I believe that people don't buy consoles for performance/graphics reasons. They buy them because of the simplicity of use, which is manifold:
1) you don't care whether the game will run on your particular setup
2) you don't care how to set up that game (yeah, "average" people don't have time/will/skill to know what driver version they are running, etc)
3) your game is integrated with the (unified) platform online services - so it picks up your account settings, parental settings, friends and whatnot.
PCs are surely faster, but average players these days don't even object to games being run at 30 fps, which is commonplace on the consoles:(
Well, I believe that the will was essential, and not just the fact that you threw in more money. Most progress happened after the war, mind you.
That should be probably attributed to experiences of recent war, which gave people (at least in Soviet Union) really unique and hard-to-reproduce feeling of being able to achieve any goal. Those people then moved on to rebuilding their country, and they carried on that boost of energy. My granddad told me that in 1945-1946 he lived in some ruined basement without running water and electricity, had three part-time jobs at once to make enough money, but it felt great. People were working like crazy, because they were used to work 16h daily during the war - some were brought up like this.
They all shared the spirit probably similar to that of Americans in 1776 or Israelis in 1948 - they felt that they accomplished something really great as a nation, they understood their own role in that accomplishment and they were partly compensated by the participation in organizing their life anew.
To sum it up, that generation was completely different and I don't believe that you can easily create the same conditions by increasing the funding. You need a major stressful, but ultimately successful event in population's life to motivate them like that.
P.S. Of course, I come from Russia where money is traditionally considered a "dirty" and "sinister" thing, and explaining people's behavior in purely financial terms may be subconsciously unacceptable for me (although I'm pretty cynical otherwise), so your mileage may vary.
Lenin said: they will sell us the rope with which we will hang them.
Business should not only care about profits, which implies short-term policy. There should be some more general idea behind the business which will help it to avoid "local extrema".
Freedom is about having choices. Security is about limiting the choices. They are inherently in conflict.
I don't see how you can have both. You can have some trade-off between those, but not both. Law is one of such trade-offs - you aren't free to kill people, but you're safe from being killed yourself.
And as a former Soviet citizen I can testify that living in a police state IS safe, if you agree to follow the rules (e.g. if you limit your freedom...). Whether or not this means that the state itself is safe is a separate question, but living without any personal responsibility for your actions (everything is "programmed" for you for years to come) IS safe.
I don't like security news precisely because they result in such overreactions like yours one.
We should not care about security too much. Security is the opposite of freedom, and by concentrating our efforts on security we may end up with completely locked environment.
It's better to tolerate certain threshold of hijacked/owned computers than to require hardware verify the software.
I don't see how current Windows situation is different from DOS times. "Classical" viruses (in DOS) also required user interaction to be run. What makes virus a virus is just an ability to replicate.
Well, while you are allowed to see other people on the street (naked or not), making photos of them without asking for their permission may be objectionable.
"Stupid things"? I disagree, as a low-level console programmer. Reading things directly to memory (or better, mapping them) is the best choice available for certain tasks and/or conditions (e.g. reading from device with extremely expensive seeks or other overhead per single I/O operation, reading without ability to allocate memory dynamically etc). You should know when not to abstract away from the hardware you're running on... Abstractions never improve performance and sometimes are just plain impossible.
I'm also eager to run my GPGPU code on that, but since it uses single-precision floats only, I'm OK with NVIDIA's decision.
People still think about cards in terms of "traditional" graphics API. DirectX 11 bleh... These days, you can again render pixels your own way, and rasterization and polygon-based graphics API can be completely bypassed... without losing performance!
Sorry for delay, been busy. If you are interested in further discussion, we can continue by e-mail - mine one is easy to decrypt.
Well, making something "work" and making something "work out-of-the box for dumb user" are pretty different things.
Both of which have happened, for every Linux game I've tried in recent memory. Do you have a counterexample?
With games, no - I don't really play Linux games. But I can complain on Linux demos - try downloading not-so-recent ones from Pouet and see for yourself.
I have a recent counter-example with [recent] system software;) NVidia-supplied OpenCL profiler (binary only, shipped with its own copy of Qt 4.6 libraries) crashes at start under my Kubuntu 9.10;) Okay, we were talking about slightly different problems, but see, that's also a compatibility issue and it does not work "out of the box":)
You can do compatibility testing with XP SP3, Vista SP1, Win7 to make sure that game runs on each of those (sometimes you need hacks, too). With Ubuntu, you may make much less assumptions.
How so?
Because user is free to install different kernel? Or whatever else he wishes, even replacing GNOME with a later version? It isn't probably supported by Canonical (I hope so!) so that's why I stated that you should stick with Canonical-supported stuff... or you are going to bundle everything that your game relies on - and that's a problem if you want to integrate it with the rest of system (in more or less Games For Windows way).
Realistic option is to stick to particular Ubuntu release (perhaps an LTS one) and only declare support for things installed from Canonical-supported repositories.
Yeah, which kind of falls under a "duh" heading.
Of course, it's very likely that it will continue to work with future releases -- again, I cite pretty much every game released for Linux, ever.
See above why I mentioned Canonical-supported stuff only. As for "very likely that it will continue to work" it's probably not a good definition of a platform. Microsoft, security issues aside, goes to great lengths to ensure that their updates break minimum amount of software, sometimes even creating separate "hacked" system libraries for specific products. It also has a certification program which you can submit your application to (for money of course) and be sure that you are (and will stay) compatible. Why major distros (like Canonical) don't do these things? (sure, it's all about the money).
weaker backward compatibility between releases (some distros have no well-defined release at all).
Weaker than Win7's "backwards compatibility" called "XP Mode"?
You will have to link statically as much as possible (to minimize damage of incompatible or just newer libraries)
...just like on Windows. Or include the libraries inside your installation directory, just like OS X -- I'd prefer that, actually, as it allows them to be patched individually, or by users -- but again, I cannot remember that ever being needed.
Sure, but there's a clear border what you should bundle with your Windows game and what you shouldn't.
you are still facing problems like significant driver changes that break sound or video for you.
...just like on Windows. And again, a counterexample: Quake 3 still works. In case you've forgotten, Quake 3 was released in 1999.
Well, Quake is a C program which only depends on OpenGL - IIRC it has zip (for packages) and whatever they used for cinematics linked in.
Nowadays, games have grown a lot large... There's plenty of third-party software, like various voice-over-ip libraries, codecs, telemetry, physics, cus
Ubuntu, Fedora, particular LSB release might be a platform, but still, it's too vague for games.
Wait, what? Ubuntu is "too vague for games"?
WTF do you call "XP or newer" then?
And frankly, there seem to be plenty of games which work well across many platforms, even without relying on things like LSB. Games from years ago still work. Just what is it that's so different about desktop Linux distros?
Well, making something "work" and making something "work out-of-the box for dumb user" are pretty different things.
You can do compatibility testing with XP SP3, Vista SP1, Win7 to make sure that game runs on each of those (sometimes you need hacks, too). With Ubuntu, you may make much less assumptions.
Realistic option is to stick to particular Ubuntu release (perhaps an LTS one) and only declare support for things installed from Canonical-supported repositories. The other option is just not to support anything at all and let user figure out things himself/herself.
What's so different about Linux distros? Well, weaker backward compatibility between releases (some distros have no well-defined release at all). You will have to link statically as much as possible (to minimize damage of incompatible or just newer libraries) and you are still facing problems like significant driver changes that break sound or video for you.
Those games that work between distros... well, good for them. But do they work right after unpacking/installing or do you need to find things like older libstdc++ yourself?
with binary NVIDIA drivers - there're no other options currently
The binary ATI drivers don't work?
May be, I haven't really tested (haven't used ATI with Linux for ages). I was making comparison mostly to that new "nouveau" thing which is gaining popularity.
set "realtime" priority for our process
Do you actually need that? Or would it work to simply give you a non-bloated system with an intelligent scheduler? I run plenty of games without realtime that seem to work well, and it's nice to know that if they freeze, my system won't.
Well, this was just an example. Actually, setting realtime priority is discouraged by Microsoft as well, but I wanted to show that developers need much more control over the system than traditional Linux/Unix program is allowed to have.
Not only because of copy protection (though this is also a strong reason), but because of games trying to provide smooth, "console"-like experience on the PC. We want sustainable FPS rate and go to great lengths to fight occasional FPS hitches which usually happen because of resource streaming. Sure, if one just wants to get the game working under Linux, one may lower his/her quality standards, and most Linux games I have tried actually do exactly that...
To begin with, Linux is not a platform. Ubuntu, Fedora, particular LSB release might be a platform, but still, it's too vague for games. There's enough incompatibilities between Windows versions and Windows drivers - actually, that's one of two major reasons why PC does not look specially attractive for large gamedev studios (the other reason is rampant piracy).
I work as a game developer and being a Linux enthusiast, I'd really love to support it natively, but the only realistic option for me is to make sure that the game runs in Wine (with binary NVIDIA drivers - there're no other options currently). Games are (and always were) highly dependent on decent graphics drivers and predictable behavior of system components (yes, we need the ability to know what is in Video RAM and what isn't, use vendor-specific hacks to fine-tune application behavior, set "realtime" priority for our process and do other not-always-pleasant low-level stuff to solve real-world problems and get the game shipped on time), and Linux unfortunately cannot guarantee all that predictability and control over the system (for developers), as of today.
Actually MacOS X prior to 10.6 seems to lack posix_memalign() function in its libc. Well, their ABI mandates 16-byte stack and malloc() alignment, so they got away with that because alignment other than 16 bytes is rarely needed, but still... I wonder how can a system comply with Unix/POSIX specification and lack POSIX-mandated call.
Actually, as of today (that is, Tuesday 09.03.2010), NVIDIA Linux drivers (195.36.03, 195.36.08) support OpenCL better than Windows ones (196.34), that is, they have less bugs. Not sure about Mac, as I don't have access to Mac with OpenCL-capable hardware.
I think they are misusing already public (or is it? aren't twitter posts intellectual property of their authors?) information to aid criminals (perhaps inadvertedly, but still). I believe it is punishable: it's like giving the keys to someone's locked house (even already freely available because of owner's stupidity) to anyone who asks for them (or even worse, encouraging to rob the house at the same time). Sure, the criminals could have found the keys themselves, but by making their lives easier you are increasing the chances of robbery, even if your intent was to warn the potential victim.
I believe that from evolutionary standpoint women should seek not quantity but quality. That is, it makes less sense to raise children from different (and perhaps unsuccessful) males than invest their time in winning the most successful one and then reproduce with him.
And I also believe that the whole concept of forming stable couples is later development, because apes do not exhibit such behavior. The scheme you mention is valid for humans, but I think it is an adaptation of underlying primal behavior to constraints imposed by current civilization (I also wonder how does it look like for polygamic societies).
Women do compete between themselves more than men do. Men are able to organize large and hierarchical structures (e.g. military, church etc) while women usually prefer horizontal relations exactly because they are less willing to subordinate.
Women are just not that keen on taking risks, they prefer long-term stability - that is probably why they are not numerous in risky businesses like being a VC. And I do agree, that's natural: males are nature's way to experiment while female's role is to pick the most successful one among them and reproduce his genes. Somehow it is akin to VC's role in business, though.
By "low quality games" I was referring to games that crash, that look like shit, that contain malicious code etc - everything that would be below publisher's threshold.
Secondly, I'm not sure that it is possible to secure an open platform (i.e. one that you can freely develop for) - I can't think of any example at least. Every OS I used has been successfully exploited.
I agree that a lot of people (especially non-technical) like filling their phones/computers/etc with crap. But most people also run anti-viruses to filter malicious "crapware" - do you want antiviruses for phones, too?
Returning to the start of discussion - given two choices: "reasonably open" platform controlled by its vendor (like iPhone) or free-for-all PC-like environment where you can download software from the web, but you have to use (perhaps commercial) anti-virus, participating in never ending arms race between virus and anti-virus writers, what would you choose?
I'm not sure that would be more popular. Among developers, certainly, but mere users might not like turning their phones into "sort of" PCs with spyware, adware, low quality games and etc.
It would turn phone's software ecosystem from an asphalted city into tropic jungle. Sure, lifeforms are much more diverse in jungles than they are in a city, but I personally prefer living in sterile rooms and not being afraid of parasites.
Slashdot Mirror
I believe that people don't buy consoles for performance/graphics reasons. They buy them because of the simplicity of use, which is manifold:
:(
1) you don't care whether the game will run on your particular setup
2) you don't care how to set up that game (yeah, "average" people don't have time/will/skill to know what driver version they are running, etc)
3) your game is integrated with the (unified) platform online services - so it picks up your account settings, parental settings, friends and whatnot.
PCs are surely faster, but average players these days don't even object to games being run at 30 fps, which is commonplace on the consoles
Well, I believe that the will was essential, and not just the fact that you threw in more money. Most progress happened after the war, mind you.
That should be probably attributed to experiences of recent war, which gave people (at least in Soviet Union) really unique and hard-to-reproduce feeling of being able to achieve any goal. Those people then moved on to rebuilding their country, and they carried on that boost of energy. My granddad told me that in 1945-1946 he lived in some ruined basement without running water and electricity, had three part-time jobs at once to make enough money, but it felt great. People were working like crazy, because they were used to work 16h daily during the war - some were brought up like this.
They all shared the spirit probably similar to that of Americans in 1776 or Israelis in 1948 - they felt that they accomplished something really great as a nation, they understood their own role in that accomplishment and they were partly compensated by the participation in organizing their life anew.
To sum it up, that generation was completely different and I don't believe that you can easily create the same conditions by increasing the funding. You need a major stressful, but ultimately successful event in population's life to motivate them like that.
P.S. Of course, I come from Russia where money is traditionally considered a "dirty" and "sinister" thing, and explaining people's behavior in purely financial terms may be subconsciously unacceptable for me (although I'm pretty cynical otherwise), so your mileage may vary.
Lenin said: they will sell us the rope with which we will hang them.
Business should not only care about profits, which implies short-term policy. There should be some more general idea behind the business which will help it to avoid "local extrema".
Freedom is about having choices. Security is about limiting the choices. They are inherently in conflict.
I don't see how you can have both. You can have some trade-off between those, but not both. Law is one of such trade-offs - you aren't free to kill people, but you're safe from being killed yourself.
And as a former Soviet citizen I can testify that living in a police state IS safe, if you agree to follow the rules (e.g. if you limit your freedom...). Whether or not this means that the state itself is safe is a separate question, but living without any personal responsibility for your actions (everything is "programmed" for you for years to come) IS safe.
I don't like security news precisely because they result in such overreactions like yours one.
We should not care about security too much. Security is the opposite of freedom, and by concentrating our efforts on security we may end up with completely locked environment.
It's better to tolerate certain threshold of hijacked/owned computers than to require hardware verify the software.
Actually it's more like large Hollywood studio coming to nudist beach and starting filming the surroundings...
I don't see how current Windows situation is different from DOS times. "Classical" viruses (in DOS) also required user interaction to be run. What makes virus a virus is just an ability to replicate.
Well, while you are allowed to see other people on the street (naked or not), making photos of them without asking for their permission may be objectionable.
"Stupid things"? I disagree, as a low-level console programmer. Reading things directly to memory (or better, mapping them) is the best choice available for certain tasks and/or conditions (e.g. reading from device with extremely expensive seeks or other overhead per single I/O operation, reading without ability to allocate memory dynamically etc). You should know when not to abstract away from the hardware you're running on... Abstractions never improve performance and sometimes are just plain impossible.
I'm also eager to run my GPGPU code on that, but since it uses single-precision floats only, I'm OK with NVIDIA's decision.
People still think about cards in terms of "traditional" graphics API. DirectX 11 bleh... These days, you can again render pixels your own way, and rasterization and polygon-based graphics API can be completely bypassed... without losing performance!
Well, making something "work" and making something "work out-of-the box for dumb user" are pretty different things.
Both of which have happened, for every Linux game I've tried in recent memory. Do you have a counterexample?
With games, no - I don't really play Linux games. But I can complain on Linux demos - try downloading not-so-recent ones from Pouet and see for yourself. I have a recent counter-example with [recent] system software ;) NVidia-supplied OpenCL profiler (binary only, shipped with its own copy of Qt 4.6 libraries) crashes at start under my Kubuntu 9.10 ;) Okay, we were talking about slightly different problems, but see, that's also a compatibility issue and it does not work "out of the box" :)
You can do compatibility testing with XP SP3, Vista SP1, Win7 to make sure that game runs on each of those (sometimes you need hacks, too). With Ubuntu, you may make much less assumptions.
How so?
Because user is free to install different kernel? Or whatever else he wishes, even replacing GNOME with a later version? It isn't probably supported by Canonical (I hope so!) so that's why I stated that you should stick with Canonical-supported stuff... or you are going to bundle everything that your game relies on - and that's a problem if you want to integrate it with the rest of system (in more or less Games For Windows way).
Realistic option is to stick to particular Ubuntu release (perhaps an LTS one) and only declare support for things installed from Canonical-supported repositories.
Yeah, which kind of falls under a "duh" heading.
Of course, it's very likely that it will continue to work with future releases -- again, I cite pretty much every game released for Linux, ever.
See above why I mentioned Canonical-supported stuff only. As for "very likely that it will continue to work" it's probably not a good definition of a platform. Microsoft, security issues aside, goes to great lengths to ensure that their updates break minimum amount of software, sometimes even creating separate "hacked" system libraries for specific products. It also has a certification program which you can submit your application to (for money of course) and be sure that you are (and will stay) compatible. Why major distros (like Canonical) don't do these things? (sure, it's all about the money).
weaker backward compatibility between releases (some distros have no well-defined release at all).
Weaker than Win7's "backwards compatibility" called "XP Mode"?
You will have to link statically as much as possible (to minimize damage of incompatible or just newer libraries)
...just like on Windows. Or include the libraries inside your installation directory, just like OS X -- I'd prefer that, actually, as it allows them to be patched individually, or by users -- but again, I cannot remember that ever being needed.
Sure, but there's a clear border what you should bundle with your Windows game and what you shouldn't.
you are still facing problems like significant driver changes that break sound or video for you.
...just like on Windows. And again, a counterexample: Quake 3 still works. In case you've forgotten, Quake 3 was released in 1999.
Well, Quake is a C program which only depends on OpenGL - IIRC it has zip (for packages) and whatever they used for cinematics linked in.
Nowadays, games have grown a lot large... There's plenty of third-party software, like various voice-over-ip libraries, codecs, telemetry, physics, cus
Ubuntu, Fedora, particular LSB release might be a platform, but still, it's too vague for games.
Wait, what? Ubuntu is "too vague for games"?
WTF do you call "XP or newer" then?
And frankly, there seem to be plenty of games which work well across many platforms, even without relying on things like LSB. Games from years ago still work. Just what is it that's so different about desktop Linux distros?
Well, making something "work" and making something "work out-of-the box for dumb user" are pretty different things.
You can do compatibility testing with XP SP3, Vista SP1, Win7 to make sure that game runs on each of those (sometimes you need hacks, too). With Ubuntu, you may make much less assumptions.
Realistic option is to stick to particular Ubuntu release (perhaps an LTS one) and only declare support for things installed from Canonical-supported repositories. The other option is just not to support anything at all and let user figure out things himself/herself.
What's so different about Linux distros? Well, weaker backward compatibility between releases (some distros have no well-defined release at all). You will have to link statically as much as possible (to minimize damage of incompatible or just newer libraries) and you are still facing problems like significant driver changes that break sound or video for you.
Those games that work between distros... well, good for them. But do they work right after unpacking/installing or do you need to find things like older libstdc++ yourself?
with binary NVIDIA drivers - there're no other options currently
The binary ATI drivers don't work?
May be, I haven't really tested (haven't used ATI with Linux for ages). I was making comparison mostly to that new "nouveau" thing which is gaining popularity.
set "realtime" priority for our process
Do you actually need that? Or would it work to simply give you a non-bloated system with an intelligent scheduler? I run plenty of games without realtime that seem to work well, and it's nice to know that if they freeze, my system won't.
Well, this was just an example. Actually, setting realtime priority is discouraged by Microsoft as well, but I wanted to show that developers need much more control over the system than traditional Linux/Unix program is allowed to have.
Not only because of copy protection (though this is also a strong reason), but because of games trying to provide smooth, "console"-like experience on the PC. We want sustainable FPS rate and go to great lengths to fight occasional FPS hitches which usually happen because of resource streaming. Sure, if one just wants to get the game working under Linux, one may lower his/her quality standards, and most Linux games I have tried actually do exactly that...
To begin with, Linux is not a platform. Ubuntu, Fedora, particular LSB release might be a platform, but still, it's too vague for games. There's enough incompatibilities between Windows versions and Windows drivers - actually, that's one of two major reasons why PC does not look specially attractive for large gamedev studios (the other reason is rampant piracy).
I work as a game developer and being a Linux enthusiast, I'd really love to support it natively, but the only realistic option for me is to make sure that the game runs in Wine (with binary NVIDIA drivers - there're no other options currently).
Games are (and always were) highly dependent on decent graphics drivers and predictable behavior of system components (yes, we need the ability to know what is in Video RAM and what isn't, use vendor-specific hacks to fine-tune application behavior, set "realtime" priority for our process and do other not-always-pleasant low-level stuff to solve real-world problems and get the game shipped on time), and Linux unfortunately cannot guarantee all that predictability and control over the system (for developers), as of today.
Actually MacOS X prior to 10.6 seems to lack posix_memalign() function in its libc. Well, their ABI mandates 16-byte stack and malloc() alignment, so they got away with that because alignment other than 16 bytes is rarely needed, but still... I wonder how can a system comply with Unix/POSIX specification and lack POSIX-mandated call.
Actually, as of today (that is, Tuesday 09.03.2010), NVIDIA Linux drivers (195.36.03, 195.36.08) support OpenCL better than Windows ones (196.34), that is, they have less bugs. Not sure about Mac, as I don't have access to Mac with OpenCL-capable hardware.
I think they are misusing already public (or is it? aren't twitter posts intellectual property of their authors?) information to aid criminals (perhaps inadvertedly, but still). I believe it is punishable: it's like giving the keys to someone's locked house (even already freely available because of owner's stupidity) to anyone who asks for them (or even worse, encouraging to rob the house at the same time). Sure, the criminals could have found the keys themselves, but by making their lives easier you are increasing the chances of robbery, even if your intent was to warn the potential victim.
When Motorola was a processor brand. And a good one. Ehh...
Agreed. Let the code reveal itself.
Try to single-step it in debugger from the beginning up to main loop.
I believe that from evolutionary standpoint women should seek not quantity but quality. That is, it makes less sense to raise children from different (and perhaps unsuccessful) males than invest their time in winning the most successful one and then reproduce with him.
And I also believe that the whole concept of forming stable couples is later development, because apes do not exhibit such behavior. The scheme you mention is valid for humans, but I think it is an adaptation of underlying primal behavior to constraints imposed by current civilization (I also wonder how does it look like for polygamic societies).
;) Thanks. You're cool, too.
I'm doing my best but sometimes my English is confusing :> Glad that my point was understood :)
Women do compete between themselves more than men do. Men are able to organize large and hierarchical structures (e.g. military, church etc) while women usually prefer horizontal relations exactly because they are less willing to subordinate.
Women are just not that keen on taking risks, they prefer long-term stability - that is probably why they are not numerous in risky businesses like being a VC. And I do agree, that's natural: males are nature's way to experiment while female's role is to pick the most successful one among them and reproduce his genes. Somehow it is akin to VC's role in business, though.
By "low quality games" I was referring to games that crash, that look like shit, that contain malicious code etc - everything that would be below publisher's threshold.
Secondly, I'm not sure that it is possible to secure an open platform (i.e. one that you can freely develop for) - I can't think of any example at least. Every OS I used has been successfully exploited.
I agree that a lot of people (especially non-technical) like filling their phones/computers/etc with crap. But most people also run anti-viruses to filter malicious "crapware" - do you want antiviruses for phones, too?
Returning to the start of discussion - given two choices: "reasonably open" platform controlled by its vendor (like iPhone) or free-for-all PC-like environment where you can download software from the web, but you have to use (perhaps commercial) anti-virus, participating in never ending arms race between virus and anti-virus writers, what would you choose?
I'm not sure that would be more popular. Among developers, certainly, but mere users might not like turning their phones into "sort of" PCs with spyware, adware, low quality games and etc. It would turn phone's software ecosystem from an asphalted city into tropic jungle. Sure, lifeforms are much more diverse in jungles than they are in a city, but I personally prefer living in sterile rooms and not being afraid of parasites.