Read a Usenet group or follow any online forum for a few months. Posts and responses to posts become painfully predictable after a while. Heck, you can predict the general flow of responses to almost any Slashdot story, not even counting the obvious Natalie Portman, Beowulf cluster, insensitive clod, etc.
Re:Great for GC and dynamically typed languages
on
Athlon 64 Debuts
·
· Score: 1
That being said, taking 2 or 3 bits on a 32 bit field is a lot, but it is very interesting to realize that that cost go away on a 64 bit machine.
But they don't. As it stands, you could switch to native 64 bit values in most any dynamic language without much trouble, and at the expense of very few instructions. Just use 32-bits for the value and 32-bits for the tag. But then you realize that wasting 32-bits for the tag is crazy, because now all of a sudden you're using twice as much memory as before. A cons cell in Lisp and Scheme jumps from 8 bytes to 16 bytes. And now you're blowing the cache twice as often, leading to an overall slowdown.
The preferred method is to use a staged tag scheme, so you only need the bottom two bits to identify pointers, and you use more bits for less important datatypes.
You've got me there. You don't have to learn any of vim's advanced features if you don't use vim. Thanks for clearing that up.
And the original author's point was that he spent too much time fiddling with configurable word processors, therefore he wanted something that was fixed and simple. That doesn't describe vim at all. In fact, configuring vim is a major pain in the ass compared with, say, Microsoft Word.
Okay, maybe the classic version of vi is (which I have written tens of thousands of lines of code with), but not the more recent incarnations like vim. Surem vim doesn't have help balloons and all that, but you'll still spend the next five years figuring out everything there is. If you're the kind of person who loves to fiddle with fonts and colors and settings, then vim is like crack. You'll kill even more time configuring vim than most GUI editors because it's all so byzantine.
Realistically, if you want straightforward and simple, just use one of the many Notepad-like editors. Notepad is too simplistic, but there are many free, enhanced versions (like EditPad Lite). You still don't have to worry about fonts and help balloons and automatic reformatting, but something like EditPad Lite (or EditPlus) is much less invisible to use than vi.
He says: "People who are serious about getting the job done on time and under budget will use tools such as Visual Basic." No further comment necessary
Of course, Greenspun is right. For what Visual Basic was designed for--writing GUI front ends to other appplications--is is absolutely brilliant. You'd be crazy to use Java or C++ or MFC (or whatever) instead.
x86 has always had 64-bit instructions...
on
Is Prescott 64-bit?
·
· Score: 1
...on the FPU. The Pentium FPU can not only load and store 64-bit values directly--over a 64-bit bus--but can load and store 80-bit values. In fact, all internal FPU operations are 80-bit.
They require lots of power, and the capabilities provided by that power consumption aren't needed by 99% of the people who buy them. That' pretty much describes SUVs.
If the auto market were like the PC market, though, then you wouldn't be able to buy anything except a SUV.
A decently configured G4 PowerBook is $2599. That's as much as a G5 desktop already. I fully expect G5 PowerBooks to start at $2999 and extend past $4000.
Please check the SPEC web site [spec.org] for a performance evaluation of the Pentium 4's floating-point (FP) performance. In particular, it outperforms the UltraSPARC III even though the latter has a 2-to-1 advantage in the width of its databus -- 64 bits versus 32 bits.
Note that: 1. The Pentium ONE had a 64-bit data bus. This is nothing new. The FPU was able to load and store 64-bit values directly. 2. The x86 FPU is EIGHTY bits internally. Not 32, not 64, but 80. Again, this is not new.
I agree whole-heartedly. This is 2003, and the biggest game of the year locks you down to a propriotary platform. This is an inexcusable insult on Valve's behalf.
How is this any different than GTA3 being a PS2 exclusive for over a year, Halo only being released for the Xbox, FFX only being released for the PS2, Metroid Prime and Mario Sunshine only being released for the Game Cube...sounds like business as usual to me.
For hobby sites, no. For proper sites, definitely. Far too many people build a site without any understanding of how the browser talks to the server.
But these kind of things are mostly dwarfed by making a site overly graphic-heavy. That's the number one problem. Make a site that's lean, and you're 90% of the way there.
Microbenchmarks about string concatenation and method calling are mostly irrelevant. What's important is too see how an implementation scales up to huge programs, say 500,000 lines of code and 1GB of data. There have been plenty of cases of fast languages not scaling up well (like C++), because you have to add so much cruft to hande the kind of problems you hit at that level, and much "slower" languages outrunning them because they were architected for that kind of problem.
For over five years now, Borland has divided its development products into three classes: personal, professional, and enterprise. It used to be that all three had similar licenses, but each level had more features than the previous version. All were suitable for professional development. The personal edition was ~$100, the professional ~$500, and enterprise ~$2500.
Then, a few years ago, Borland changed this scheme. The professional version jumped up to ~$1000, and the enterprise beyond that. The catch is that at the same time they changed the license of the personal version so it cannot be used commercially or to develop commercial products--even low cost products. So now you have the $100 product that's essentially crippled, and to go to the next level, just to remove that one clause from the license, you have to spend $900. That's completely ridiculous.
Please Borland, give it up. Why should I have to pay $1000 when there are other products at half the price? The answer is "I won't." I'll buy your competitor's products.
An oft heard sentiment is that choice is one of the key tenets of Linux: there are many window managers to choose from, many text editors, desktop environments are optional, etc. What's disappointing is how, after all this time, it has come down to UNIXalikes vs. Windows. That's choice, I guess, but it's more like having to pick from Republicans and Democrats (many Americans think it's a choice, but everyone else laughs at how ridiculous it has become). Yeah, there's BSD and OS X, but again, they're just variants of UNIX. If Linux proponents have their way--and I admit that they are well-meaning--they we'll all be using UNIX. This does not strike me as a good thing.
Remember, the GNU project and the original Linux kernel were both originally attempts to clone UNIX, an operating system that was 20 years old when Linus started hacking. What if Stallman and Torvalds had been enamored of VMS or OS/360 instead? Would it have been a good thing to have those on our desktops? I respect the historical importance of UNIX, and many of its fundamental principles, but after putting the "Let's take down Microsoft!" battle aside, I'm not sure it's what we really want. Perhaps the real problem is that we shouldn't be so concerned with the archaic concept of "operating systems," and instead have computing systems designed the other way around, from a usability point of view (and, no, that doesn't mean we should abolish the command line and have Clippy-like interfaces).
Yes, sadly, many programs for non-MS operating systems, like Linux, are blatant clones of that from Microsoft. It's always been a bit ironic, but livable. But cloning.NET--as Mono--is by far one of the dumbest moves ever, and I mean that in a non-trolling way. At its heart,.NET is a way to break free from the aged Win32 API and old fashioned languages like C and C++. This is they key point of.NET, not web services. Now you can use Visual Basic-like forms from any language. Now you can have garbage collection. Now you can have true modules, not the FORTRAN-era separate compilation of C. None of this is new; none of this was invented by Microsoft. But is all so much better than building apps with MFC or raw Win32 calls. Windows programmers are flocking to.NET for this reason.
But there are other ways to reach the same end. Python + a UI toolkit is a biggie. It's even more modern than C#, which is hopelessly mired in the 1990s philosophy of very strict object-orientation (Python is much looser in this regard). And it's interpreted, so you can incrementally build and test code, while still having all the same general benefits of.NET and C#. So anyone promoting Mono for Linux is putting their effort in entirely the wrong place. This is the one spot in which open source is already far superior, but for some reasons some zealots want to copy the inferfior solution, most likely just to spite Microsoft. What a complete waste of time.
If you play some of the newer games, a GF3 isn't adequate. If you want to play the newest DX9 games then a GF3 is completely inadequate (for the full experience). Go look at the framerates coming out of HL2
You can't talk about performance of games that haven't been released yet, like HL2. That's a total fanboy realm.
My point is that we're essentially talking about a handful of games here, and these are not games that are particularly well optimized. If 3D game XYZ was targeted for the Xbox, it would rock and roll and be plenty fast. What happens on the PC is that developers start shooting for a GF3-class card, then they all upgrade to the FX, then they all upgrade to the latest ATI. Then they shoot backwards and tag the GF3 as the low end, even though they're essentially blowing off that card. The sad part is that high-end PC gaming is moribund, except for a small class of fanboy games. Everyone else has either given up or bought a console. Mass market PC gaming is growing like crazy, though.
I suppose a 33% improvement in frame rate is "incremental". Especially when you're talking about 45 fps vs 60 fps. How meaningless. Power or expense? You mean that because the GeForce FX consumes more power and costs more than the ATI cards that it's better, right? Gotcha.
I never said the FX was better. I said that it's a damn nice card, whose only offense isn't being as fast as a newer card. But that's like arguing that one fighter jet is 200Mph faster than another. Who cares, especially if that jet sits in the hangar 90% of the time anyway?
This article seems to reiterate what everyone has been saying (Carmack, Valve, everyone). The GeforceFX architecture can only be made competitive for 3d engines using modern shaders with herculean effort. This is to be competitive, not dominantly superior.
Now wait a minute. The GeForceFX is essentially faster than anything out there, except for the newest Radeon cards. That makes it the second fastest 3D hardware solution for PCs. And it is certainly faster than past nVidia cards, card which were already insanely fast. Considering that no one has even remotely pushed the limits of the GeForce 3, a card that's still dramatically faster than what ships in most Dell PCs, it's not like anyone can put down the FX as "slow" by any stretch of the imagination.
The FX is only "slow" in the minds of fanboys who live for incremental performance increases without regard to power consumption or expense, like saying that a Ferarri is slow compared to a rocked powered car with no brakes or steering.
Dude, it's called "trickle-down." Same thing happens at car-shows. The industry puts out "concept cars" that are truly revolutionary, cost a small (or large) fortune, and are bought by very few (if any.) Eventually, some of those innovations make their way down to consumer-level designs, where they have become refined and affordable. People are happy.
But PC buyers don't give a damn about 3D graphics except for the 10-20% that play hardcore games. No one else cares. 3D PC games failed at being mass market, because it's too hard to keep up with system requirements, drivers, patches, etc. Most people who want to play that kind of game buy a console, because the headaches are so much fewer.
That a Dell ships with Intel integrated video or a GeForce FX makes no difference to most people, even most techies, except for people who do 3D modeling and such for a living.
You may not need a 10GHz CPU. I know I don't, at least, not yet. But as interfaces improve, as software becomes more complicated, hardware like this may well a requirement. After all, monitor images are still nothing like as detailed as a printed page, and even a very high-end monitor has only a fraction of the resolution that can be produced by a $100 printer. It's going to take serious hardware to acheive that kind of image clarity in real time. 10GHz CPUs may be just the beginning.
Maybe. But some technology is good enough, like terminals used to display flight information in airports and LCD games at Toys 'R Us. You could replace LCD games with much fancier alternatives, but they're cheap, low power, have a bit of charm, and people like them.
The general problem with PC technology is that it's too broad. We don't know what we want, so we expand in every direction. Now we have PCs to two pound heatsinks, four fans, 400W power supplies, and need giant, impossibly complex operating systems to run (in this regard, Linux is in the same category as Windows, especially once you factor everything on top of the kernel into the picture, like XFree86). In the end, we have boxes sitting on desks in insurance companies and corporate reception areas that have 80-bit bit FPUs (standard on the x86 since the 486), 64-bit data bases (standard on the x86 since the Pentium I), MMX instructions (standard on the x86 since then Pentium MMX), SSE2 vector units (standard on the x86 since the Pentium 4), and so on.
Slashdot Mirror
My mac owning friend assures me this is all true, and anyone who tells you different is a dirty liar!
:)
I hate to break it to you, but obviously all of those were invented by Commodore and first appeared in the Amiga.
Read a Usenet group or follow any online forum for a few months. Posts and responses to posts become painfully predictable after a while. Heck, you can predict the general flow of responses to almost any Slashdot story, not even counting the obvious Natalie Portman, Beowulf cluster, insensitive clod, etc.
That being said, taking 2 or 3 bits on a 32 bit field is a lot, but it is very interesting to realize that that cost go away on a 64 bit machine.
But they don't. As it stands, you could switch to native 64 bit values in most any dynamic language without much trouble, and at the expense of very few instructions. Just use 32-bits for the value and 32-bits for the tag. But then you realize that wasting 32-bits for the tag is crazy, because now all of a sudden you're using twice as much memory as before. A cons cell in Lisp and Scheme jumps from 8 bytes to 16 bytes. And now you're blowing the cache twice as often, leading to an overall slowdown.
The preferred method is to use a staged tag scheme, so you only need the bottom two bits to identify pointers, and you use more bits for less important datatypes.
You've got me there. You don't have to learn any of vim's advanced features if you don't use vim. Thanks for clearing that up.
And the original author's point was that he spent too much time fiddling with configurable word processors, therefore he wanted something that was fixed and simple. That doesn't describe vim at all. In fact, configuring vim is a major pain in the ass compared with, say, Microsoft Word.
Ahh, but the beauty is, you don't have to.
You don't have to with any other text editor either.
Okay, maybe the classic version of vi is (which I have written tens of thousands of lines of code with), but not the more recent incarnations like vim. Surem vim doesn't have help balloons and all that, but you'll still spend the next five years figuring out everything there is. If you're the kind of person who loves to fiddle with fonts and colors and settings, then vim is like crack. You'll kill even more time configuring vim than most GUI editors because it's all so byzantine.
Realistically, if you want straightforward and simple, just use one of the many Notepad-like editors. Notepad is too simplistic, but there are many free, enhanced versions (like EditPad Lite). You still don't have to worry about fonts and help balloons and automatic reformatting, but something like EditPad Lite (or EditPlus) is much less invisible to use than vi.
He says: "People who are serious about getting the job done on time and under budget will use tools such as Visual Basic." No further comment necessary
Of course, Greenspun is right. For what Visual Basic was designed for--writing GUI front ends to other appplications--is is absolutely brilliant. You'd be crazy to use Java or C++ or MFC (or whatever) instead.
...on the FPU. The Pentium FPU can not only load and store 64-bit values directly--over a 64-bit bus--but can load and store 80-bit values. In fact, all internal FPU operations are 80-bit.
They require lots of power, and the capabilities provided by that power consumption aren't needed by 99% of the people who buy them. That' pretty much describes SUVs.
If the auto market were like the PC market, though, then you wouldn't be able to buy anything except a SUV.
A decently configured G4 PowerBook is $2599. That's as much as a G5 desktop already. I fully expect G5 PowerBooks to start at $2999 and extend past $4000.
Please check the SPEC web site [spec.org] for a performance evaluation of the Pentium 4's floating-point (FP) performance. In particular, it outperforms the UltraSPARC III even though the latter has a 2-to-1 advantage in the width of its databus -- 64 bits versus 32 bits.
Note that:
1. The Pentium ONE had a 64-bit data bus. This is nothing new. The FPU was able to load and store 64-bit values directly.
2. The x86 FPU is EIGHTY bits internally. Not 32, not 64, but 80. Again, this is not new.
I agree whole-heartedly. This is 2003, and the biggest game of the year locks you down to a propriotary platform. This is an inexcusable insult on Valve's behalf.
How is this any different than GTA3 being a PS2 exclusive for over a year, Halo only being released for the Xbox, FFX only being released for the PS2, Metroid Prime and Mario Sunshine only being released for the Game Cube...sounds like business as usual to me.
SDL is nothing more than very basic glue, barely enough to provide a portable framework.
And it all honesty, it's not even very good, just portable.
But they must have, because there are no bugs in any software that runs under Linux. There never have been, and there never will be.
For hobby sites, no. For proper sites, definitely. Far too many people build a site without any understanding of how the browser talks to the server.
But these kind of things are mostly dwarfed by making a site overly graphic-heavy. That's the number one problem. Make a site that's lean, and you're 90% of the way there.
From the Devil's Dictionary:
FUD: The sound made by someone attempting to wish away inconvenient facts.
http://www.eod.com/devil/archive/fud.html
This is what you want.
Microbenchmarks about string concatenation and method calling are mostly irrelevant. What's important is too see how an implementation scales up to huge programs, say 500,000 lines of code and 1GB of data. There have been plenty of cases of fast languages not scaling up well (like C++), because you have to add so much cruft to hande the kind of problems you hit at that level, and much "slower" languages outrunning them because they were architected for that kind of problem.
One interesting case is Yet Another Web Server (YAWS), which is written in an interpreted language and completely blows the socks off of Apache under high load.
For over five years now, Borland has divided its development products into three classes: personal, professional, and enterprise. It used to be that all three had similar licenses, but each level had more features than the previous version. All were suitable for professional development. The personal edition was ~$100, the professional ~$500, and enterprise ~$2500.
Then, a few years ago, Borland changed this scheme. The professional version jumped up to ~$1000, and the enterprise beyond that. The catch is that at the same time they changed the license of the personal version so it cannot be used commercially or to develop commercial products--even low cost products. So now you have the $100 product that's essentially crippled, and to go to the next level, just to remove that one clause from the license, you have to spend $900. That's completely ridiculous.
Please Borland, give it up. Why should I have to pay $1000 when there are other products at half the price? The answer is "I won't." I'll buy your competitor's products.
An oft heard sentiment is that choice is one of the key tenets of Linux: there are many window managers to choose from, many text editors, desktop environments are optional, etc. What's disappointing is how, after all this time, it has come down to UNIXalikes vs. Windows. That's choice, I guess, but it's more like having to pick from Republicans and Democrats (many Americans think it's a choice, but everyone else laughs at how ridiculous it has become). Yeah, there's BSD and OS X, but again, they're just variants of UNIX. If Linux proponents have their way--and I admit that they are well-meaning--they we'll all be using UNIX. This does not strike me as a good thing.
Remember, the GNU project and the original Linux kernel were both originally attempts to clone UNIX, an operating system that was 20 years old when Linus started hacking. What if Stallman and Torvalds had been enamored of VMS or OS/360 instead? Would it have been a good thing to have those on our desktops? I respect the historical importance of UNIX, and many of its fundamental principles, but after putting the "Let's take down Microsoft!" battle aside, I'm not sure it's what we really want. Perhaps the real problem is that we shouldn't be so concerned with the archaic concept of "operating systems," and instead have computing systems designed the other way around, from a usability point of view (and, no, that doesn't mean we should abolish the command line and have Clippy-like interfaces).
Yes, sadly, many programs for non-MS operating systems, like Linux, are blatant clones of that from Microsoft. It's always been a bit ironic, but livable. But cloning .NET--as Mono--is by far one of the dumbest moves ever, and I mean that in a non-trolling way. At its heart, .NET is a way to break free from the aged Win32 API and old fashioned languages like C and C++. This is they key point of .NET, not web services. Now you can use Visual Basic-like forms from any language. Now you can have garbage collection. Now you can have true modules, not the FORTRAN-era separate compilation of C. None of this is new; none of this was invented by Microsoft. But is all so much better than building apps with MFC or raw Win32 calls. Windows programmers are flocking to .NET for this reason.
.NET and C#. So anyone promoting Mono for Linux is putting their effort in entirely the wrong place. This is the one spot in which open source is already far superior, but for some reasons some zealots want to copy the inferfior solution, most likely just to spite Microsoft. What a complete waste of time.
But there are other ways to reach the same end. Python + a UI toolkit is a biggie. It's even more modern than C#, which is hopelessly mired in the 1990s philosophy of very strict object-orientation (Python is much looser in this regard). And it's interpreted, so you can incrementally build and test code, while still having all the same general benefits of
If you play some of the newer games, a GF3 isn't adequate. If you want to play the newest DX9 games then a GF3 is completely inadequate (for the full experience). Go look at the framerates coming out of HL2
You can't talk about performance of games that haven't been released yet, like HL2. That's a total fanboy realm.
My point is that we're essentially talking about a handful of games here, and these are not games that are particularly well optimized. If 3D game XYZ was targeted for the Xbox, it would rock and roll and be plenty fast. What happens on the PC is that developers start shooting for a GF3-class card, then they all upgrade to the FX, then they all upgrade to the latest ATI. Then they shoot backwards and tag the GF3 as the low end, even though they're essentially blowing off that card. The sad part is that high-end PC gaming is moribund, except for a small class of fanboy games. Everyone else has either given up or bought a console. Mass market PC gaming is growing like crazy, though.
I suppose a 33% improvement in frame rate is "incremental". Especially when you're talking about 45 fps vs 60 fps. How meaningless. Power or expense? You mean that because the GeForce FX consumes more power and costs more than the ATI cards that it's better, right? Gotcha.
I never said the FX was better. I said that it's a damn nice card, whose only offense isn't being as fast as a newer card. But that's like arguing that one fighter jet is 200Mph faster than another. Who cares, especially if that jet sits in the hangar 90% of the time anyway?
This article seems to reiterate what everyone has been saying (Carmack, Valve, everyone). The GeforceFX architecture can only be made competitive for 3d engines using modern shaders with herculean effort. This is to be competitive, not dominantly superior.
Now wait a minute. The GeForceFX is essentially faster than anything out there, except for the newest Radeon cards. That makes it the second fastest 3D hardware solution for PCs. And it is certainly faster than past nVidia cards, card which were already insanely fast. Considering that no one has even remotely pushed the limits of the GeForce 3, a card that's still dramatically faster than what ships in most Dell PCs, it's not like anyone can put down the FX as "slow" by any stretch of the imagination.
The FX is only "slow" in the minds of fanboys who live for incremental performance increases without regard to power consumption or expense, like saying that a Ferarri is slow compared to a rocked powered car with no brakes or steering.
Dude, it's called "trickle-down." Same thing happens at car-shows. The industry puts out "concept cars" that are truly revolutionary, cost a small (or large) fortune, and are bought by very few (if any.) Eventually, some of those innovations make their way down to consumer-level designs, where they have become refined and affordable. People are happy.
But PC buyers don't give a damn about 3D graphics except for the 10-20% that play hardcore games. No one else cares. 3D PC games failed at being mass market, because it's too hard to keep up with system requirements, drivers, patches, etc. Most people who want to play that kind of game buy a console, because the headaches are so much fewer.
That a Dell ships with Intel integrated video or a GeForce FX makes no difference to most people, even most techies, except for people who do 3D modeling and such for a living.
You may not need a 10GHz CPU. I know I don't, at least, not yet. But as interfaces improve, as software becomes more complicated, hardware like this may well a requirement. After all, monitor images are still nothing like as detailed as a printed page, and even a very high-end monitor has only a fraction of the resolution that can be produced by a $100 printer. It's going to take serious hardware to acheive that kind of image clarity in real time. 10GHz CPUs may be just the beginning.
Maybe. But some technology is good enough, like terminals used to display flight information in airports and LCD games at Toys 'R Us. You could replace LCD games with much fancier alternatives, but they're cheap, low power, have a bit of charm, and people like them.
The general problem with PC technology is that it's too broad. We don't know what we want, so we expand in every direction. Now we have PCs to two pound heatsinks, four fans, 400W power supplies, and need giant, impossibly complex operating systems to run (in this regard, Linux is in the same category as Windows, especially once you factor everything on top of the kernel into the picture, like XFree86). In the end, we have boxes sitting on desks in insurance companies and corporate reception areas that have 80-bit bit FPUs (standard on the x86 since the 486), 64-bit data bases (standard on the x86 since the Pentium I), MMX instructions (standard on the x86 since then Pentium MMX), SSE2 vector units (standard on the x86 since the Pentium 4), and so on.