t has been my impression of the modern games industry that the resources available are uncanny... it seems to me that the execution engines could be quite sophisticated with that order of effort.
I don't completely believe fragmentation would be that much of an issue. Could it?
Yes, it could. Next question?
If you allocate based on the spacial position of the artifacts being loaded, then you can pretty much guarantee that fragmentation can be avoided.
Interesting idea -- you'd to take the entire resource arena as a circular buffer, and do rolling loads through it, in effect. Yeah, that could actually work if the game was perfectly linear (and set us up for the next Slashdot post, "Why Are Games So Linear?", in the comments of which some clever slashdotter will explain how easy it is to do a nonlinear game by merely loading whatever's near the player whichever direction the player goes, and I hate you all).
If you can defragment at all; most game engine resource managers don't have the ability to relocate loaded resources without unloading and reloading a load unit.
This kind of thing always sounds like a great idea, but it has a lot of engineering subtlety and complexity that might not be obvious at first. It's all solvable, of course, but sometimes a game development team decides to solve a different set of problems.
Several games do this to make individual levels big, then have multiple levels on top of that. (Lord of the Rings: Return of the King and From Russia With Love, PS2 and XBox, to name two).
Assume you're breaking up and entire game into small loadable chunks, and you have a linear system where five chunks are loaded at any given time, two in front of and two behind the player. If there's a basic grunt enemy you fight over and over throughout the game, then you can either put him into a common, always-loaded area of memory, or you can put an extra copy of him into each loadable chunk. Too much of the former and the common area gets to be game-sized and you don't have enough memory left for the chunks. Too much of the latter, and you're wasting memory on multiple copies of resources in memory at once. Now say you're using a grunt widely in the first third and the last third of the game, but not so much in the middle. Breaking the game up into levels lets you put the grunt into a "level common" region instead of in a game common, which is more efficient.
Very few loading screens, actually. But OP is whining about having to wait a few seconds for his game to continue because he's no better than a rat addicted to cocaine hitting a lever over and over, not complaining about the division of the narrative structure of the game.
Um. They stop the game for several seconds at that point, even if they don't render a loading screen. It's levels; it just happens that the exit foyer for one level looks exactly like the entry foyer for the next.
As a game programmer who is currently having to deal with the complexity of memory management in a streaming open-world environment, I'd like to say shut up, I hate you.
Or to put it a little more politely, once you take away the known-memory state checkpoints that you reach between levels, you start having to worry about fragmentation of memory, so you start instituting fixed-size memory "slots" for assets, which deals with the fragmentation problem, but then you sometimes aren't optimally using memory, and then the designers start wanting things to follow you through the world, or allowing you to carry things back and forth through the world, so you have to manage memory outside of the slot system as well as within it, so you have the fragmentation problem again, and then you have to sneak into the designer's house late at night and stab him to death with an icicle.
Nyquist doesn't say, but I do: approximately 6dB of SNR per linear sampling bit, so a Redbook CD can do 96dB SNR, whereas I understand that good vinyl is somewhere under 70dB.
I bet you twenty bucks you're still going to have your ass parked in a chair in front of a half-kilowatt computer saying the same damn thing on the same damn slashdot in ten years, not living "off-grid and low-tech".
Best part is, if you're living off-grid and low-tech in ten years, you won't be able to reach me to collect on the bet.
You can reproduce any analog effect digitally. The way a simply built tube saturates when overdriven just happens to be nice, and the way a simply built digital amplifier saturates when overdriven just happens to be awful. If you have enough computing power to compute something as simple as y = tanh( x * gain ) instead of y = clip( x * gain ), digital sounds "warm".
Nyquist is an exclusive boundary and a theoretical limit -- at 44100Hz, you can reproduce any signal up to and not including 22050Hz for the reason you describe. However, for any frequency less than that, the amplitude of the signal *can* be accurately reconstructed if the reconstruction filter is good enough. What looks, in a naive reconstruction, like frequency modulation due to taking samples at different phases of the source signal when the signal is near Nyquist actually goes away in the filter if the filter is designed properly.
The article specifically states that she ignored questions about the shirt and putty then walked away.
And newspaper articles are always accurate. And people who overreact to apparent security threats never misreport the facts in order to justify their overreactions.
Strange, on my last cross-country flight I watched "L.A. Story" and two episodes of Lost without coming close to killing the battery, on my first-rev iPod Video. Perhaps you're speaking ex recto?
Yes, the phone still rings if it's sleeping.
If you're expecting something else, then you're thinking about four power modes: Off, Standby (non-functional but able to instantly wake up), Sleep (black screen, not accepting touch screen input, but receiving phone calls), and On.
This would be even more confusing than the apparently-too-confusing three modes the iPhone has.
I think I was the first one to buy an iphone for myself. Almost everyone in line was buying them in quantity to either sell at a profit via ebay (haven't heard of success at that).
Considering that, to the best of my knowledge, Apple stores were limiting sales to two per customer, and at least one AT&T store limited them to one, that sounds unlikely. Which kind of makes the rest of your story suspect as well.
Development is not a sunk fixed cost except to a very very crude approximation. It increases roughly linearly with time-on-market. Bug discovery and feature demands are going to come in proportional to roughly log(number-of-users); Apple now has a bunch of fulltime permanent employees working on improving iPhone.
Where's the monopoly, FFS? Every third comment on any iPhone-related post is telling me how there are a dozen other better cheaper touch screen cell phone PDAs available.
Slashdot Mirror
*sob*
I don't completely believe fragmentation would be that much of an issue. Could it? Yes, it could. Next question? If you allocate based on the spacial position of the artifacts being loaded, then you can pretty much guarantee that fragmentation can be avoided. Interesting idea -- you'd to take the entire resource arena as a circular buffer, and do rolling loads through it, in effect. Yeah, that could actually work if the game was perfectly linear (and set us up for the next Slashdot post, "Why Are Games So Linear?", in the comments of which some clever slashdotter will explain how easy it is to do a nonlinear game by merely loading whatever's near the player whichever direction the player goes, and I hate you all).
If you can defragment at all; most game engine resource managers don't have the ability to relocate loaded resources without unloading and reloading a load unit.
This kind of thing always sounds like a great idea, but it has a lot of engineering subtlety and complexity that might not be obvious at first. It's all solvable, of course, but sometimes a game development team decides to solve a different set of problems.
Several games do this to make individual levels big, then have multiple levels on top of that. (Lord of the Rings: Return of the King and From Russia With Love, PS2 and XBox, to name two). Assume you're breaking up and entire game into small loadable chunks, and you have a linear system where five chunks are loaded at any given time, two in front of and two behind the player. If there's a basic grunt enemy you fight over and over throughout the game, then you can either put him into a common, always-loaded area of memory, or you can put an extra copy of him into each loadable chunk. Too much of the former and the common area gets to be game-sized and you don't have enough memory left for the chunks. Too much of the latter, and you're wasting memory on multiple copies of resources in memory at once. Now say you're using a grunt widely in the first third and the last third of the game, but not so much in the middle. Breaking the game up into levels lets you put the grunt into a "level common" region instead of in a game common, which is more efficient.
Very few loading screens, actually. But OP is whining about having to wait a few seconds for his game to continue because he's no better than a rat addicted to cocaine hitting a lever over and over, not complaining about the division of the narrative structure of the game.
Um. They stop the game for several seconds at that point, even if they don't render a loading screen. It's levels; it just happens that the exit foyer for one level looks exactly like the entry foyer for the next.
As a game programmer who is currently having to deal with the complexity of memory management in a streaming open-world environment, I'd like to say shut up, I hate you. Or to put it a little more politely, once you take away the known-memory state checkpoints that you reach between levels, you start having to worry about fragmentation of memory, so you start instituting fixed-size memory "slots" for assets, which deals with the fragmentation problem, but then you sometimes aren't optimally using memory, and then the designers start wanting things to follow you through the world, or allowing you to carry things back and forth through the world, so you have to manage memory outside of the slot system as well as within it, so you have the fragmentation problem again, and then you have to sneak into the designer's house late at night and stab him to death with an icicle.
Nyquist doesn't say, but I do: approximately 6dB of SNR per linear sampling bit, so a Redbook CD can do 96dB SNR, whereas I understand that good vinyl is somewhere under 70dB.
I bet you twenty bucks you're still going to have your ass parked in a chair in front of a half-kilowatt computer saying the same damn thing on the same damn slashdot in ten years, not living "off-grid and low-tech". Best part is, if you're living off-grid and low-tech in ten years, you won't be able to reach me to collect on the bet.
You can reproduce any analog effect digitally. The way a simply built tube saturates when overdriven just happens to be nice, and the way a simply built digital amplifier saturates when overdriven just happens to be awful. If you have enough computing power to compute something as simple as y = tanh( x * gain ) instead of y = clip( x * gain ), digital sounds "warm".
Nyquist is an exclusive boundary and a theoretical limit -- at 44100Hz, you can reproduce any signal up to and not including 22050Hz for the reason you describe. However, for any frequency less than that, the amplitude of the signal *can* be accurately reconstructed if the reconstruction filter is good enough. What looks, in a naive reconstruction, like frequency modulation due to taking samples at different phases of the source signal when the signal is near Nyquist actually goes away in the filter if the filter is designed properly.
If you don't want me to read your HTML source, I recommend you don't send it over the intertubes to my web browser when I ask for it.
And newspaper articles are always accurate. And people who overreact to apparent security threats never misreport the facts in order to justify their overreactions.
With 4-way SIMD, multiple execution units, and dual core, you could probably get 14 billion ops/sec at maximum, not minimum, yeah.
Strange, on my last cross-country flight I watched "L.A. Story" and two episodes of Lost without coming close to killing the battery, on my first-rev iPod Video. Perhaps you're speaking ex recto?
Guess what? Automatic email check defaults to off, roaming or not, on the iPhone.
Yes, the phone still rings if it's sleeping. If you're expecting something else, then you're thinking about four power modes: Off, Standby (non-functional but able to instantly wake up), Sleep (black screen, not accepting touch screen input, but receiving phone calls), and On. This would be even more confusing than the apparently-too-confusing three modes the iPhone has.
There is no line-and-circle icon anywhere on an iPhone. But hey, don't let the facts get in the way of your righteous anger!
And the call was coming from INSIDE THE AIRPLANE! *shriek*
What an ass. Who spends $600 on a phone without checking to see if it does what you consider "crucial"?
How much is the contribution of gravitation (weight->pressure->heat) to geothermal activity? I would have guessed it exceeded that of radioactivity.
Development is not a sunk fixed cost except to a very very crude approximation. It increases roughly linearly with time-on-market. Bug discovery and feature demands are going to come in proportional to roughly log(number-of-users); Apple now has a bunch of fulltime permanent employees working on improving iPhone.
Where's the monopoly, FFS? Every third comment on any iPhone-related post is telling me how there are a dozen other better cheaper touch screen cell phone PDAs available.