Okay, I was partially joking, so I can understand your points. Please try to understand mine: this game shouldn't require this much memory to run - unless, of course, they're using some fantastic new form of rendering.
Usually High Res texture are at least around 512x512 at 32 bits, so around 8 megs per tetxures, a bit higher than 16k.
Um, 512 x 512 = 262144 * 4 bytes/pixel = 1048576 ~ 1MB per texture.
Okay, so that's just off by a factor of 8. But then you say:
That's why videocards have 256/512mb ram now alone, mostly for the framebuffers and textures.
Which just contradicts your earlier point. If I've got 256MB on the video card, that's ~ 2 framebuffers and 200 textures on the card, not in main memory.
Having developed FPS games, I'm kind of skeptical of your 512x512 texture claim. Granted, you'd use these for large objects, but generally speaking, you want to minimize the size of tiles because of the problem with memory bandwidth: if you've only got 32MB per frame, that's 32 1MB textures you can render per frame.
There are different algorithms for rendering 3d scenes, and I've found that the difference in time and memory requirements between them can vary by an order of magnitude. The 384MB requirement makes it look as if they've chosen a less-optimized, more generalized approach for the sake of developing the game on time, rather than producing the most efficient rendering. (Insert Duke Nukem forever joke here...)
It was written in Java, right? Then the breakdown would be:
Windows XP : uses 256 of your 384, leaving 128 MB.
Java VM : Uses 96 MB of your remaining 128, leaving 32.
Actual DOOM graphics, maps, tiles and program code: 32 MB.
There, that sounds about right.
Why they think they need 384 MB and a 1.5 GHz Pentium III is beyond me. Only the most inefficient of rendering techniques would require this kind of horsepower - heck, Descent only needed 586k! Maybe they're rendering scenes by voxel, rather than using polygon or wireframe models...
Here's what I would expect:
For 32 bit color rendered at 1600x1200 you would need a frame buffer of 7.68MB (approx)
You would probably want two, or possibly three framebuffers. 24MB so far.
Each 64x64 bitmapped tile would take up 16k. If you had even 2,000 of these, you'd still only use up an additional 32MB.
So, if you use 8MB for your maps, code, and data structures, the combined total is still only 64 MB.
What are they doing that takes so much memory?
Even with an 800MHz FSB, rendering 100 fps leaves only enough time to access 32 MB per frame. Why would they need 384 MB?
It is conceivable that Microsoft might start going after FOSS authors for patent infringement, but...
It is likewise conceivable that once they start doing this, every FOSS author could sue Microsoft as well. Remember the EOLAs case? $500 million from one patent.
The pendulum swings both ways. The thousands of FOSS developers could, regardless of actual merit, file so many lawsuits that the cost of MS defending their patents would be more trouble than its worth.
This is what is more likely to happen:
Microsoft sues FOSS developer for patent infringement, spending $500k in the process.
MS wins, gets the $12.57 in developer's bank acount. Developer files bankruptcy.
Someone else takes said developer's source code, modifies it a little, and re-releases it under a different name.
MS sues that developer, spends another $500k.
That developer declares bankruptcy...
Another developer comes along, takes some of the source code.... Repeat ad infinitum.
As Dan Lancaster's case against patents points out, patent enforcement is a farce. Yes, it does work when the litigants are large companies, but, more often than not, patent infringement suits are a waste of money. The plaintiff almost always pays as much or more than the patent is actually worth to win the lawsuit, and each lawsuit only stops that particular defendant. Dan points out how easy it is for a patent-infringing CEO to be sued into bankruptcy, only to turn around and start another company doing the exact same thing - after which, the lawsuit process must be started all over again!
He also points out how easy it is to get a patent declared invalid. If anyone can find any prior art, or even show that the invention was "obvious", there goes the plaintiff's case.
And Microsoft, I believe, won't pursue many cases against FOSS software. The reason is simple: they understand that the value of their product is derived primarily from the fact that it can connect to the internet, and that the internet is built primarily upon open standards. If a FOSS developers were to create and patent a better internet protocol - one which required a GPL-like license to use - they could effectively shut Microsoft out of the market simply by introducing an incompatibility between UNIX and Windows machines. In much the same way that Microsoft engineered incompatibilities into Windows so that Lotus wouldn't run, the UNIX world could shut Microsoft out of the market by refusing a connection from a Windows machine.
And the real rub - we could do it by creating a "secure" TCP/IP protocol - one that required a cryptographic handshake generated by a patented algorithm. Where would Microsoft be if using the internet meant that you had to run a Linux machine? (or FreeBSD, etc...)
Why pay....
on
TMBG on DRM
·
· Score: 3, Insightful
Why would I pay for music I can freely download from the internet when I'm going to be treated like a criminal either way? At least in the latter case there's no pretense of legitimacy on either side, and I might save myself some cash. If I buy a DRM crippled CD or MP3, I'm basically supporting the record companies' assertion that music fans are criminals.
Here's a hint: As long as the RIAA views music fans as parasites, they'll never offer them anything of true value. The problem isn't DRM; the problem is that the RIAA has an adversarial attitude toward the public which engenders a spirit of retribution among music fans. After having seen themselves and their favorite bands treated like dirt by the record companies, it's easy for the average fan to justify downloading against the RIAA's wishes. Professional musicianship has now become a con game between the RIAA, the band, and the fans. The fans love the bands, the bands love the fans, and the RIAA hates them both. Is it any wonder people turn a blind eye toward illegal downloading?
Yes, but the idea behind using interpreted code is that you are willing to sacrifice some performance loss for the sake of portability. If you use P/Invoke, you might as well compile to a native executable, because your code won't be portable across differing platforms. And Mono is even worse than C/C++ in this regard - even though #ifdefs are ugly, at least I can compile the same source tree on different platforms. Without the #ifdefs, a Mono programmer has to split source trees if he uses even a single P/Invoke. Worse, his code will still run slower than the C++ equivalent.
And, incidentally, this is one of the reasons why I think.Net was such a bad idea. Why would I bother to accept the performance hit of a virtual machine if my code would only run on Windows? I'm better off using a native executable, because, at least in that case my users don't have to install the.Net runtime (which would require them accepting any EULA, no matter how onerous it may be).
Regarding #2, I am willing to learn new technologies, but I don't see the point if it doesn't do anything for me.
In the beginning, programmers wrote in machine code. But using toggle switches and numbers was difficult and error prone, so assembly language was born.
Assembly language was good, but it was still hard to read. Many common structures, such as loops, could not be visually parsed. Hence, C was born.
Because C compilers would accept inline assembly, those who switched to C could simply wrap their assembly libraries in C functions, and continue to use them. C was good, but the paradigm of structures and functions had its limitations. Hence C++ was born.
C++ compilers, likewise, would compile C code. So switching to C++ didn't mean throwing out your C function libraries. C++ added the class, and later, templates and exceptions.
C++ memory management was initially error-prone. Though later additions such as a the STL and garbage collection libraries addressed this, some programmers still thought it would be a good idea to build a language with automatic garbage collection - they called it Mono. But instead of maintaining backward compatibility with C++, they threw out some of its most useful features, such as templates and overloaded operators. So a programmer can't switch to Mono without taking a substantial loss, and, as history shows, the dominant players in the market are those who are backward compatible.
I agree with your defense of Perl. If anything, you've convinced me that I'd be better off learning Perl than Mono.
So instead of targetting the Windows platform, or the Linux platform, or the Java platform, I can now target the Mono platform, which has *GASP* a regular expression library.
Excuse me if I'm not a little underwhelmed by this. We're supposed to get excited about Mono because of the libraries? As if a good C or C++ coder couldn't write a regular expression parser in less than a week?
But no, I'm converting to Mono because (pick one or more):
I hate Microsoft. But I like their software. So, instead, I'll use a cheap knockoff of.NET, and won't pay Microsoft any money. And I'll never admit that they had a good idea with.NET, even though I'm using what amounts to a copy of it. Instead, I'll call them all sorts of names and say that.NET sucks, just because it's made by Microsoft.
I love to reinvent the wheel with every new project. I'm going to throw out all of those C and C++ function libraries and classes I've built so I can now reimplement them in Mono.
I hate fast code! How am I supposed to write slow code in C++ when my users have a 3 GHz Pentium 4s? Oh, I know, I'll use a virtual machine!
I love to learn. In fact, now that I've mastered C and C++, I'm going to throw away all of my effort invested in these languages and learn a new one.
I hate to write clean code. All my programs have memory leaks because I could never be bothered to actually follow good design principles, nor keep track of the memory I'm using. So instead, I'll just switch langauges, where I can continue to write sloppy code.
I love to file bug reports. There's nothing like using a new technology before it is fully debugged and tested. In addition, I'll complain that the garbage collector slows my program down too much. Granted, if I understood how it worked, I could write code that would minimize the work it had to do. But that would require effort, and I'd rather complain about than clean up my coding style.
I really don't see the big deal here, folks. The fact that the binaries are portable without recompile isn't going to make up for the fact that C++ is a mature, very powerful, fast, portable language.
I can't write kernel drivers in Mono. Only apps.
I couldn't write an OS in Mono, but I could in C++.
I can target Macs, Linux, and Windows now with C++, but I'll have to wait if I want to use Mono.
Mono is not sufficiently different in terms of ease of programming to justify the switch from C++. In fact, as the example code demonstrated, Mono apps often involve more lines of code simply because every single detail of the GUI can, and must be, set by the programmer. In Windows, 9 times out of 10 I can simply use the defaults and everything comes out fine.
Okay, if you've got counterexamples of the points above, I'd like to hear them. But save it if you're going to flame me for broiling your sacred cow. I'd rather see a few good reasons to switch from C++ than a flame war, and right now, Mono isn't making a compelling case. To me, it seems like its Yet Another Algebraic Language With Libraries.
We will not allow these terrorists to change our way of life.
Exactly. Why allow terrorists to change our way of life when Congress gets paid to do it? Get those lazy legislators off their asses - they're on the public payroll and they've got work to do!
They clearly don't own a "right to view" the content, because if they did, they wouldn't need the copyright holder's permission to copy the DVD content to another format or device.
They clearly don't own the media, in the sense of true ownership. Yes, they possess the media, but there are restrictions on what they may do with the media. Ownership implies the exclusive right of control of the thing owned (excepting explicitly shared ownership). Since the possessor of a DVD may not do "anything and everything" with the DVD, they clearly do not own it - neither by common definition (which the consumer is most likely to believe), nor by the full sense of the word (which lawyers like to believe).
What gets me is the whole "license" thing - apparently, the MPAA believes they can impose arbitrary terms on the consumer after the sale. If this holds legal precedent, what prevents me from sending them a check for say, $10, and after they cash it claim that I now own the company? After all, if any arbitrary claim is legally allowable, any claim made whatsoever is actionable...
...to permit legal DVD copying for use on portable devices and...
As if I have to ask for permission to copy something that I own in the first place?
I rent my apartment. I read and signed the lease prior to occupation. I crossed out the parts I didn't agree with, and the landlord accepted the modified lease. I don't pretend that I own my apartment, and the landlord didn't pretend he sold it too me.
But, this DVD thing, is apparently different.
According to the MPAA:
You pay for the cost of content, not the media, when you buy a DVD. So clearly, you don't own the media in the sense that you can do anything you want with it. According the MPAA, you are "licensing" the content, even though you never agreed to any contract, nor were aware of the "license" terms at purchase. Absent case law supporting retroactive compulsory licensing, I'm at a loss as to why the MPAA believes they have the right to do this.
Yet, as anyone who has tried to exchange a damaged or scratched DVD will tell you, the MPAA believes that you don't own the "license" either - if the media becomes unusable, you'll have to "license" the movie again - meaning pay full price for a new copy. So clearly, you don't own the license, either.
When I see the the billboard movie ads says "own it today", I think of actually owning a movie. But after I've shelled out hard cash and pop in the disk, the MPAA informs me that this movie is licensed for home viewing... Wait a minute? - I thought I was buying the DVD, as in, I NOW OWN THE MOVIE. How can the MPAA impose terms on the use of something they no longer own?
What it comes down to is this: If the MPAA can impose terms on me after I've bought something, I don't really own it. And why would I buy something I can't own?
The communists didn't believe in private ownership either. Given Hollywood's leftist leanings, the MPAA's attempt to erode private ownership of goods comes as no surprise.
I'll think about buying a DVD when the MPAA can tell me exactly what, if anything, I own after the purchase
Basically, there are multiple execution units on the die - each one specialized for a particular type of instruction. Consider the following instructions:
mul bx
mov [bx],ax
These instructions must be executed in order because of the fact that the latter depends on the result of the former. But this is not always the case:
mul cx
mov si,@myarray
mov bx,[index]
mov [si + bx],ax
cmp ax,0
jne next
In C, this would be something like: myarray[index] = a * b; In this case, the first three instructions could be executed simultaneously in parallel:
Clock Cycles:
All six instructions can be lined up in a queue. Additional following instructions may also be loaded - provided there's space in the queue. Since the first three do not have dependent operands, they can be issued immediately: the 'mul cx', 'mov si,@myarray', and 'mov bx,[index]' are executed in parallel. Furthermore, the last instruction, a conditional jump, can now be issued. Rather than actually taking the jump, what will happen is that the instruction at the jump address will be fetched and decode started.
The fourth instruction requires the addition of bx and si to form a destination address. This is done now, even though the 'mul cx' instruction may still be executing. The instruction queue can now be filled with the next set of instructions - but since the fifth instruction is a jump to a known offset, there's little point
If the result of 'mul cx' is available now, it will be moved to the address calculated in the previous clock cycle. Also, the 'cmp' instruction can now be executed. Since the execution units that handled the 'mov' instructions are unused at this point, the issue unit could start another mov instruction, absent dependencies.
The issue queue now has a quandry. It has fetched and decoded the instructions following the jump, and may already have issued some to various execution units. If the jump is taken, the issue queue continues as normal, but if not, it must reload the issue queue with the instructions following the jump, re-issue them, and discard the results of any already in-process instructions.
From the start of issue to completion, these six instructions would execute in three clock cycles, as opposed to at least six for sequential processing. Granted, there's an overhead in filling the issue unit, but since this is done as the last instruction is executing, it doesn't affect the number of clock cycles needed to perform the average sequence of instructions. However, the Pentium performed branch-prediction analysis to fill the issue queue - this analysis consists of simply taking the majority of the last four jump instructions; 3 of 4 jumps taken will result in the issue queue starting the instruction after the jump; 1 of 4 will continue with the current instruction stream. Thus, when the Pentium guesses wrong about the result of a jump, the instruction cache must be cleared and reloaded - a substantial penalty. This means that it is possible, but unlikely, for a compiler (or assembly programmer) to order instruction in such a manner that the Pentium actually runs slower than the non-pipelined 486. However, absent intentional opcode abuse, the Pentium will perform 3 instructions every 2 clock cycles. The Pentium II and later can perform 4 instructions every 2 clock cycles.
This does need to be taken with a grain of salt, though.
While compiler optimizations can increase the speed of the translated C or C++ source, a compiler can't (nor should it ever) interpret the intention of the programmer. A programmer using a FOR loop to blank out a 100 character array will cost about 400 clock cycles, as opposed to an assembly programmer using 26 with a REP STOSW.
The assembly programmer can implement algorithms unavailable to the HLL programmer.
Since differing architectures can achieve the same result with widely varying instruction counts, instructions per clock cycle and instructions per second are only useful as comparisons within the same processor family.
Is it just me, or are the terms "all-American" and "un-American" beginning to sound hypocritical and hollow, mostly due to overuse by the beloved administration and media schills?
It's just you. I guess you missed the day in poly-sci where they discussed governmental neologisms and press handling. The typical sound bite is only 7 seconds, and to fit into that time frame, the Government has adopted some official abbreviations:
Patriotic
Supporting the President, no matter how erroneous or immoral the action.
Un-American
Something that the President doesn't like.
Terrorist activity
A more serious form of being Un-American, this occurs when two or more people in a room question US foreign policy.
Terrorism Charges
Not an actual crime per se, but rather, a way of slandering a person's character without telling anyone what, exactly, the accused allegedly did wrong. Terrorism Charges may also indicate the person engaged in un-American activities.
War on Terror
A blanket justification for whatever the Government feels like doing at the moment. If a government action is identified as part of the War on Terror, this implies that criticism of said action would be unpatriotic and un-American.
Terrorist Country
A country which did not support the US invasion of Iraq.
Enemy Combatant
Someone who disagrees with the President. May also be used to denote a citizen of a Terrorist Country. Note that this implies the person is un-American, meaning they could be subject to Terrorism Charges. To qualify as an Enemy Combatant, one merely needs to disagree with US foreign policy and have visited a Terrorist Country in their lifetime. Someone who has never left the US may qualify as an Enemy Combatant if they meet certain ethnic, racial, religious or political criteria, or if they have ever had contact with such individuals.
Known Terrorist
Some the US suspects of planning terrorist activities, but remains at large because the US either cannot locate the person, or because the prosecutor needs more time to gather evidence of terrorist activity.
Links to Terrorism
A Link to Terrorism is established whenever a country, person, or organization comes in contact with a Known Terrorist. The contact need not be intentional to qualify.
With these definitions, you should be able to make sense of any government press release. In the odd circumstance when these terms aren't used, you should suspect the Government of telling the truth. Next time, I'll go into Conspiracy-speak.
Client: I'm having trouble with my Storage-Tek....
Vendor: Okay, bring it in...
A few hours pass.
Vendor: Um, we've got kind of a problem here...
Client: What is it?
Vendor: Well, it turns out that the DMCA prohibits us from accessing or repairing your hardware.
Client: So, what exactly does that mean?
Vendor: Well, per contract, you agreed that submitting a DMCA-protected device for service would render the full value of the contract due immediately. This is a DMCA-protected device, so your full 5-year service contract will be due before we return the device.
Client: Okay, but is it usuable? I mean, can we get our data back.
Vendor: Unfortunately, no. It would be illegal.
Client: So you're screwing us, right?
Vendor: No, you screwed us. You bought hardware that was illegal for us to service. We can't do anything for you without breaking the law.
Client: What do you mean, illegal?! It's our data.
Vendor: Um, yes, it is your data. But you don't own the software which controls access to it - Storage-Tek does. And since this software is restricted by the DMCA, you can't legally access your data without Storage-Tek's permission. Your only option at this point is to return the device to them.
Client: So let me get this straight: You're going to charge us a full five year contract, and you don't even fix the machine?! How can you keep clients like this?
Vendor: Truth is, we can't. That's why you agreed in the contract not to send us DMCA-restricted devices. Since we cannot legally service DMCA restricted hardware, we can't restore your data. And this is why you agreed in the contract to pay off the full value of the contract if you did so - as compensation for the fact that we can no longer do business with you.
Client: So what happens if Storage-Tek goes out of business? What would you do then?
Vendor: It's not what we would do, it is what you would do. You'd probably go out of business because no one would legally be able to service your Storage-Tek machines. The next time they failed, you would irretrievably lose all of your data.
Client: So, how could we avoid this in the future?
Vendor: Simple - buy a machine that isn't DMCA-restricted.
Client: But we obviously didn't know this when we bought these machines...
Vendor: Well, it's not my problem you want to break the law....
If I can develop UNIX apps which will run on Windows, why would I bother developing Windows Apps?
I fear that the reality of the situation will be that Microsoft's SFU will work well enough to sell Longhorn, but not well enough that existing UNIX apps can be seamlessly moved to a Windows box.
Besides, this is a dumb move by Microsoft. Remember what happened to OS/2 when IBM announced that it would run Windows apps? That's right - developers flocked to Windows (3.1) because they new that regardless of OS/2's success, they'd still be able to sell their applications. As a consequence, OS/2 bombed, where Windows took off. Seems to me like Microsoft is shooting themselves in the foot with this one.
And incidentally, I believe the same thing about WINE - I think it hindered the adoption of Linux by developers.
Really, this protects both the broadcasters and the viewing public:
Now a broadcaster has a reasonable defense against the jerks who claim to have seen something they didn't. For example, a real jerk could simply send in porno excerpts with a complaint about a local station, and without a tape of their broadcast, a local station would end up paying a fine for indecency.
In those cases where the broadcaster did transmit something indecent, the material could be reviewed and critically analyzed for exactly what constituted the indecency. Determining exactly what is indecent or not usually happens after the fact - something which doesn't sit too well with broadcasters, who would rather know beforehand. With recorded segments, there would be a large body of evidence of prior incidents, giving the broadcaster a much more precise idea of where the line between merely objectionable and getting-fined-for-indecency content lies.
And, as other posters have mentioned, people seldom record shows in advance, making it difficult, if not impossible, to provide airtime excerpts when filing a complaint. Because of this difficulty, and the tendency for exaggeration, I suspect that many indecency complaints are indeed frivolous, and having the airtime segment recorded gives the broadcaster a solid defense in cases of frivolous complaints.
In the rare even that something indecent is broadcast, this law prevents the broadcaster from ducking responsibility simply because no one had the foresight (or perhaps the technical ability - broadcast flag?!) to record a show.
Just once I'd like to see Open Source apps that didn't pretend to be a cheap knock-off of someone else's successful product.
Why is it that FS developers insist on following Microsoft's lead?.NET is a stupid idea, and.Mono even worse - why would you build a virtual machine if it would never be ported to other platforms, and why spend effort optimizing said machine when modern compilers can already generate much faster code?
Oh, I know - the challenge. Rather than writing code which actually solves problems, I'd rather reinvent the wheel, learn how to everything in this New! SHINY! Language (tm), just so my programs will run more slowly on fewer platforms.
And.Mono is insane. GCC is a very mature, very good compiler, and it will be a long, long time before any virtual machine can come close to the performance of gcc-generated code. Why would you waste talent on a project which is going to produce something worse than what you've already got? Because it's popular? Because Microsoft is doing it? Think about it - Free Software will never be a leader as long as they look to an unethical corporation for ideas. Microsoft implements ideas, but it doesn't create them.
Well, yes, functionality wins hands down. There's little point in installing software you can't really use.
But the whole security through "code auditing" thing is a farce. What does auditing matter when the company itself is willing to steal the code of others and infringe patents at will?
The Timeline case is a perfect example. Timeline gave Microsoft their code, and someone at Microsoft decided to ship it with SQL server. It's not as if it was an accident (and if it was, there goes your auditing argument). Rather, someone at Microsoft thought they could "bend" the rules, and shipped Timeline's code. When Timeline pressed Microsoft about the issue, Microsoft claimed that their license gave them the right to redistribute code; Timeline thought they were selling them a license for internal use.
You have to admire the slick way Microsoft pulled this off - they duped Timeline into giving them their code for a song, claiming they were going to use it internally. But then they shipped it, and now the burden of proof is on Timeline to show damages.
I know that only the stingiest managers would actually care to do this.
The problem is that even if you do shortchange customers, you still don't make any money selling gas. A popular gas station can sell a few hundered to a few thousand gallons a day. They'll make a few pennies per gallon of profit. Even if they earn, say, 9 cents per gallon, that's only 9,000 cents - $90.00 for a full day of sales. Now, suppose the station decides to cheat - they undermeter by 10% - 23.1 ci per gallon. How much would they make? - $99.00, still less than what they've paid out in cashier wages.
Who would risk being charged with fraud to make a paltry $9? The problem is that the pumps are mechanical, and they wear out. Gas is what draws the customers, and gas stations make far more money on auto service and general sales than they do on gasoline. The gasoline is just there to get the customers into the convenience store. And since it doesn't make much money, the managers have a hard time justifying pump maintainence. And that is why the pumps are inaccurate in the vast majority of cases - not because of outright fraud.
And yes, there was a guy who would occasionally come by with a one gallon gas can, pump a gallon of gas, and then tell us that we were shortchanging him because the can wasn't full. The guy didn't know that gas cans are designed with a certain amount of airspace at the top so they didn't overflow when rising temperatures caused the gas to expand... Fortunately, my state does regular tests, and we had our pumps certified. My boss told him to take a hike; he threatened to sue, but apparently someone clued him in before he filed the papers.
Ah, right, but what exactly constitutes a decent education is a matter of considerable debate. If my idea of a decent education is offensive to your morals, do I still have the right to subject your children to it?
Microsoft's customers were sued by Timeline when Microsoft distributed code with SQL server that Timeline had only license to Microsoft for internal use.
I have never heard of a Linux vendor suing their customers. Yet fairly recently, Microsoft sued the LA school district for about $400,000. They sued a school.... Think about that one for a while.
Microsoft was convicted of abusing their monopoly position to hurt competitors. Misgivings about code written by volunteers are understandable, but it would be downright foolish to eschew code written by someone with altruistic motives in favor of code written by a convicted felon.
The idea that a convicted felon's code - code which is intentionally kept secret - is somehow less likely to infringe on IP rights is laughable, at best.
Furthermore, the fact that the source is closed means a company cannot even begin to evaluate the risk of infringement. One simply cannot determine how much or from whom Microsoft has stolen code - or if they've stolen any at all.
One could make some very good objections to using Linux, but liability is not one of them. If anything, the fact that the source code is freely available means that absent frivolous plaintiffs (*cough* SCO *cough*) there is a very small risk of being sued. Unlike the proprietary, closed source model, cases of actual infringement can be mitigated by the end user. If I was sued for IP infringement and didn't have the source code, my only option (assuming that infringement really took place) is to pay royalties and licensing fees. But if I do have the source code, I can simply remove the infringing material, substantially reducing the damages that a plaintiff could collect.
And for all you Microsoft-vs-Linux trolls, save it. Microsoft and Linux are just good examples of the relative strengths and weaknesses of open versus closed source. The argument would apply equally well to Adobe Photoshop vs. Gimp or Oracle vs Postgresql, etc...
Well, given that neither paying taxes nor sending your children are optional activities, I'd say your argument doesn't apply.
But there's something insidious about someone taking money from you (via taxes), and then using that money to undermine your parenting. Those who view children as a nuisance and the school district as a babysitter could care less what happens there. It is the parents who are actually being parents who take greatest offense at their own taxes being used to undermine their parenting efforts. It isn't an issue of free-speech rights, but rather of the rights of a parent to raise their own child in the manner they see fit. I don't tell you how to raise your children; why do you think you have the right to tell me how to raise mine?
but if its a faulty display or a bad CD drive, this might actually work...
Actually, no, it won't. I bought a Toshiba Satellite last year and in spite of an under-warranty replacement of the CD burner, it still can't burn more than about 400 MB to a CD-R or CD-RW.
There's no use in warranty work when the fault is in the product design.
Slashdot Mirror
Okay, I was partially joking, so I can understand your points. Please try to understand mine: this game shouldn't require this much memory to run - unless, of course, they're using some fantastic new form of rendering.
Usually High Res texture are at least around 512x512 at 32 bits, so around 8 megs per tetxures, a bit higher than 16k.
Um, 512 x 512 = 262144 * 4 bytes/pixel = 1048576 ~ 1MB per texture.Okay, so that's just off by a factor of 8. But then you say:
That's why videocards have 256/512mb ram now alone, mostly for the framebuffers and textures.Which just contradicts your earlier point. If I've got 256MB on the video card, that's ~ 2 framebuffers and 200 textures on the card, not in main memory.
Having developed FPS games, I'm kind of skeptical of your 512x512 texture claim. Granted, you'd use these for large objects, but generally speaking, you want to minimize the size of tiles because of the problem with memory bandwidth: if you've only got 32MB per frame, that's 32 1MB textures you can render per frame.
There are different algorithms for rendering 3d scenes, and I've found that the difference in time and memory requirements between them can vary by an order of magnitude. The 384MB requirement makes it look as if they've chosen a less-optimized, more generalized approach for the sake of developing the game on time, rather than producing the most efficient rendering. (Insert Duke Nukem forever joke here...)
Or am I the only one who read the headline as:
Microsoft Expands Access with Windows Source Code...
I was thinking, "OK, as if Access wasn't already bloated enough, they're going to build their OS into it?"
It was written in Java, right? Then the breakdown would be:
- Windows XP : uses 256 of your 384, leaving 128 MB.
- Java VM : Uses 96 MB of your remaining 128, leaving 32.
- Actual DOOM graphics, maps, tiles and program code: 32 MB.
There, that sounds about right.Why they think they need 384 MB and a 1.5 GHz Pentium III is beyond me. Only the most inefficient of rendering techniques would require this kind of horsepower - heck, Descent only needed 586k! Maybe they're rendering scenes by voxel, rather than using polygon or wireframe models...
Here's what I would expect:
- For 32 bit color rendered at 1600x1200 you would need a frame buffer of 7.68MB (approx)
- You would probably want two, or possibly three framebuffers. 24MB so far.
- Each 64x64 bitmapped tile would take up 16k. If you had even 2,000 of these, you'd still only use up an additional 32MB.
So, if you use 8MB for your maps, code, and data structures, the combined total is still only 64 MB.What are they doing that takes so much memory? Even with an 800MHz FSB, rendering 100 fps leaves only enough time to access 32 MB per frame. Why would they need 384 MB?
It is conceivable that Microsoft might start going after FOSS authors for patent infringement, but...
It is likewise conceivable that once they start doing this, every FOSS author could sue Microsoft as well. Remember the EOLAs case? $500 million from one patent.
The pendulum swings both ways. The thousands of FOSS developers could, regardless of actual merit, file so many lawsuits that the cost of MS defending their patents would be more trouble than its worth.
This is what is more likely to happen:
As Dan Lancaster's case against patents points out, patent enforcement is a farce. Yes, it does work when the litigants are large companies, but, more often than not, patent infringement suits are a waste of money. The plaintiff almost always pays as much or more than the patent is actually worth to win the lawsuit, and each lawsuit only stops that particular defendant. Dan points out how easy it is for a patent-infringing CEO to be sued into bankruptcy, only to turn around and start another company doing the exact same thing - after which, the lawsuit process must be started all over again!
He also points out how easy it is to get a patent declared invalid. If anyone can find any prior art, or even show that the invention was "obvious", there goes the plaintiff's case.
And Microsoft, I believe, won't pursue many cases against FOSS software. The reason is simple: they understand that the value of their product is derived primarily from the fact that it can connect to the internet, and that the internet is built primarily upon open standards. If a FOSS developers were to create and patent a better internet protocol - one which required a GPL-like license to use - they could effectively shut Microsoft out of the market simply by introducing an incompatibility between UNIX and Windows machines . In much the same way that Microsoft engineered incompatibilities into Windows so that Lotus wouldn't run, the UNIX world could shut Microsoft out of the market by refusing a connection from a Windows machine.
And the real rub - we could do it by creating a "secure" TCP/IP protocol - one that required a cryptographic handshake generated by a patented algorithm. Where would Microsoft be if using the internet meant that you had to run a Linux machine? (or FreeBSD, etc...)
Why would I pay for music I can freely download from the internet when I'm going to be treated like a criminal either way? At least in the latter case there's no pretense of legitimacy on either side, and I might save myself some cash. If I buy a DRM crippled CD or MP3, I'm basically supporting the record companies' assertion that music fans are criminals.
Here's a hint: As long as the RIAA views music fans as parasites, they'll never offer them anything of true value. The problem isn't DRM; the problem is that the RIAA has an adversarial attitude toward the public which engenders a spirit of retribution among music fans. After having seen themselves and their favorite bands treated like dirt by the record companies, it's easy for the average fan to justify downloading against the RIAA's wishes. Professional musicianship has now become a con game between the RIAA, the band, and the fans. The fans love the bands, the bands love the fans, and the RIAA hates them both. Is it any wonder people turn a blind eye toward illegal downloading?
Yes, but the idea behind using interpreted code is that you are willing to sacrifice some performance loss for the sake of portability. If you use P/Invoke, you might as well compile to a native executable, because your code won't be portable across differing platforms. And Mono is even worse than C/C++ in this regard - even though #ifdefs are ugly, at least I can compile the same source tree on different platforms. Without the #ifdefs, a Mono programmer has to split source trees if he uses even a single P/Invoke. Worse, his code will still run slower than the C++ equivalent.
And, incidentally, this is one of the reasons why I think .Net was such a bad idea. Why would I bother to accept the performance hit of a virtual machine if my code would only run on Windows? I'm better off using a native executable, because, at least in that case my users don't have to install the .Net runtime (which would require them accepting any EULA, no matter how onerous it may be).
Regarding #2, I am willing to learn new technologies, but I don't see the point if it doesn't do anything for me.
So instead of targetting the Windows platform, or the Linux platform, or the Java platform, I can now target the Mono platform, which has *GASP* a regular expression library.
Excuse me if I'm not a little underwhelmed by this. We're supposed to get excited about Mono because of the libraries? As if a good C or C++ coder couldn't write a regular expression parser in less than a week?
But no, I'm converting to Mono because (pick one or more):
I really don't see the big deal here, folks. The fact that the binaries are portable without recompile isn't going to make up for the fact that C++ is a mature, very powerful, fast, portable language.
Okay, if you've got counterexamples of the points above, I'd like to hear them. But save it if you're going to flame me for broiling your sacred cow. I'd rather see a few good reasons to switch from C++ than a flame war, and right now, Mono isn't making a compelling case. To me, it seems like its Yet Another Algebraic Language With Libraries.
Exactly. Why allow terrorists to change our way of life when Congress gets paid to do it? Get those lazy legislators off their asses - they're on the public payroll and they've got work to do!
What gets me is the whole "license" thing - apparently, the MPAA believes they can impose arbitrary terms on the consumer after the sale. If this holds legal precedent, what prevents me from sending them a check for say, $10, and after they cash it claim that I now own the company? After all, if any arbitrary claim is legally allowable, any claim made whatsoever is actionable...
As if I have to ask for permission to copy something that I own in the first place?
I rent my apartment. I read and signed the lease prior to occupation. I crossed out the parts I didn't agree with, and the landlord accepted the modified lease. I don't pretend that I own my apartment, and the landlord didn't pretend he sold it too me.
But, this DVD thing, is apparently different. According to the MPAA:
When I see the the billboard movie ads says "own it today", I think of actually owning a movie. But after I've shelled out hard cash and pop in the disk, the MPAA informs me that this movie is licensed for home viewing... Wait a minute? - I thought I was buying the DVD, as in, I NOW OWN THE MOVIE. How can the MPAA impose terms on the use of something they no longer own?
What it comes down to is this: If the MPAA can impose terms on me after I've bought something, I don't really own it. And why would I buy something I can't own?
The communists didn't believe in private ownership either. Given Hollywood's leftist leanings, the MPAA's attempt to erode private ownership of goods comes as no surprise.
I'll think about buying a DVD when the MPAA can tell me exactly what, if anything, I own after the purchase
For the worms which will detect and disable AV software....
What will the Windows community do when a virus disables the AV software and prevents it from loading?
It's actually simpler than it sounds.
Basically, there are multiple execution units on the die - each one specialized for a particular type of instruction. Consider the following instructions:
mul bx
mov [bx],ax
These instructions must be executed in order because of the fact that the latter depends on the result of the former. But this is not always the case:
mul cx
In C, this would be something like: myarray[index] = a * b; In this case, the first three instructions could be executed simultaneously in parallel: Clock Cycles:mov si,@myarray
mov bx,[index]
mov [si + bx],ax
cmp ax,0
jne next
From the start of issue to completion, these six instructions would execute in three clock cycles, as opposed to at least six for sequential processing. Granted, there's an overhead in filling the issue unit, but since this is done as the last instruction is executing, it doesn't affect the number of clock cycles needed to perform the average sequence of instructions. However, the Pentium performed branch-prediction analysis to fill the issue queue - this analysis consists of simply taking the majority of the last four jump instructions; 3 of 4 jumps taken will result in the issue queue starting the instruction after the jump; 1 of 4 will continue with the current instruction stream. Thus, when the Pentium guesses wrong about the result of a jump, the instruction cache must be cleared and reloaded - a substantial penalty. This means that it is possible, but unlikely, for a compiler (or assembly programmer) to order instruction in such a manner that the Pentium actually runs slower than the non-pipelined 486. However, absent intentional opcode abuse, the Pentium will perform 3 instructions every 2 clock cycles. The Pentium II and later can perform 4 instructions every 2 clock cycles.
This does need to be taken with a grain of salt, though.
It's just you. I guess you missed the day in poly-sci where they discussed governmental neologisms and press handling. The typical sound bite is only 7 seconds, and to fit into that time frame, the Government has adopted some official abbreviations:
Patriotic Supporting the President, no matter how erroneous or immoral the action. Un-American Something that the President doesn't like. Terrorist activity A more serious form of being Un-American, this occurs when two or more people in a room question US foreign policy. Terrorism Charges Not an actual crime per se, but rather, a way of slandering a person's character without telling anyone what, exactly, the accused allegedly did wrong. Terrorism Charges may also indicate the person engaged in un-American activities. War on Terror A blanket justification for whatever the Government feels like doing at the moment. If a government action is identified as part of the War on Terror, this implies that criticism of said action would be unpatriotic and un-American. Terrorist Country A country which did not support the US invasion of Iraq. Enemy Combatant Someone who disagrees with the President. May also be used to denote a citizen of a Terrorist Country. Note that this implies the person is un-American, meaning they could be subject to Terrorism Charges. To qualify as an Enemy Combatant, one merely needs to disagree with US foreign policy and have visited a Terrorist Country in their lifetime. Someone who has never left the US may qualify as an Enemy Combatant if they meet certain ethnic, racial, religious or political criteria, or if they have ever had contact with such individuals. Known Terrorist Some the US suspects of planning terrorist activities, but remains at large because the US either cannot locate the person, or because the prosecutor needs more time to gather evidence of terrorist activity. Links to Terrorism A Link to Terrorism is established whenever a country, person, or organization comes in contact with a Known Terrorist. The contact need not be intentional to qualify.With these definitions, you should be able to make sense of any government press release. In the odd circumstance when these terms aren't used, you should suspect the Government of telling the truth. Next time, I'll go into Conspiracy-speak.
Client: I'm having trouble with my Storage-Tek....
Vendor: Okay, bring it in...
A few hours pass.
Vendor: Um, we've got kind of a problem here...
Client: What is it?
Vendor: Well, it turns out that the DMCA prohibits us from accessing or repairing your hardware.
Client: So, what exactly does that mean?
Vendor: Well, per contract, you agreed that submitting a DMCA-protected device for service would render the full value of the contract due immediately. This is a DMCA-protected device, so your full 5-year service contract will be due before we return the device.
Client: Okay, but is it usuable? I mean, can we get our data back.
Vendor: Unfortunately, no. It would be illegal.
Client: So you're screwing us, right?
Vendor: No, you screwed us. You bought hardware that was illegal for us to service. We can't do anything for you without breaking the law.
Client: What do you mean, illegal?! It's our data.
Vendor: Um, yes, it is your data. But you don't own the software which controls access to it - Storage-Tek does. And since this software is restricted by the DMCA, you can't legally access your data without Storage-Tek's permission. Your only option at this point is to return the device to them.
Client: So let me get this straight: You're going to charge us a full five year contract, and you don't even fix the machine?! How can you keep clients like this?
Vendor: Truth is, we can't. That's why you agreed in the contract not to send us DMCA-restricted devices. Since we cannot legally service DMCA restricted hardware, we can't restore your data. And this is why you agreed in the contract to pay off the full value of the contract if you did so - as compensation for the fact that we can no longer do business with you.
Client: So what happens if Storage-Tek goes out of business? What would you do then?
Vendor: It's not what we would do, it is what you would do. You'd probably go out of business because no one would legally be able to service your Storage-Tek machines. The next time they failed, you would irretrievably lose all of your data.
Client: So, how could we avoid this in the future?
Vendor: Simple - buy a machine that isn't DMCA-restricted.
Client: But we obviously didn't know this when we bought these machines...
Vendor: Well, it's not my problem you want to break the law....
If I can develop UNIX apps which will run on Windows, why would I bother developing Windows Apps?
I fear that the reality of the situation will be that Microsoft's SFU will work well enough to sell Longhorn, but not well enough that existing UNIX apps can be seamlessly moved to a Windows box.
Besides, this is a dumb move by Microsoft. Remember what happened to OS/2 when IBM announced that it would run Windows apps? That's right - developers flocked to Windows (3.1) because they new that regardless of OS/2's success, they'd still be able to sell their applications. As a consequence, OS/2 bombed, where Windows took off. Seems to me like Microsoft is shooting themselves in the foot with this one.
And incidentally, I believe the same thing about WINE - I think it hindered the adoption of Linux by developers.
Really, this protects both the broadcasters and the viewing public:
Just once I'd like to see Open Source apps that didn't pretend to be a cheap knock-off of someone else's successful product.
Why is it that FS developers insist on following Microsoft's lead? .NET is a stupid idea, and .Mono even worse - why would you build a virtual machine if it would never be ported to other platforms, and why spend effort optimizing said machine when modern compilers can already generate much faster code?
Oh, I know - the challenge. Rather than writing code which actually solves problems, I'd rather reinvent the wheel, learn how to everything in this New! SHINY! Language (tm), just so my programs will run more slowly on fewer platforms.
And .Mono is insane. GCC is a very mature, very good compiler, and it will be a long, long time before any virtual machine can come close to the performance of gcc-generated code. Why would you waste talent on a project which is going to produce something worse than what you've already got? Because it's popular? Because Microsoft is doing it? Think about it - Free Software will never be a leader as long as they look to an unethical corporation for ideas. Microsoft implements ideas, but it doesn't create them.
Okay, flame away.
Well, yes, functionality wins hands down. There's little point in installing software you can't really use.
But the whole security through "code auditing" thing is a farce. What does auditing matter when the company itself is willing to steal the code of others and infringe patents at will?
The Timeline case is a perfect example. Timeline gave Microsoft their code, and someone at Microsoft decided to ship it with SQL server. It's not as if it was an accident (and if it was, there goes your auditing argument). Rather, someone at Microsoft thought they could "bend" the rules, and shipped Timeline's code. When Timeline pressed Microsoft about the issue, Microsoft claimed that their license gave them the right to redistribute code; Timeline thought they were selling them a license for internal use.
You have to admire the slick way Microsoft pulled this off - they duped Timeline into giving them their code for a song, claiming they were going to use it internally. But then they shipped it, and now the burden of proof is on Timeline to show damages.
I know that only the stingiest managers would actually care to do this.
The problem is that even if you do shortchange customers, you still don't make any money selling gas. A popular gas station can sell a few hundered to a few thousand gallons a day. They'll make a few pennies per gallon of profit. Even if they earn, say, 9 cents per gallon, that's only 9,000 cents - $90.00 for a full day of sales. Now, suppose the station decides to cheat - they undermeter by 10% - 23.1 ci per gallon. How much would they make? - $99.00, still less than what they've paid out in cashier wages.
Who would risk being charged with fraud to make a paltry $9? The problem is that the pumps are mechanical, and they wear out. Gas is what draws the customers, and gas stations make far more money on auto service and general sales than they do on gasoline. The gasoline is just there to get the customers into the convenience store. And since it doesn't make much money, the managers have a hard time justifying pump maintainence. And that is why the pumps are inaccurate in the vast majority of cases - not because of outright fraud.
And yes, there was a guy who would occasionally come by with a one gallon gas can, pump a gallon of gas, and then tell us that we were shortchanging him because the can wasn't full. The guy didn't know that gas cans are designed with a certain amount of airspace at the top so they didn't overflow when rising temperatures caused the gas to expand... Fortunately, my state does regular tests, and we had our pumps certified. My boss told him to take a hike; he threatened to sue, but apparently someone clued him in before he filed the papers.
Ah, right, but what exactly constitutes a decent education is a matter of considerable debate. If my idea of a decent education is offensive to your morals, do I still have the right to subject your children to it?
For your consideration:
One could make some very good objections to using Linux, but liability is not one of them. If anything, the fact that the source code is freely available means that absent frivolous plaintiffs (*cough* SCO *cough*) there is a very small risk of being sued. Unlike the proprietary, closed source model, cases of actual infringement can be mitigated by the end user. If I was sued for IP infringement and didn't have the source code, my only option (assuming that infringement really took place) is to pay royalties and licensing fees. But if I do have the source code, I can simply remove the infringing material, substantially reducing the damages that a plaintiff could collect.
And for all you Microsoft-vs-Linux trolls, save it. Microsoft and Linux are just good examples of the relative strengths and weaknesses of open versus closed source. The argument would apply equally well to Adobe Photoshop vs. Gimp or Oracle vs Postgresql, etc...
Well, given that neither paying taxes nor sending your children are optional activities, I'd say your argument doesn't apply.
But there's something insidious about someone taking money from you (via taxes), and then using that money to undermine your parenting. Those who view children as a nuisance and the school district as a babysitter could care less what happens there. It is the parents who are actually being parents who take greatest offense at their own taxes being used to undermine their parenting efforts. It isn't an issue of free-speech rights, but rather of the rights of a parent to raise their own child in the manner they see fit. I don't tell you how to raise your children; why do you think you have the right to tell me how to raise mine?
Actually, no, it won't. I bought a Toshiba Satellite last year and in spite of an under-warranty replacement of the CD burner, it still can't burn more than about 400 MB to a CD-R or CD-RW.
There's no use in warranty work when the fault is in the product design.
I feel like starting an ISP and offering free email accounts to congressmen, judges, FBI agents, etc...
The time difference between an embarrassing email leak and legislation outlawing reading another's email is left as an exercise for the reader....