Yeah, his conclusions were also a bit too simple when he got to the size factor:
However, the difference wasn't all that huge, only around 16% to 54% larger for 64-bit than 32-bit. Unless the system was an embedded system with very limited storage space, I can't see this being all that much of a negative factor.
Withholding important facts is just another way of lying.
Sivaram_Velauthapill's post is quite excellent. The point is, of course no one should really care if that moderator takes some weeks of vacation and telecommutes. But why not be honest about it? What did he have to hide? OK, it wouldn't sound so good if he said "Brr, it's so cold in Boston... thank God I'm in Florida right now... haha, you losers!" But I think now that people found out another way, it sounds even worse. And he's lost quite some credibility there.
Saying honestly that he's not there where he seems to be is one bit of information that I think the public should have a right to know. And it wouldn't have been such a big deal. I mean, it's not like he reports from Florida all year (that would change the case a bit).
So you're saying that if I publish a Java program with code stolen from your GPLed program, I'm not violating the license because I "effectively include the source code." Firstly, that is BS, as others have pointed out already in this thread; secondly, the point is that my program is only valid if I still put it under the GPL. If I don't do that, I can redistrubite the source code as much as I want to---it's still a violation. I am not permitted to take your GPL code, put my name on it, change the license to, say, BSD or even Public Domain, without violating your rights to the original code.
Probably neither. My Canon PowerShot A70 shoots 2048x1536 = 3,145,728 pixels, and claims to have 3.2 megapixels. The specs say it has 3.34 million photo detectors, so it seems like Canon took the 'average' of these two numbers... much like for the A60 which claims 2.0 megapixels, and has in fact 1.92 million vs. 2.11 million.
So, aside from rounding for marketing purposes, it seems like the 'mega' is decimal (MPixels, not MiPixels).
Hm. Like KISS would ever publicize their CVS code, if it exists... sounds not much more likely than SCO opening up their code. The only proper way for MPlayer to prove their innocence (if they ever had to) would be to find the oldest copy of MPlayer code (hey, their CVS is open...) containing the code in question and to hope that it's older than KISS's code, or even older than KISS themselves. I would guess that's not too unlikely...
Probably not only that. Let's suppose that Java's trig functions are indeed native. The problem is, they're still functions which are called as standard class methods. In C, the trig functions themselves are compiled and thus of course native, but the complete overhead for making the function calls goes away, because ideally the compiler will even in-line the resulting code. I don't think this is done in Java.
This would also explain why the standard math operations (+,-,*,/) are indeed fast -- they're mapped directly onto processor instructions. The trig functions, however, are functions, and are apparently not optimized like one would optimize such a function in a C/C++ compiler.
The article was actually written by Christopher W. Cowell-Shah. Also, as others noted, except for the I/O benchmark (where gcc was actually faster) the benchmarks couldn't profit much from the OS around it since they were just computationally intensive, but didn't make much use of OS-specific library functions.
Hm, I don't really think that it has much to do with the fact that it runs on Windows. Except for maybe the I/O bound benchmarks, the others, couldn't be expected to gain much from being 'optimized for Windows'. And gcc was actually better on I/O. VC++ wins on the math benchmarks -- so MS may just have the better set of Intel optimizations.
Besides, it would've been kinda difficult comparing the same set of compilers under native Linux, don't you think?
Actually, it's more of a surprise that Java performed/equally well/ in all benchmarks -- except the trig benchmark. The dismal performance in the trig benchmark is the second, much bigger surprise. It would be good to see the source code used to find out whether the code is BS or Java actually totally fails to produce efficient trig code.
Hmm, interesting. Someone should really do this. Make it a 2- or 4-processor Efficeon PC (8 would be a bit too many IMO) with accompanying energy-saving hardware around it, put it in a compact box, equip it with a well pre-configured Linux, and it might actually sell pretty well (OK, only among geeks, with Linux not being mass-market ready...). That would be one nice box. Low power consumption, proper performance, no fan noise while Transmeta's Thermal Extensions are active... I might buy something like that.
Well, at least high-power, high-heat, high-speed. Have you ever seen a computer magazine publishing a serious article about 'efficient computers' or even 'portable, energy-saving laptops'? I haven't. Even when it comes to laptops, all you see is Tom's-Hardware-style performance benchmarks, CPU performance, graphics performance, etc. This really pisses me off. Only gamers (and a few others) really need that much performance, in my opinion. Most Joe Average users are just caught by ads and 'reviews' of high-performance systems... "you really need to have this graphics card... well, it requires a 1000W power supply, but you won't need heating in your house anymore... you really need it for your word processor, spread sheet, and net surfing."
I, for one, am very happy with my LifeBook P-2120. It has everything I need (ok, maybe for some people it's missing Bluetooth), it's fast enough to watch DVDs and DivX, with all batteries installed it runs 9-10 hours, and it's still extremely compact and light-weight, and it runs Linux alright. What more could I need?
Transmeta has found a good niche. But it deserves much more attention.
...to answer this survey would be, "we use Linux because it is Free as in Speech." Any other answer is just a cheap way of giving MS helpful tips on how to make more money. Too bad the survey does not really allow this kind of answer. But, I think, if everybody participating said that (GNU/)Linux is the single choice because it is Free, we could leave a good impression... at least I suppose RMS would agree.
So, if it is possible in any way, fill out the survey saying, "none of the given reasons are why I/really/ use GNU/Linux (although, of course, I get all these reasons as a side effect). I use it because it is Free." I wonder what MS would do if everybody answered the survey like that?
Nice idea, but keep in mind that THC can be traced at other places in your body than just in your blood. Hair comes to mind, for instance. So for now, this technology, if it ever becomes practical, won't save you from the consequences of a drug test...
Hm, well at least in Europe (especially Germany), ISDN is very widely used . The infrastructure is there, everybody can have an ISDN line set up. And many people do, especially businesses are running on ISDN. OK, the bandwidth thing was of course nonsense; but the best-sold solution in Germany right now is ISDN+DSL.
Slashdot Mirror
Somehow, the word compromise looks wrong in this place... but maybe it describes the security level appropriately? :)
Where's QuantumFTL when you need him? Too bad he doesn't have screenshots on his homepage. At least he's on Slashdot. Somebody go wake him up. :)
Hmm, well, she's a Queen, right?
:)
Yeah, his conclusions were also a bit too simple when he got to the size factor:
Two words: cache footprint.
AntiVir (free-as-in-beer version here) detects it too. Their removal tool also takes care of it.
You mean, besides this, this, or this?
I have no idea, but have you checked the website? :)
Withholding important facts is just another way of lying.
Sivaram_Velauthapill's post is quite excellent. The point is, of course no one should really care if that moderator takes some weeks of vacation and telecommutes. But why not be honest about it? What did he have to hide? OK, it wouldn't sound so good if he said "Brr, it's so cold in Boston... thank God I'm in Florida right now... haha, you losers!" But I think now that people found out another way, it sounds even worse. And he's lost quite some credibility there.
Saying honestly that he's not there where he seems to be is one bit of information that I think the public should have a right to know. And it wouldn't have been such a big deal. I mean, it's not like he reports from Florida all year (that would change the case a bit).
Hey, what does Slashdot have to do with this?
:)
Heh, I've always wanted to reply to a troll...
:)
So you're saying that if I publish a Java program with code stolen from your GPLed program, I'm not violating the license because I "effectively include the source code." Firstly, that is BS, as others have pointed out already in this thread; secondly, the point is that my program is only valid if I still put it under the GPL. If I don't do that, I can redistrubite the source code as much as I want to---it's still a violation. I am not permitted to take your GPL code, put my name on it, change the license to, say, BSD or even Public Domain, without violating your rights to the original code.
But hey, nice troll. At least people are biting
Probably neither. My Canon PowerShot A70 shoots 2048x1536 = 3,145,728 pixels, and claims to have 3.2 megapixels. The specs say it has 3.34 million photo detectors, so it seems like Canon took the 'average' of these two numbers... much like for the A60 which claims 2.0 megapixels, and has in fact 1.92 million vs. 2.11 million.
So, aside from rounding for marketing purposes, it seems like the 'mega' is decimal (MPixels, not MiPixels).
Aww... this thread is getting too messy and unmaintainable. You should've done a complete rewrite instead:
It was too messy and unmaintainable. I'll wait until the rewrite comes out to fix all the grammar and spelling bugs.
Hm. Like KISS would ever publicize their CVS code, if it exists... sounds not much more likely than SCO opening up their code. The only proper way for MPlayer to prove their innocence (if they ever had to) would be to find the oldest copy of MPlayer code (hey, their CVS is open...) containing the code in question and to hope that it's older than KISS's code, or even older than KISS themselves. I would guess that's not too unlikely...
Probably not only that. Let's suppose that Java's trig functions are indeed native. The problem is, they're still functions which are called as standard class methods. In C, the trig functions themselves are compiled and thus of course native, but the complete overhead for making the function calls goes away, because ideally the compiler will even in-line the resulting code. I don't think this is done in Java.
This would also explain why the standard math operations (+,-,*,/) are indeed fast -- they're mapped directly onto processor instructions. The trig functions, however, are functions, and are apparently not optimized like one would optimize such a function in a C/C++ compiler.
The article was actually written by Christopher W. Cowell-Shah. Also, as others noted, except for the I/O benchmark (where gcc was actually faster) the benchmarks couldn't profit much from the OS around it since they were just computationally intensive, but didn't make much use of OS-specific library functions.
Hm, I don't really think that it has much to do with the fact that it runs on Windows. Except for maybe the I/O bound benchmarks, the others, couldn't be expected to gain much from being 'optimized for Windows'. And gcc was actually better on I/O. VC++ wins on the math benchmarks -- so MS may just have the better set of Intel optimizations.
Besides, it would've been kinda difficult comparing the same set of compilers under native Linux, don't you think?
Actually, it's more of a surprise that Java performed /equally well/ in all benchmarks -- except the trig benchmark. The dismal performance in the trig benchmark is the second, much bigger surprise. It would be good to see the source code used to find out whether the code is BS or Java actually totally fails to produce efficient trig code.
Looks like it's time to remind everyone of Bush in 30 Seconds...
Hmm, interesting. Someone should really do this. Make it a 2- or 4-processor Efficeon PC (8 would be a bit too many IMO) with accompanying energy-saving hardware around it, put it in a compact box, equip it with a well pre-configured Linux, and it might actually sell pretty well (OK, only among geeks, with Linux not being mass-market ready...). That would be one nice box. Low power consumption, proper performance, no fan noise while Transmeta's Thermal Extensions are active... I might buy something like that.
Well, at least high-power, high-heat, high-speed. Have you ever seen a computer magazine publishing a serious article about 'efficient computers' or even 'portable, energy-saving laptops'? I haven't. Even when it comes to laptops, all you see is Tom's-Hardware-style performance benchmarks, CPU performance, graphics performance, etc. This really pisses me off. Only gamers (and a few others) really need that much performance, in my opinion. Most Joe Average users are just caught by ads and 'reviews' of high-performance systems... "you really need to have this graphics card... well, it requires a 1000W power supply, but you won't need heating in your house anymore... you really need it for your word processor, spread sheet, and net surfing."
I, for one, am very happy with my LifeBook P-2120. It has everything I need (ok, maybe for some people it's missing Bluetooth), it's fast enough to watch DVDs and DivX, with all batteries installed it runs 9-10 hours, and it's still extremely compact and light-weight, and it runs Linux alright. What more could I need?
Transmeta has found a good niche. But it deserves much more attention.
...to answer this survey would be, "we use Linux because it is Free as in Speech." Any other answer is just a cheap way of giving MS helpful tips on how to make more money. Too bad the survey does not really allow this kind of answer. But, I think, if everybody participating said that (GNU/)Linux is the single choice because it is Free, we could leave a good impression... at least I suppose RMS would agree.
/really/ use GNU/Linux (although, of course, I get all these reasons as a side effect). I use it because it is Free." I wonder what MS would do if everybody answered the survey like that?
So, if it is possible in any way, fill out the survey saying, "none of the given reasons are why I
I'd give you a .torrent, but after downloading 4294967295 bytes, it simply went back to 0...
1 PRINT "0WN3D"
:-)
2 GOTO 1
Better start out at 1, while you're at it
Nice idea, but keep in mind that THC can be traced at other places in your body than just in your blood. Hair comes to mind, for instance. So for now, this technology, if it ever becomes practical, won't save you from the consequences of a drug test...
Careful, I think you're infringing Darl's copyright on that piece of code.
Hm, well at least in Europe (especially Germany), ISDN is very widely used . The infrastructure is there, everybody can have an ISDN line set up. And many people do, especially businesses are running on ISDN. OK, the bandwidth thing was of course nonsense; but the best-sold solution in Germany right now is ISDN+DSL.