Let me preface all this by saying I'm only talking about console games, my data come from sales figures tallied by NPD, which is a commercial reporting service and I define commercial success as sell-through of over 250,000 units.
90% of all videogames launching new properties are commercial failures. This is well above the overall failure rate of 78%. The sad truth is that most gamers stick with familiar themes and simple concepts. The quirky, unusual, original games you and I like (over 20 of the PS2 games on my shelf sold under 100,000 units) just do not pay the bills.
Therefore it behooves us to think about how to make games that we can enjoy but which will also speak to the mass market. My personal opinion is that the real inventiveness is best served in gameplay mechanics and control. Use the licensed theme as a base to build from. The game market is coming close to maturity. It grew and grew through the 80s and early 90s and now it's near a plateau in terms of userbase. One sign of this is the gamer's average age increasing year on year.
Right now there are plenty of relatively new gamers for whom simple action-shooter, driving and hack-slash gameplay is appealing. They'll quickly grow into more sophisticated play concepts, just as today's hardcore gamers did in years past. The people who figure out how to keep that large pool of gamers interested will be the people who succeed in the industry.
I've played right through the game once and got almost halfway through a second time before growing bored with it. Offhand I'd guess this is about 20 hours of play. My system (bought in April 2001 and used a great deal) is not modded nor do I own a gameshark.
I've seen the loading screen crash once - very early on in fact. When I ejected the disc and put it back in the game recovered perfectly. I've seen the missing tiles problem quite a bit more. The game is almost constantly loading from the DVD and in my opinion makes heavier demands of the PS2 optical drive than any other game - certainly much more than I would ever dare to do in my own code. As such I am unsurprised that it has disc-related issues.
There will be a great many false-positives and false-negatives about these errors. You're right about stupid user syndrome but for each of those people there is at least one person who will assume their PS2 overheated or there was a power spike or they bumped the table.
The one part of this that I don't really understand is why they used a dual-layer DVD. It's a pain in the ass to make and test a dual-layer game and the disc doesn't appear to hold more data than would fir on a single-layer disc. I was also slightly surprised to see the disc doesn't have the 6 PS logos on its underside.
It's actually a 487Mhz PPC. The real keys are that clock-for-clock the PPC is better than the EE (MIPS R5900 as you mention), the CPU cache is much bigger (256k) on the PPC than the EE (8k), the system bus bandwidth is lower - EE is 128bit but only 150Mhz and RDRAM has a long setup time vs the 1T-SRAM used with the PPC - and core floating point performence is much better on the PPC than the EE (eg. physics, AI).
Most PS2 games are EE-limited. The VUs are great for medium (eg. skinning) and low (eg. transform, clipping, lighting) level vertex processing but that's only a relatively small piece of the total work done by a modern game. They have so little integer and flow control support that you can't do much higher level work on them even if you wanted to deal with writing such things in dual-issue microcode.
The Crash platform games were very popular
on
Where Did the Games Go?
·
· Score: 2, Insightful
The first 3 Crash Bandicoot games sold 8.35 million units between them in North America, making them respectively the number 6, number 2 and number 4 all time sellers on the PSone. Though you didn't mention them but someone who replied to you did, the 3 Spyro games sold 5 million units between them with the first title being number 7 on the PSone all time list. I've finished all six of those games and I would definitely say they are extremely high quality products.
The key here is the teams behind the two franchises, namely Naughty Dog and Insomniac. Once they moved on to new things and Universal handed the licenses to other teams, things went south.
It's not what individual publishers expect but it's what the industry collectively must expect when it puts out so many titles in such a short span of time. The industry's output quintuples (or more) during November and December but the available shelf space only grows moderately if at all and the consumer dollars spent probably only triples or quadruples.
We actually decided for our game that, since it's a new title rather than a sequel, it would be better to release it after Christmas when things have calmed down. It meant we got more time to polish the game, more time to get the marketing campaign focused and it was easier to get it onto retail shelves. From what I'm hearing about our pre-order levels it seems like it was the right choice to me - obviously the real test will be when the game goes on sale.
I haven't picked up NPD life-to-date numbers for a few months now but the ones I do have say that Escape From Monkey Island on PS2 had sold through just over 40,000 units (one of which is in my living room). I would not consider that "quite well" although it may have been profitable for them.
I think the biggest problem with these kinds of games is limited shelf space especially for PC titles. Retailers are very risk-averse and it's very hard to get them to carry a title unless they feel really sure it's going to sell fast - whether it's ultimately a good or bad game is largely irrelevant to that equation. It's a shame because it makes the 'sleeper hits' much rarer and as such makes our industry less adventurous (in more ways than one).
You've clearly never met David or Neil for that matter - I know both and they know me. I've worked for this organisation for almost 8 years and I've seen good times and bad times. I say David Zucker is a great CEO, a great motivator and has made a positive impression on just about everybody. As for managers vanishing, look at his own comments and you'll see he's made the most staff changes in marketing, not product development. Right now he has the company focused on making games people in the real world actually want to buy and play. Judge for yourself in March with The Suffering and in April with NBA Ballers (full disclosure: my current project).
I used Exmon on the Beeb and that didn't simulate the 6502 - that would've been much, much too slow. It just swapped the memory at the breakpoint location with an SWI or a JMP (I forget exactly) and put the memory back after the breakpoint was hit.
I didn't find the Exile/Repton Infinity protection very hard to break - took me a week of evenings if I remember rightly. The disc looked like it was only 1 track long normally but there was another catalog sector hidden and encrypted for the loader to use.
The much harder protection to break, in my opinion, was Kevin Edward's 6522 timer based stuff. The code did operations using the 1MHz timer, the encrypted payload and the decryptor machine code for inputs. If you changed the decryptor you'd throw the timer reads off and to correctly simulate the behaviour you needed to know exactly which cycle of a 5-cycle LDA would fetch the low byte of the VIA timer. Nasty!
I'd love to get licenses from Superior for a bunch of their old games and do a compilation on GBA (or better yet PS2/GCN/Xbox), but I think my bosses have more pressing plans for my time...
This was in early 1998 so forgive me if my memory is a little hazy. The first production MK4 machines went out with EPROMs with a 1.0 rev of the software. Those were much more expensive than masked ROMs of the same size but there wasn't the 8 week lead time on production. Later, when the team had done everything else they wanted to they released a 2.0 rev on masked ROM and got the owners of the early games to ship back the EPROMs.
Hobbyist programmers can make their game or more realistically a game prototype on PC or Mac and demonstrate that to potential publishers. It's not hard to move a game's technology from one platform to another for most cases. Making a sufficiently compelling and graphically appealing game is pretty hard these days no matter what platform you're on. I would say that supporting this platform or that platform is only 10% of the programming effort at most. If your game is good enough you'll get funding and access to hardware. If it's not you won't get the backing anyway.
We bought a bunch of some no-name USB adapter so we could plug PS2 controllers into development PCs. They worked but the update rate was only about 15Hz so it was easy to have missed button presses.
I'm sure there are many different models - this particular one also had an N64 controller port with a slider to choose between them and two LEDs to indicate which was active (red for PSX, green for N64).
That's a mauling waiting to happen. There's a great book called "Bear Attacks: Their Causes And Avoidance" by Stephen Herrero (http://www.amazon.com/exec/obidos/tg/detail/-/158 574557X/) which explains the kind of things that happen whenbears get used to humans and human food. One of those kids will probably get his/her head bitten clean off.
The book is a great read, by the way. It alternates between very dry, clinical explanations of statistics and brutally graphic retelings of people being gnawed and clawed and generally brutalised by wild bears. Bill Bryson mentions it in his travelogue "A Walk In The Woods". If it's good enough for him to read, it's good enough for me:-)
Taking a console game and making it run on PC is usually very feasible from a purely technical point of view. There are, however, four things that stand in the way of turning a console game into a PC game.
Firstly, the art is designed for 640x480 on a TV. This means it usually looks bad at 1024x768 on a PC monitor and PC gamers react very negatively to bad art because they spend a bunch of money on their systems in order to have their games look as beautiful as possible.
Secondly, the game is designed to be played with a console gamepad. To sell a game on PC it needs to play well with keyboard and mouse - some people have gamepads but the vast majority do not. This can mean making changes to a lot of game systems.
Thirdly, PCs vary in their specifications. The game must be able to run at different frame rates, at different resolutions and so on. Loading and saving works differently. Data may be coming off CD or off the hard drive. Users can task switch away from the game. Basically, the environment is just different. Depending on how the game code is written these things may be easy to accomodate, or extremely hard.
Lastly, the business environment for PC games is different. You need a much bigger tech support department (all those users with their disparate configurations). The retail channel is somewhat different to work with (margins, major buyers, cost structure, packaging, promotional material).
The bottom line from all of this is that it takes a bunch of money and time to turn a console game into a PC game that anyone would actually want to pay for. Either you do it after the console version is done, or you do it along the way by keeping the game cross-platform. You also need a different set of business resources than you do for console games. It's unclear that, for most console games, the return on investment would exceed the costs of this. Perhaps more importantly the effort spent on PC versions can instead be spent making the console version better.
In my opinion the above are why you don't see many console games come to PC. That's not to say it never happens, of course. Most of the conversions that I'm aware of were done by relatively small developers well after the console version was known to be successful and in many cases the PC versions didn't sell well and were negatively perceived basically because of their console-ness.
The article uses the phrase "system LSI". To me that implies this chip is actually a support chip, for example a DMA and memory controller. It would make sense to test the fab process with something smaller than a full CPU. Besides, they haven't announced that Cell has taped out and since they did that for the PS2 EE and GS I would expect a similar announcement before sampling begins.
The board is indeed much, much cheaper to make. The Pac-man/Galaga board is 3 ASICs and 2 EPROMs. However, the board itself was only ever about 20% of the build cost of most arcade games. Low-res 19" monitors haven't changed much in price over the years, nor has particle board, nor have coin doors, nor has labour.
That said, the real reason the game is 50c/play is that most of the units will wind up in street locations (movie theaters, pizza takeouts, laundromats and so on) where people just want a little distraction to pass the time. Very few of them will care enough about the price that they wouldn't pay 50c. I'd probably add a bill validator and put the game on 3/$1 bonus to catch people without pocket change or to entice repeat play.
I'm sure Two Bit Score or one of the other companies has a coin-doubler that will work with the game for those operators who absolutely insist on putting the thing on 25c play.
XSN's only for first party titles to my knowledge. For reference, I'm the lead programmer on our Xbox SKU and I did the analysis of what it would take to add Xbox Live support to the game.
We could implement our own league system on Xbox Live but it would be really expensive and time-consuming (such servers have to live inside Microsoft's data center so the testing is pretty exhaustive). On PS2 we get all that stuff basically for free because we use Gamespy for our frontend (ie. lobbies, chat, matchmaking).
Of course gamers can do whatever they want informally but that doesn't compare to something integrated and supported by the developer. We can run online tournaments on PS2 and even give away prizes but currently we can't do that on Xbox Live.
Who said anything about infrastructure? The expense in doing an Xbox Live game isn't related to technology or backend, it's the ~100 user interface screens you have to lay out, code and test. None of it's hard to do but it's time-consuming and therefore costly since human resources are the dominant cost of making a game these days.
You'd have to pay extra. Each Live subscription gets you only a single gamertag (your online identity). Everything is done through the gamertag - friends, feedback, stats and so on.
The advantage of the friends list is that it integrates with the rest of the system. If you're online and I log in for some Crimson Skies I can invite you to come play with me no matter what game you're currently playing. You'll get a popup showing the invite and if you accept your system will prompt you for your Crimson Skies disc and reboot and take you to my session. On a more mundane level, you can see how you rank in a game relative to your friends list. If you're creating a game session you can reserve slots for friends (if your friends don't want to fill those slots you can invite whoever else to come play). When you play an open game and meet someone cool you can just add them to the friends list, no need to figure out which IM service (if any) they use, go to the PC, add them, etc.
This is the heart of it. Adding online play to a game takes a lot of engineering effort and testing time and therefore adds to the development cost of the game. This is only worthwhile if it increases sales more than the other potential uses of that development effort. If your expected sell-through is relatively small anyhow because of a smaller hardware base the problem is compounded.
Xbox Live, while certainly a great thing for users, only increases this problem for developers and publishers because it's basically an all-or-nothing service. Worse, it doesn't support some features that publishers want. For example, EA only lets the most recent iteration of each franchise play online - my understanding is that's actually the deal-breaker with Microsoft. There's also no real ability to run leagues or tournaments via Xbox Live.
I mostly agree about the bandwidth demand but there are times when RAM will save you. More RAM means you can preload and/or precalculate more things, or arrange data in a more friendly format. On PS2 people waste some amount of VU1 time packing and unpacking data so their DMA chains use less space in system RAM, for example. We'd also be able to spend the time that is taken up straining to get the game to fit in memory and use it to improve other areas of performance or just generally implement cool, fun things that gamers actually experience.
Gamecube doesn't use RDRAM to my knowledge. The system RAM is some special 1-transistor SRAM design that NEC came up with, and the audio RAM is ordinary PC100.
PS2 has 4megs of local memory for the graphics processor. This is used for frame buffers, Z buffer and textures. DMA from the 32megs of system memory is pretty fast - it's rarely the limiting factor in game performance.
Gamecube has 3megs of local memory, just enough for 1 frame buffer, 1 Z buffer and whichever textures are being used to rasterise the current primitve. Textures are automatically DMAed from the 24megs of (extremely fast and low-latency) system memory as they are needed. There's 16megs of (PC100) audio memory and some games use that as a backing store for extra data - DMA to/from it is quite slow though.
You're right about Xbox's 64megs but it should be noted that the memory is divided by the crossbar memory controller into 4 16meg chunks and by carefully arranging what is in what chunk you get better performance because each chunk can be accessed at full speed (ie. as if the overall memory wasn't shared).
Game consoles really don't need a 64bit address space yet, but they do need a very wide data bus and wide CPU registers. Right now I'd say raw CPU performance and pixel fillrate are actually the two most limiting factors for games. More RAM would certainly be nice but throughput is a bigger concern, at least with the games I've worked on.
We don't make games for Korea on any of the platforms, but this is my understanding of the situation.
Sony localised the PS2 browser for Korean and provides support for game companies to release Korean-language versions of their games (eg. providing 'approved' Korean-language messages for things like "Now saving. Do not remove MEMORY CARD (PS2) in MEMORY CARD slot 1, reset or switch off console." As such PS2 games released in Korea are well-localised. Microsoft, on the other hand, is simply releasing English-language systems and software in Korea.
I am under the impression English is prevalent in Korea but I have a hard time believing Korean gamers won't choose Korean-language games over English ones.
Microsoft was talking about releasing western games in Japan without localising them. The idea was that, since the market was small and almost exclusively very hardcore gamers, they wouldn't mind the games being in English. I don't know if they are still planning to do this.
First a disclaimer: This is my personal understanding of the company history. It's not official Midway comment and it may not be correct.
Atari Inc was split into two companies in 1984. The home game and computer part, Atari Corp, and the arcade game part, Atari Games. Williams acquired Atari Games in 1996. Atari Corp wound up in the hands of Infogrames via JTS and Hasbro. Atari Corp had the rights to do home translations of the pre-split arcade games but Atari Games owned the rights to the originals. Each could use the Atari name and logo but only in their own markets (ie. home vs coin-op).
Rights to the coin-op games included rights to publish emulated versions for home systems and WMS did this via Williams Home Entertainment and Midway Games. Infogrames published updated games based on the original home versions (Pong, Space Invaders, Asteroids) and collections of emulated 2600 games.
Midway scaled back its use of Atari as a brand name because it was transitioning to making home games, where it could not use that brand. Note that when Atari Games made NES versions of its games it used the Tengen brand. The last few coin-op games made in Milpitas were Midway-branded (eg. Gauntlet Dark Legacy, Skins Game) and the studio was renamed Midway Games West.
Since no-one is making coin-op any more, the important rights are only home rights. Basically, if it's from before 1984 Infogrames owns it (with the exception of exact emulation of arcade versions). If it's after 1984 then Midway owns it if it's an arcade game or a Tengen game and Infogrames owns basically everything else.
Complicated, but the company covers the whole life of the videogame era so that's perhaps unsurprising.
Slashdot Mirror
Let me preface all this by saying I'm only talking about console games, my data come from sales figures tallied by NPD, which is a commercial reporting service and I define commercial success as sell-through of over 250,000 units.
90% of all videogames launching new properties are commercial failures. This is well above the overall failure rate of 78%. The sad truth is that most gamers stick with familiar themes and simple concepts. The quirky, unusual, original games you and I like (over 20 of the PS2 games on my shelf sold under 100,000 units) just do not pay the bills.
Therefore it behooves us to think about how to make games that we can enjoy but which will also speak to the mass market. My personal opinion is that the real inventiveness is best served in gameplay mechanics and control. Use the licensed theme as a base to build from. The game market is coming close to maturity. It grew and grew through the 80s and early 90s and now it's near a plateau in terms of userbase. One sign of this is the gamer's average age increasing year on year.
Right now there are plenty of relatively new gamers for whom simple action-shooter, driving and hack-slash gameplay is appealing. They'll quickly grow into more sophisticated play concepts, just as today's hardcore gamers did in years past. The people who figure out how to keep that large pool of gamers interested will be the people who succeed in the industry.
I've played right through the game once and got almost halfway through a second time before growing bored with it. Offhand I'd guess this is about 20 hours of play. My system (bought in April 2001 and used a great deal) is not modded nor do I own a gameshark.
I've seen the loading screen crash once - very early on in fact. When I ejected the disc and put it back in the game recovered perfectly. I've seen the missing tiles problem quite a bit more. The game is almost constantly loading from the DVD and in my opinion makes heavier demands of the PS2 optical drive than any other game - certainly much more than I would ever dare to do in my own code. As such I am unsurprised that it has disc-related issues.
There will be a great many false-positives and false-negatives about these errors. You're right about stupid user syndrome but for each of those people there is at least one person who will assume their PS2 overheated or there was a power spike or they bumped the table.
The one part of this that I don't really understand is why they used a dual-layer DVD. It's a pain in the ass to make and test a dual-layer game and the disc doesn't appear to hold more data than would fir on a single-layer disc. I was also slightly surprised to see the disc doesn't have the 6 PS logos on its underside.
It's actually a 487Mhz PPC. The real keys are that clock-for-clock the PPC is better than the EE (MIPS R5900 as you mention), the CPU cache is much bigger (256k) on the PPC than the EE (8k), the system bus bandwidth is lower - EE is 128bit but only 150Mhz and RDRAM has a long setup time vs the 1T-SRAM used with the PPC - and core floating point performence is much better on the PPC than the EE (eg. physics, AI).
Most PS2 games are EE-limited. The VUs are great for medium (eg. skinning) and low (eg. transform, clipping, lighting) level vertex processing but that's only a relatively small piece of the total work done by a modern game. They have so little integer and flow control support that you can't do much higher level work on them even if you wanted to deal with writing such things in dual-issue microcode.
The first 3 Crash Bandicoot games sold 8.35 million units between them in North America, making them respectively the number 6, number 2 and number 4 all time sellers on the PSone. Though you didn't mention them but someone who replied to you did, the 3 Spyro games sold 5 million units between them with the first title being number 7 on the PSone all time list. I've finished all six of those games and I would definitely say they are extremely high quality products.
The key here is the teams behind the two franchises, namely Naughty Dog and Insomniac. Once they moved on to new things and Universal handed the licenses to other teams, things went south.
It's not what individual publishers expect but it's what the industry collectively must expect when it puts out so many titles in such a short span of time. The industry's output quintuples (or more) during November and December but the available shelf space only grows moderately if at all and the consumer dollars spent probably only triples or quadruples.
We actually decided for our game that, since it's a new title rather than a sequel, it would be better to release it after Christmas when things have calmed down. It meant we got more time to polish the game, more time to get the marketing campaign focused and it was easier to get it onto retail shelves. From what I'm hearing about our pre-order levels it seems like it was the right choice to me - obviously the real test will be when the game goes on sale.
I haven't picked up NPD life-to-date numbers for a few months now but the ones I do have say that Escape From Monkey Island on PS2 had sold through just over 40,000 units (one of which is in my living room). I would not consider that "quite well" although it may have been profitable for them.
I think the biggest problem with these kinds of games is limited shelf space especially for PC titles. Retailers are very risk-averse and it's very hard to get them to carry a title unless they feel really sure it's going to sell fast - whether it's ultimately a good or bad game is largely irrelevant to that equation. It's a shame because it makes the 'sleeper hits' much rarer and as such makes our industry less adventurous (in more ways than one).
You've clearly never met David or Neil for that matter - I know both and they know me. I've worked for this organisation for almost 8 years and I've seen good times and bad times. I say David Zucker is a great CEO, a great motivator and has made a positive impression on just about everybody. As for managers vanishing, look at his own comments and you'll see he's made the most staff changes in marketing, not product development. Right now he has the company focused on making games people in the real world actually want to buy and play. Judge for yourself in March with The Suffering and in April with NBA Ballers (full disclosure: my current project).
I used Exmon on the Beeb and that didn't simulate the 6502 - that would've been much, much too slow. It just swapped the memory at the breakpoint location with an SWI or a JMP (I forget exactly) and put the memory back after the breakpoint was hit.
I didn't find the Exile/Repton Infinity protection very hard to break - took me a week of evenings if I remember rightly. The disc looked like it was only 1 track long normally but there was another catalog sector hidden and encrypted for the loader to use.
The much harder protection to break, in my opinion, was Kevin Edward's 6522 timer based stuff. The code did operations using the 1MHz timer, the encrypted payload and the decryptor machine code for inputs. If you changed the decryptor you'd throw the timer reads off and to correctly simulate the behaviour you needed to know exactly which cycle of a 5-cycle LDA would fetch the low byte of the VIA timer. Nasty!
I'd love to get licenses from Superior for a bunch of their old games and do a compilation on GBA (or better yet PS2/GCN/Xbox), but I think my bosses have more pressing plans for my time...
This was in early 1998 so forgive me if my memory is a little hazy. The first production MK4 machines went out with EPROMs with a 1.0 rev of the software. Those were much more expensive than masked ROMs of the same size but there wasn't the 8 week lead time on production. Later, when the team had done everything else they wanted to they released a 2.0 rev on masked ROM and got the owners of the early games to ship back the EPROMs.
Your value of r at the Earth's surface is 6400km. Therefore at an orbit of 1000km above the Earth's surface r should be 7400km.
Hobbyist programmers can make their game or more realistically a game prototype on PC or Mac and demonstrate that to potential publishers. It's not hard to move a game's technology from one platform to another for most cases. Making a sufficiently compelling and graphically appealing game is pretty hard these days no matter what platform you're on. I would say that supporting this platform or that platform is only 10% of the programming effort at most. If your game is good enough you'll get funding and access to hardware. If it's not you won't get the backing anyway.
We bought a bunch of some no-name USB adapter so we could plug PS2 controllers into development PCs. They worked but the update rate was only about 15Hz so it was easy to have missed button presses.
I'm sure there are many different models - this particular one also had an N64 controller port with a slider to choose between them and two LEDs to indicate which was active (red for PSX, green for N64).
That's a mauling waiting to happen. There's a great book called "Bear Attacks: Their Causes And Avoidance" by Stephen Herrero (http://www.amazon.com/exec/obidos/tg/detail/-/158 574557X/) which explains the kind of things that happen whenbears get used to humans and human food. One of those kids will probably get his/her head bitten clean off.
:-)
The book is a great read, by the way. It alternates between very dry, clinical explanations of statistics and brutally graphic retelings of people being gnawed and clawed and generally brutalised by wild bears. Bill Bryson mentions it in his travelogue "A Walk In The Woods". If it's good enough for him to read, it's good enough for me
Taking a console game and making it run on PC is usually very feasible from a purely technical point of view. There are, however, four things that stand in the way of turning a console game into a PC game.
Firstly, the art is designed for 640x480 on a TV. This means it usually looks bad at 1024x768 on a PC monitor and PC gamers react very negatively to bad art because they spend a bunch of money on their systems in order to have their games look as beautiful as possible.
Secondly, the game is designed to be played with a console gamepad. To sell a game on PC it needs to play well with keyboard and mouse - some people have gamepads but the vast majority do not. This can mean making changes to a lot of game systems.
Thirdly, PCs vary in their specifications. The game must be able to run at different frame rates, at different resolutions and so on. Loading and saving works differently. Data may be coming off CD or off the hard drive. Users can task switch away from the game. Basically, the environment is just different. Depending on how the game code is written these things may be easy to accomodate, or extremely hard.
Lastly, the business environment for PC games is different. You need a much bigger tech support department (all those users with their disparate configurations). The retail channel is somewhat different to work with (margins, major buyers, cost structure, packaging, promotional material).
The bottom line from all of this is that it takes a bunch of money and time to turn a console game into a PC game that anyone would actually want to pay for. Either you do it after the console version is done, or you do it along the way by keeping the game cross-platform. You also need a different set of business resources than you do for console games. It's unclear that, for most console games, the return on investment would exceed the costs of this. Perhaps more importantly the effort spent on PC versions can instead be spent making the console version better.
In my opinion the above are why you don't see many console games come to PC. That's not to say it never happens, of course. Most of the conversions that I'm aware of were done by relatively small developers well after the console version was known to be successful and in many cases the PC versions didn't sell well and were negatively perceived basically because of their console-ness.
The article uses the phrase "system LSI". To me that implies this chip is actually a support chip, for example a DMA and memory controller. It would make sense to test the fab process with something smaller than a full CPU. Besides, they haven't announced that Cell has taped out and since they did that for the PS2 EE and GS I would expect a similar announcement before sampling begins.
The board is indeed much, much cheaper to make. The Pac-man/Galaga board is 3 ASICs and 2 EPROMs. However, the board itself was only ever about 20% of the build cost of most arcade games. Low-res 19" monitors haven't changed much in price over the years, nor has particle board, nor have coin doors, nor has labour.
That said, the real reason the game is 50c/play is that most of the units will wind up in street locations (movie theaters, pizza takeouts, laundromats and so on) where people just want a little distraction to pass the time. Very few of them will care enough about the price that they wouldn't pay 50c. I'd probably add a bill validator and put the game on 3/$1 bonus to catch people without pocket change or to entice repeat play.
I'm sure Two Bit Score or one of the other companies has a coin-doubler that will work with the game for those operators who absolutely insist on putting the thing on 25c play.
It's more expensive to add Xbox Live support to a game than to add PS2 online support - more expensive by a large margin, in fact.
XSN's only for first party titles to my knowledge. For reference, I'm the lead programmer on our Xbox SKU and I did the analysis of what it would take to add Xbox Live support to the game.
We could implement our own league system on Xbox Live but it would be really expensive and time-consuming (such servers have to live inside Microsoft's data center so the testing is pretty exhaustive). On PS2 we get all that stuff basically for free because we use Gamespy for our frontend (ie. lobbies, chat, matchmaking).
Of course gamers can do whatever they want informally but that doesn't compare to something integrated and supported by the developer. We can run online tournaments on PS2 and even give away prizes but currently we can't do that on Xbox Live.
Who said anything about infrastructure? The expense in doing an Xbox Live game isn't related to technology or backend, it's the ~100 user interface screens you have to lay out, code and test. None of it's hard to do but it's time-consuming and therefore costly since human resources are the dominant cost of making a game these days.
You'd have to pay extra. Each Live subscription gets you only a single gamertag (your online identity). Everything is done through the gamertag - friends, feedback, stats and so on.
The advantage of the friends list is that it integrates with the rest of the system. If you're online and I log in for some Crimson Skies I can invite you to come play with me no matter what game you're currently playing. You'll get a popup showing the invite and if you accept your system will prompt you for your Crimson Skies disc and reboot and take you to my session. On a more mundane level, you can see how you rank in a game relative to your friends list. If you're creating a game session you can reserve slots for friends (if your friends don't want to fill those slots you can invite whoever else to come play). When you play an open game and meet someone cool you can just add them to the friends list, no need to figure out which IM service (if any) they use, go to the PC, add them, etc.
This is the heart of it. Adding online play to a game takes a lot of engineering effort and testing time and therefore adds to the development cost of the game. This is only worthwhile if it increases sales more than the other potential uses of that development effort. If your expected sell-through is relatively small anyhow because of a smaller hardware base the problem is compounded.
Xbox Live, while certainly a great thing for users, only increases this problem for developers and publishers because it's basically an all-or-nothing service. Worse, it doesn't support some features that publishers want. For example, EA only lets the most recent iteration of each franchise play online - my understanding is that's actually the deal-breaker with Microsoft. There's also no real ability to run leagues or tournaments via Xbox Live.
I mostly agree about the bandwidth demand but there are times when RAM will save you. More RAM means you can preload and/or precalculate more things, or arrange data in a more friendly format. On PS2 people waste some amount of VU1 time packing and unpacking data so their DMA chains use less space in system RAM, for example. We'd also be able to spend the time that is taken up straining to get the game to fit in memory and use it to improve other areas of performance or just generally implement cool, fun things that gamers actually experience.
Gamecube doesn't use RDRAM to my knowledge. The system RAM is some special 1-transistor SRAM design that NEC came up with, and the audio RAM is ordinary PC100.
PS2 has 4megs of local memory for the graphics processor. This is used for frame buffers, Z buffer and textures. DMA from the 32megs of system memory is pretty fast - it's rarely the limiting factor in game performance.
Gamecube has 3megs of local memory, just enough for 1 frame buffer, 1 Z buffer and whichever textures are being used to rasterise the current primitve. Textures are automatically DMAed from the 24megs of (extremely fast and low-latency) system memory as they are needed. There's 16megs of (PC100) audio memory and some games use that as a backing store for extra data - DMA to/from it is quite slow though.
You're right about Xbox's 64megs but it should be noted that the memory is divided by the crossbar memory controller into 4 16meg chunks and by carefully arranging what is in what chunk you get better performance because each chunk can be accessed at full speed (ie. as if the overall memory wasn't shared).
Game consoles really don't need a 64bit address space yet, but they do need a very wide data bus and wide CPU registers. Right now I'd say raw CPU performance and pixel fillrate are actually the two most limiting factors for games. More RAM would certainly be nice but throughput is a bigger concern, at least with the games I've worked on.
We don't make games for Korea on any of the platforms, but this is my understanding of the situation.
Sony localised the PS2 browser for Korean and provides support for game companies to release Korean-language versions of their games (eg. providing 'approved' Korean-language messages for things like "Now saving. Do not remove MEMORY CARD (PS2) in MEMORY CARD slot 1, reset or switch off console." As such PS2 games released in Korea are well-localised. Microsoft, on the other hand, is simply releasing English-language systems and software in Korea.
I am under the impression English is prevalent in Korea but I have a hard time believing Korean gamers won't choose Korean-language games over English ones.
Microsoft was talking about releasing western games in Japan without localising them. The idea was that, since the market was small and almost exclusively very hardcore gamers, they wouldn't mind the games being in English. I don't know if they are still planning to do this.
First a disclaimer: This is my personal understanding of the company history. It's not official Midway comment and it may not be correct.
Atari Inc was split into two companies in 1984. The home game and computer part, Atari Corp, and the arcade game part, Atari Games. Williams acquired Atari Games in 1996. Atari Corp wound up in the hands of Infogrames via JTS and Hasbro. Atari Corp had the rights to do home translations of the pre-split arcade games but Atari Games owned the rights to the originals. Each could use the Atari name and logo but only in their own markets (ie. home vs coin-op).
Rights to the coin-op games included rights to publish emulated versions for home systems and WMS did this via Williams Home Entertainment and Midway Games. Infogrames published updated games based on the original home versions (Pong, Space Invaders, Asteroids) and collections of emulated 2600 games.
Midway scaled back its use of Atari as a brand name because it was transitioning to making home games, where it could not use that brand. Note that when Atari Games made NES versions of its games it used the Tengen brand. The last few coin-op games made in Milpitas were Midway-branded (eg. Gauntlet Dark Legacy, Skins Game) and the studio was renamed Midway Games West.
Since no-one is making coin-op any more, the important rights are only home rights. Basically, if it's from before 1984 Infogrames owns it (with the exception of exact emulation of arcade versions). If it's after 1984 then Midway owns it if it's an arcade game or a Tengen game and Infogrames owns basically everything else.
Complicated, but the company covers the whole life of the videogame era so that's perhaps unsurprising.