgaming, they have a chance to get to the market before it *completely* dries up.
Another, Linux-based dedicated game Distros. You pop in the CD and boot, start the game, no install required. Make the PC into kinda a game console. The wide variety of PC hardware makes this not work that well however...
Finally, a really good game maker with exclusive Linux releases. A manufacturer willing to essentially shoot themselves in the foot in the name of Linux popularity. Not happening, but wouldn't it be neat to see, say, Squaresoft Linux exclusive titles?
You could even, in theory add Windows to that list. If you design the application around SDL and OpenGL, it should be minimal work to get it to work under Windows as well as other platforms. One huge problem with the approach you suggest is that the game would have to cater to the lowest common denominator for all platforms. Running a linux layer on top of PS1 would not leave much room for game performance, and thus every version would have to look essentially as crappy as the PS1 version. In any case, for all but the most popular games, one platform is enough and further platforms aren't really worth degrading your flagship platform for. So in essence, it is possible, but I don't think you would have any takers with good games...
Well, judging from my new research, at least the 2D and TV functionality work fine, but 3D support is uncertain.....
The Radeon All-In-Wonder 7500 will probably have full support in short order following its release, so I may just go with that instead. ATI doesn't seem to be too supportive of dri so much, but they do seem to be supporting the GATOS efforts thoroughly. If 8500DV will not have 3D support in the near future, there is no point in me getting it, but the 7500 might fit my purposes....