Until developers program games with PPC code in mind, then no, it will not be a "gaming machine" other than the card you put into it. In the world of drawing polygons, MHz is MHz. No fancy shortened pipeline trickery (something as simple and straightforward as keeping track of polygonal models is never really fraught with data branches) or ridiculous 1GHz front side bus bandwidth is going to change that. In fact, the rawest power to render with brute force is vastly preferable over any kind of efficiency that the G5 may provide over its competition.
So a 2GHz G5 is pretty much equivalent to a 2GHz P4 when running any modern game (Quake III is now 4 years old). And let's not pretend duals actually matter (the only time I've EVER seen a game that threaded was when Omni Group sunk their teeth into someone else's game to do a Cocoa version).
EDIT: And by the way, Ikaruga is among my top games this year. Everything about that game is utter brilliance, particularly the "condensed excellence" of the level design. The game may be about 25 minutes long, but there's more unique and interesting content in that game than just about any other I know of.
While the G5 will be way faster than a G4 on games, I wouldn't consider it a gaming machine.
Interestingly, processing power has rarely had much effect on a platform's success as the preferred gaming platform. We care about the specs, but in the end, it mostly comes down to game selection, cost, and overall experience. Console systems are always much slower than computers but still offer an exceptional gaming experience. This is despite them costing less than a PC's graphics card.
The G5 will run the Mac's small game-library really well, but it isn't a 'gaming machine'.
In order for the G5 to develop as a gaming machine developers need to follow the lead of Blizzard and Id, making openGL games that come out for both systems at once...
At the heart of that issue is having OpenGL be more powerful, robust, and generally more accepted than DirectX - thats really all that prevents this crossover from happening. Though as long as M$ systems dominate gaming systems it will be a long time coming before DirectX starts to die out.
In order for the G5 to develop as a gaming machine developers need to follow the lead of Blizzard and Id, making openGL games that come out for both systems at once...
At the heart of that issue is having OpenGL be more powerful, robust, and generally more accepted than DirectX - thats really all that prevents this crossover from happening. Though as long as M$ systems dominate gaming systems it will be a long time coming before DirectX starts to die out.
OpenGL 1.5 is a step in the right direction and 2.0 will take it even further. Apple should get on the horn and make sure Panther and it's successors have current support for 1.5 and 2.0 asap.
Until developers program games with PPC code in mind, then no, it will not be a "gaming machine" other than the card you put into it. In the world of drawing polygons, MHz is MHz. No fancy shortened pipeline trickery (something as simple and straightforward as keeping track of polygonal models is never really fraught with data branches) or ridiculous 1GHz front side bus bandwidth is going to change that. In fact, the rawest power to render with brute force is vastly preferable over any kind of efficiency that the G5 may provide over its competition.
So you say that for games, mhz is the only factor of performance, and how the processor is architected is irrelevant?
What is your definition of "raw power to render with brute force"?
Interesting... I'll take this one step further and assert that processing power and architectural speed concerns have little bearing on the Mac games market. Of more significance is cost and game selection, no matter the speed.
I disagree. The ability of the hardware to run the games will, in my opinion, have a direct effect on which games get ported to the Mac and how quickly and how well. Game developers want to beat up your hardware, make it scream for mercy. Don't think for one second the capabilities of the G5 have escaped their notice. I expect the robustness of the G5 hardware to make a difference in the gaming market on the Mac side.
Processing power has a lot to do with the mac and gaming. I built my first PC because the top of the line mac at that time did not have the ability to play Unreal Tournament at a good frame rate.
The new dual G5 with a 9800pro should be able to handle anything that the wentel machines can handle.
Of course, what would really help Mac gaming would be decent horse power and video cards in the iMac line. (Consumers like to play games -- I've never understood why Apple always puts such low end cards in the iMac.)
The GeForce4MX is decent enough for what games need. The iMac can easily handle UT2003. Plus, a better graphics card would mean higher price, and the iMac is already overpriced.
The GF4MX were possibly enough for games two years ago. It might well run any game, but not at competitive performance. It is enough to run Warcraft 3, but not at decent resolutions at decent, competitive framerates. It will choke on Doom 3 in the future. It is, in difference to the G5, far behind the Wintel side at running games, even when forgetting the abysmal selection of games.
Res is right, the iMac needs, and has always needed (but never gotten) a decent graphics chip. It's targeted at families, consumers, which are very fond of games. When it is unfeasible to get it as a machine for gaming in addition to everything else, why develop games for it? If people have little intencive to buy games because they will run slowly on their iMacs, the market is shrunk.
When games runs on fast hardware, it makes gaming (more) a pleasure. Thus, when you have fast hardware, you're more likely to buy more games, because it's fun to play them at the highest resolution, with fantastic graphics, at good framerates. This generates higher sales, not instantly - slowly, but surely.
So you say that for games, mhz is the only factor of performance, and how the processor is architected is irrelevant?
What is your definition of "raw power to render with brute force"?
That's correct. Most game engines are designed to get the most out of the decades-old X86 architecture, tossing as many polygons as possible onto the screen while the graphics card handled effects and other such things. Polygon-drawing is, rudimentarily, a math process. It's all integer calculations, and the G5 actually has less basic integer oomph than a Pentium 4. Whereas scaling images and converting video can be handled in chunks and benefit from AltiVec and the overall insane floating point power of the G5, this isn't so with polygons.
What this means is, polygons will draw as fast as you can draw them (as fast the processor's clock frequency is). You may have caught Jon Rubinstein's explanation on data branching, and the ephemeral nature of certain data properties: as data travels down the processor's pipeline, there is a tendency for the entire pipeline to drain if the data you're handling is dynamic. This is the case, again, with Photoshop work, but not something as basic as drawing polygons. Even though the Pentium 4 has a twenty-or-so-stage pipeline, it will rarely drain so much as to put a strain on the processor.
This would be different if developers used NURBs or a very floating-point-intensive type of rendering; in that case, the G5 would totally own any Intel chip. But because most of the world still uses an archaic architecture, it will be years before you ever see that happen. We've already begun to see NURBs and other floating-point-intensive methods appear on consoles (GameCube and PS2; not Xbox, since that is based on X86), because those are closed architectures, and, for the most part, comparatively modern ones.
The GeForce4MX is decent enough for what games need. The iMac can easily handle UT2003. Plus, a better graphics card would mean higher price, and the iMac is already overpriced.
The GeForce4MX in an iMac easily handle UT2003??? That card can't run UT2003 decently in a 2+Ghz P4. Even the GeForce4 Ti 4600 can't handle UT2003 well. You need a Radeon 9500pro or better for that game (and if you want to use any anti-aliasing or anisotropic filtering with it you really need the 9700pro or better).
Apple has always saddled the iMac with a slow video-card, and it has truly hurt the Mac gaming market.
That's correct. Most game engines are designed to get the most out of the decades-old X86 architecture, tossing as many polygons as possible onto the screen while the graphics card handled effects and other such things. Polygon-drawing is, rudimentarily, a math process. It's all integer calculations, and the G5 actually has less basic integer oomph than a Pentium 4. Whereas scaling images and converting video can be handled in chunks and benefit from AltiVec and the overall insane floating point power of the G5, this isn't so with polygons.
What about physics, AI, and sound-engines in games? They (especially physics and AI) steadily increase in complexity too.
What about physics, AI, and sound-engines in games? They (especially physics and AI) steadily increase in complexity too.
Again, real-time AI and physics routines were created specifically for X86 hardware, and there is no benefit from the G5's modern architecture. The extraordinary physics and altered displacement mapping of the upcoming Half-Life 2, for example, can run on very low-end hardware, such as a 700MHz Pentium III. This is also the case with the artificial intelligence, scripting, facial animation, lip-synching, and the assembly of these components into a seamless presentation unbreakable by user input. No, the purpose of all this new, powerful hardware isn't to fuel any new game design or smarter enemy patterning, but just to make everything all purdy.
Sound engines require a bit more power, because of new processes that synchronize multiple tracks of streaming audio to the user's actions and intertwine and fade between them as needed, and it requires some processing power to keep track of all this stuff and make it dynamic. Not to mention 5.1 separation. Fortunately, though, most PC's have sound cards to handle all of this work. Though the G5 supports 5.1 through an on-board optical input, I don't believe it has a dedicated sound processor, just like every other Mac. Sure, you can run out and buy one (one company whose name escapes me has a really nice 7.1 card), but it's odd that a $1999 machine ships without a sound card.
Though the G5 supports 5.1 through an on-board optical input, I don't believe it has a dedicated sound processor, just like every other Mac. Sure, you can run out and buy one (one company whose name escapes me has a really nice 7.1 card), but it's odd that a $1999 machine ships without a sound card.
The card you're thinking of is the M-Audio Revolution.
As for Apple, when you can get more tracks, more effects,more flexibility and better latency from software, why bother with hardware? All that means is that your capabilities are rigidly defined, and you're dependent on drivers. If the machine's a duallie, then OS X can just put the game on one CPU and CoreAudio on the other.
As for game development, the Mac does have advantages as a platform: When the CRT iMac reigned, it did not escape the game devs that it acted a lot like a console: The video was always ATi, the CPU was always a G3, the screen was always the same - a lot of the variables that frustrate developers were simply moot. This is also why id released the Q3 preview first on Macs: The hardware's more controlled, so the problem of developing for them is much simpler, especially in a performance-critical application.
Raw performance is a secondary consideration. Very few developers really push the envelope, because that just means that they're shrinking their market. (Think of all the games that are built on two and three year old engines.) If they're smart, they're have the game push the hardware when all the settings are maxed out, but ship the game with the settings lower. The iMac was an easy target with adequate performance and a base of millions, and that's going to attract a lot of attention.
One of the big obstacles, in fact, is this simple combination of facts: Most developers in game houses use MS Visual Studio, which of course makes using all the MS libraries easy. Most managers know this, and they know that most of their games will sell on MS platforms, so they see it as a significant advantage in time and cost to stay all MS and punt the Mac problem until the game's pulled in enough money to cover the risk.
As id discovered, there's also a political consideration: If you do the sensible thing from a game development POV, and (pre-)release your game on the Mac first (because the Mac is a far more controlled platform) the Windows users who form your principle customer base will become enraged.
As for x86-optimized code: Yes, that's a problem. Fortunately, the 970 has a capability to rearrange crap code to its liking that the G4 never did, the deep pipelines to chew through P4-optimized code, and the caches to make up for x86 code that assumes a paucity of architectural registers (although the 7455's L1 cache is considerably nicer than the 970's).
The performance of the G5 will help to some extent, but the real issue is the white elephant no one wants to mention - market share. It was a sad day when I found out we were having trouble holding on to 3%. These levels are not sustainable! Apple should be doing whatever it has to in order to grow market share. Losses in the short term would be negligible in comparison to the gains of a couple of market share percentage points.
The G5 is here and it's great, but the G5 won't be hot and new for too long. Soon enough it will become the status quo. What is going to drive long term demand? I love Apple, but I really think it's about time to increase the risk level in pursuit of marketshare. The retail stores are great, but they're simply not enough. We need more Mac users - a lot more - and once they're converted they won't go back to the PC. Apple continues to struggle for short term profits instead of looking out for long term survival. We desperately need market share.
The performance of the G5 will help to some extent, but the real issue is the white elephant no one wants to mention - market share.
This is along the lines of what I was alluding to. There are more significant factors than processing power affecting the Mac gaming experience.
\t
Game selection and performance is greatly dependent on market size. A larger market means a higher return on investment for Mac game developers. If more profit is expected, then more games will be written/ported and more time will be spent optimizing these games.
While the PS2 isn't a speed demon compared to even last year's middle of the line machines, it offers a great gaming experience. There are tons of developers with experience in efficiently and quickly coding for the platform. This is more responsible than raw processing power for making the PS2 a good 'gaming machine'.
Comments
So a 2GHz G5 is pretty much equivalent to a 2GHz P4 when running any modern game (Quake III is now 4 years old). And let's not pretend duals actually matter (the only time I've EVER seen a game that threaded was when Omni Group sunk their teeth into someone else's game to do a Cocoa version).
EDIT: And by the way, Ikaruga is among my top games this year. Everything about that game is utter brilliance, particularly the "condensed excellence" of the level design. The game may be about 25 minutes long, but there's more unique and interesting content in that game than just about any other I know of.
Ikaruga looks interesting!
Interestingly, processing power has rarely had much effect on a platform's success as the preferred gaming platform. We care about the specs, but in the end, it mostly comes down to game selection, cost, and overall experience. Console systems are always much slower than computers but still offer an exceptional gaming experience. This is despite them costing less than a PC's graphics card.
The G5 will run the Mac's small game-library really well, but it isn't a 'gaming machine'.
The G4 wasn't bad. Initially Macs did not utilize Write Combining on the AGP bus. Once they did the speed of games on Macs took a leap forward.
The rest of the differences are Game Code and Driver Development which moves at a more brisk pace on the PC.
At the heart of that issue is having OpenGL be more powerful, robust, and generally more accepted than DirectX - thats really all that prevents this crossover from happening. Though as long as M$ systems dominate gaming systems it will be a long time coming before DirectX starts to die out.
Originally posted by Shanksta
In order for the G5 to develop as a gaming machine developers need to follow the lead of Blizzard and Id, making openGL games that come out for both systems at once...
At the heart of that issue is having OpenGL be more powerful, robust, and generally more accepted than DirectX - thats really all that prevents this crossover from happening. Though as long as M$ systems dominate gaming systems it will be a long time coming before DirectX starts to die out.
OpenGL 1.5 is a step in the right direction and 2.0 will take it even further. Apple should get on the horn and make sure Panther and it's successors have current support for 1.5 and 2.0 asap.
Originally posted by Steve
Until developers program games with PPC code in mind, then no, it will not be a "gaming machine" other than the card you put into it. In the world of drawing polygons, MHz is MHz. No fancy shortened pipeline trickery (something as simple and straightforward as keeping track of polygonal models is never really fraught with data branches) or ridiculous 1GHz front side bus bandwidth is going to change that. In fact, the rawest power to render with brute force is vastly preferable over any kind of efficiency that the G5 may provide over its competition.
So you say that for games, mhz is the only factor of performance, and how the processor is architected is irrelevant?
What is your definition of "raw power to render with brute force"?
The new dual G5 with a 9800pro should be able to handle anything that the wentel machines can handle.
Of course, what would really help Mac gaming would be decent horse power and video cards in the iMac line. (Consumers like to play games -- I've never understood why Apple always puts such low end cards in the iMac.)
Res is right, the iMac needs, and has always needed (but never gotten) a decent graphics chip. It's targeted at families, consumers, which are very fond of games. When it is unfeasible to get it as a machine for gaming in addition to everything else, why develop games for it? If people have little intencive to buy games because they will run slowly on their iMacs, the market is shrunk.
When games runs on fast hardware, it makes gaming (more) a pleasure. Thus, when you have fast hardware, you're more likely to buy more games, because it's fun to play them at the highest resolution, with fantastic graphics, at good framerates. This generates higher sales, not instantly - slowly, but surely.
Originally posted by Zapchud
So you say that for games, mhz is the only factor of performance, and how the processor is architected is irrelevant?
What is your definition of "raw power to render with brute force"?
That's correct. Most game engines are designed to get the most out of the decades-old X86 architecture, tossing as many polygons as possible onto the screen while the graphics card handled effects and other such things. Polygon-drawing is, rudimentarily, a math process. It's all integer calculations, and the G5 actually has less basic integer oomph than a Pentium 4. Whereas scaling images and converting video can be handled in chunks and benefit from AltiVec and the overall insane floating point power of the G5, this isn't so with polygons.
What this means is, polygons will draw as fast as you can draw them (as fast the processor's clock frequency is). You may have caught Jon Rubinstein's explanation on data branching, and the ephemeral nature of certain data properties: as data travels down the processor's pipeline, there is a tendency for the entire pipeline to drain if the data you're handling is dynamic. This is the case, again, with Photoshop work, but not something as basic as drawing polygons. Even though the Pentium 4 has a twenty-or-so-stage pipeline, it will rarely drain so much as to put a strain on the processor.
This would be different if developers used NURBs or a very floating-point-intensive type of rendering; in that case, the G5 would totally own any Intel chip. But because most of the world still uses an archaic architecture, it will be years before you ever see that happen. We've already begun to see NURBs and other floating-point-intensive methods appear on consoles (GameCube and PS2; not Xbox, since that is based on X86), because those are closed architectures, and, for the most part, comparatively modern ones.
Originally posted by CubeDude
The GeForce4MX is decent enough for what games need. The iMac can easily handle UT2003. Plus, a better graphics card would mean higher price, and the iMac is already overpriced.
The GeForce4MX in an iMac easily handle UT2003??? That card can't run UT2003 decently in a 2+Ghz P4. Even the GeForce4 Ti 4600 can't handle UT2003 well. You need a Radeon 9500pro or better for that game (and if you want to use any anti-aliasing or anisotropic filtering with it you really need the 9700pro or better).
Apple has always saddled the iMac with a slow video-card, and it has truly hurt the Mac gaming market.
Originally posted by Steve
That's correct. Most game engines are designed to get the most out of the decades-old X86 architecture, tossing as many polygons as possible onto the screen while the graphics card handled effects and other such things. Polygon-drawing is, rudimentarily, a math process. It's all integer calculations, and the G5 actually has less basic integer oomph than a Pentium 4. Whereas scaling images and converting video can be handled in chunks and benefit from AltiVec and the overall insane floating point power of the G5, this isn't so with polygons.
What about physics, AI, and sound-engines in games? They (especially physics and AI) steadily increase in complexity too.
Originally posted by Zapchud
What about physics, AI, and sound-engines in games? They (especially physics and AI) steadily increase in complexity too.
Again, real-time AI and physics routines were created specifically for X86 hardware, and there is no benefit from the G5's modern architecture. The extraordinary physics and altered displacement mapping of the upcoming Half-Life 2, for example, can run on very low-end hardware, such as a 700MHz Pentium III. This is also the case with the artificial intelligence, scripting, facial animation, lip-synching, and the assembly of these components into a seamless presentation unbreakable by user input. No, the purpose of all this new, powerful hardware isn't to fuel any new game design or smarter enemy patterning, but just to make everything all purdy.
Sound engines require a bit more power, because of new processes that synchronize multiple tracks of streaming audio to the user's actions and intertwine and fade between them as needed, and it requires some processing power to keep track of all this stuff and make it dynamic. Not to mention 5.1 separation. Fortunately, though, most PC's have sound cards to handle all of this work. Though the G5 supports 5.1 through an on-board optical input, I don't believe it has a dedicated sound processor, just like every other Mac. Sure, you can run out and buy one (one company whose name escapes me has a really nice 7.1 card), but it's odd that a $1999 machine ships without a sound card.
Originally posted by Steve
Though the G5 supports 5.1 through an on-board optical input, I don't believe it has a dedicated sound processor, just like every other Mac. Sure, you can run out and buy one (one company whose name escapes me has a really nice 7.1 card), but it's odd that a $1999 machine ships without a sound card.
The card you're thinking of is the M-Audio Revolution.
As for Apple, when you can get more tracks, more effects,more flexibility and better latency from software, why bother with hardware? All that means is that your capabilities are rigidly defined, and you're dependent on drivers. If the machine's a duallie, then OS X can just put the game on one CPU and CoreAudio on the other.
As for game development, the Mac does have advantages as a platform: When the CRT iMac reigned, it did not escape the game devs that it acted a lot like a console: The video was always ATi, the CPU was always a G3, the screen was always the same - a lot of the variables that frustrate developers were simply moot. This is also why id released the Q3 preview first on Macs: The hardware's more controlled, so the problem of developing for them is much simpler, especially in a performance-critical application.
Raw performance is a secondary consideration. Very few developers really push the envelope, because that just means that they're shrinking their market. (Think of all the games that are built on two and three year old engines.) If they're smart, they're have the game push the hardware when all the settings are maxed out, but ship the game with the settings lower. The iMac was an easy target with adequate performance and a base of millions, and that's going to attract a lot of attention.
One of the big obstacles, in fact, is this simple combination of facts: Most developers in game houses use MS Visual Studio, which of course makes using all the MS libraries easy. Most managers know this, and they know that most of their games will sell on MS platforms, so they see it as a significant advantage in time and cost to stay all MS and punt the Mac problem until the game's pulled in enough money to cover the risk.
As id discovered, there's also a political consideration: If you do the sensible thing from a game development POV, and (pre-)release your game on the Mac first (because the Mac is a far more controlled platform) the Windows users who form your principle customer base will become enraged.
As for x86-optimized code: Yes, that's a problem. Fortunately, the 970 has a capability to rearrange crap code to its liking that the G4 never did, the deep pipelines to chew through P4-optimized code, and the caches to make up for x86 code that assumes a paucity of architectural registers (although the 7455's L1 cache is considerably nicer than the 970's).
The G5 is here and it's great, but the G5 won't be hot and new for too long. Soon enough it will become the status quo. What is going to drive long term demand? I love Apple, but I really think it's about time to increase the risk level in pursuit of marketshare. The retail stores are great, but they're simply not enough. We need more Mac users - a lot more - and once they're converted they won't go back to the PC. Apple continues to struggle for short term profits instead of looking out for long term survival. We desperately need market share.
Originally posted by Big Mac
The performance of the G5 will help to some extent, but the real issue is the white elephant no one wants to mention - market share.
This is along the lines of what I was alluding to. There are more significant factors than processing power affecting the Mac gaming experience.
\t
Game selection and performance is greatly dependent on market size. A larger market means a higher return on investment for Mac game developers. If more profit is expected, then more games will be written/ported and more time will be spent optimizing these games.
While the PS2 isn't a speed demon compared to even last year's middle of the line machines, it offers a great gaming experience. There are tons of developers with experience in efficiently and quickly coding for the platform. This is more responsible than raw processing power for making the PS2 a good 'gaming machine'.