Your mobo/CPU combo costs $100. For $150 (including S/H) you can get a 700MHz Thunderbird Athlon +mobo from PriceWatch. That $50 difference really is just noise, and the performance gain by going to an Athlon is staggering.
A) 3D performance == fp performance for all practical considerations.
B) Why the hell cares about 2D performance? That's more of a graphics card issue than anything else! As for MMX, it's essentially a 2D-only instruction set. Nobody uses integer 3D engines anymore.
Why is this worth discussing?
A) It's an x86 chip. I'd bet that a full 90+% of/. users run x86.
B) It's cheap. I'd bet a full 90+5 of/. users like cheap.
C) It doesn't need active cooling. A full 90+% of/. users think that excessive cooling requirements are a deal breaker for a CPU (god knows why;)
D) Its runs Linux.
People said the same thing about C and ASM. Its still not true, well optimized ASM is still faster than well optimized C. Given equal optimization, Java will always be slower than C. However, you get to the point where it really doesn't matter because the underlying machine is so powerful that the couple of percent difference is neglegible. I dislike Java for different reasons, mainly that compile once run everywhere is a load of bull, and most of Java's core (ie not stuff like the class libraries) can be replicated with good portable objective C code. (Does such a thing exist?) I think the POSIX way is absolutely the correct way to go. Most new OSs these days support POSIX, to some extent, and porting between implementations is a piece of cake. If POSIX was expanded to include GUI features, cross platform issues porbably wouldn't be a Java benifet aymore.
Congratulations, another selling point for BeOS! NOT! Adding support to the nv driver (which existed before
the BeOS drivers) was as simple as adding the PCI id's of the Geforce2 (MX) cards.
>>>>>>>>>>>>>>..
But the fact remains that for a few weeks, OSS hardcores unwilling to use the NVIDIA-supplied drivers were able to use their cards in BeOS rather than Linux! Wait... the BeOS drivers are closed to... damn it, I'm confusing myself...
You're blowing smoke right?
>>>>>>>>
Don't think so. It seems that I'm correct, XAA provides a common interface for hardware acceleration. You missed my point: If XAA automatically delegates functions to hardware drivers, than any driver that implements XAA hardware acceleration will accelerate any app that uses XAA. You answered my question indirectly though, I didn't know that XRender added new functionality to XAA itself that new cards have to implement.
OT: Meanwhile Be is ready for a buyout! With a market cap of just above $30M they are ripe for the picking.
And to think JLG wanted $400M for a company that apparantly is only good at making PC demos:)
>>>>>>>>>>>>
RedHat is only good at making server OSs whats you're point? I mean there's no point fighting over it, Microsoft is still king.
Actually marketing and selling their stuff is foreign to the current management. Yep, as a former be fan myself it's really sad to see them struggle like this.
>>>>>>>>>>>>>
If you've read BeNews lately, you'll find out that they're doing pretty well for themselves. If their deals pan out, then they should become competitive.
But hey, nice demos and c00l technology simply aren't enough to survive.
>>>>>>>>>
Coming from a Linux user? What, besides c001 technology (and free stuff) IS Linux?
First, let me say that I'm so glad somebody else agrees with the fact that Linux distros are getting bloated. However, I've got a few points...
A) This is an example of the "good type of freedom" that multiple distros offers. Each one can tailored to different segments of the market. However, I think the main problem is that there is no "minimalist" distro aside from Stampede. (There's Slack, but they have a more stability rather than performance slant.) And of course, Stampede isn't exactly ready for prime time.
B) Everyone using seriously using Linux has broad-band access. While its fine for some distros to offer huge software bundles, I really think it makes more sense for the distro makers to concentrate on the core system, and leave software collections to $4 cheapbytes CDs. It would also stop the bastardization of software by distro makers. Suse and Mandrake have uglized versions of KDE, and I have not found a way to get rid of the decorations and menus.
C) I'd hesitate over all his statements about "preconfigured." The problem with many distros today isn't that stuff doesn't work "out of box" but that any changes in configuration require the user to go "against the grain" of the preconfigured system, which makes configuration harder. For example, I keep mentioning Mandrake and its inane SysV wrappers. While it may make it easier to work things out of box, it just complicates things when the user needs to do something themselves. Distro makers should not try to fix things with kludges on top (its obscene the number of kludges contained in Linux, particularly in distros like Mandrake, RedHat, and Suse) but reform the core system to make it cleaner and more coherant. Here's where the LSB could really help. If they made a good, clean core organization (without being weighed down by stupid *NIX legacy issues) while Linux is still small enough, and I am certain that distro makers would adopt it.
A) Why obther supporting ancient hardware. Mono cards? Good god! I thought palletized cards were as bad as it got! XFree86 4 is supposed to be the "second coming" if you will of X. There is no point in supporting ancient hardware. Either upgrade or stick to 3.3.6.
B) Blanking: Yes, fix it...please
C) Especially for the compile phase. The compilation instructions consist of the standard X 6.4 docs. That's just silly.
A) Who's still running PseudoColor visuals that actually want anti-aliasing? BeOS here works fine with 256 colors, anti-aliasing and all.
B) That's probably a good idea.
C) Who cares? As long as it is cross-machine compatible, I'm guessing the XFree folks have enough clout to change the extensions and have the industry follow.
I serious doubt you want only your Qt fonts anti-aliased and your Motif fonts un-anti-aliased. If it were lower than the toolkit level, you could have a config program that specified whether you wanted anti-aliasing or not.
That explains why nobody codes directly to xlib. And to tell the truth, the X protocol is just a dumb idea. Yes, you heard it here first (NOT) There is a reason for having some abstraction at the lowest levels: it allows apps to take automatic advantage of new features. The idea of putting a half-finished windowing environment (X) on top of UNIX is the whole reason you have mismatched applications, non-uniform configurations (wouldn't it be great if X display configs would work for every application?) and all the other problems X has. The sheer fact that every other windowing system has the sense to include some layer of abstractions for apps to code to should give you an idea that the X designers are alone in their thinking that a windowing system should implement *no* user-level features.
I would really enjoy it if XRender were as good as it hast he capability to be. Of course, now I know exactly what is going to happen. I'm going to be staring at Netscape's un-anti-aliased fonts until v7, XplayMidi will never give me anti-aliased fonts, and I'm going to have to eventually deal with 3 different config formats for my anti-aliased truetype fonts. (one for each toolkit)
If it were automated, (eg. pusing the reject-already submitted button gives you a certain messege) takes no time at all. And unless the guys going through the messeges aren't doing what we think they're doing (randomly posting messegse, which would account for many of the errors made recently) pushing an extra button really wouldn't increase the time factor.
Those "fucks" don't support the D-Link 530TX for the same reason those "fucks" at XFree86 aren't doing DRI drivers for the Sis630. Limit time, limited resources, limited demand.
And those "fucks" have greast SBLive! support. If you have a particular problem, just ask for help.
I'm sure that if you ever went on a userboard and asked, you'd get tons of people trying to help. (Be users groups are very friendly.) However, that's a moot point. If you don't like it don't use it. If it doesn't support your hardware, then certainly don't use it. If you are into programming, and want to help out, then go ahead to BeUnited and join a project. If you're starting a programming project, and you want cross platform support, then program with non-X OSs in mind. (Use Qt or GTK or something cross-platform.)
Exactly what I said. The NVIDIA OpenGL ICD doesn't work without the NVIDIA X driver. Meaning that you have to choose between the NVIDIA nvidia driver (3D) or the XFree86 nv driver (Xrender)
Eh? GeForce2 support was supported before the NVidia drivers were released for BeOS, but who cares, X support is much better than BeOS (*cough* OpenGL 1.1.2 through GLX
*cough*)
>>>>>>>>>>>>>>
Yea yea, shut up. I forgot about the NVIDIA drivers. Well, at least BeOS had support before the XFree guys did.
1) If it uses XAA, why does it only accelerate on MGA?
Perhaps because XAA falls back to software function if the hardware is not there (or not implemented yet), much like BeOS.
>>>>>>>>>>>>>.
Every OS falls back on software, but XAA (X Acceleration Architecture) is (by definition) not a software rendrer.
A) They can't because some of the stuff is proprietory code.
B) They can't because an ICD isn't just a driver, its a whole freaking OpenGL ICD. OpenGL ICDs are expensive and time consuming to develop, and every other consumer manufacturer is having troube with theirs. Given the fact that NVIDIA's ICD is totally kick ass, why would they give chip makers like ATI an advantage? I'd wager that if ATI's drivers were as good as NVIDIA's, then the Radeon would be at least 20-30% faster. Also, the Matrox G400MAX would have put a serious dent in RivaTNT2 Ultra sales had NVIDIA's ICD been open. Ideally, what NVIDIA would do is split the drive into three parts. An OSS kernel driver, and OSS X driver, and a closed OpenGL driver. That way they could keep the code closed, yet be able to implement extensions like this. Also, I don't think they really planned on this. Who thought that 4.0.2 would include such a innovative component? Stuff like this just dosn't get added in.x releases, much less.0.x ones.
C) They're a business. Get over it. Right now, you taking your Viper 550 out and putting it in our router probably costs them less than giving ATI free code.
D) Whining and insulting is no way to get what you want. That's why people consider the OSS market a dangerous proposition. If you want OSS drivers ask nicely, help them through it. It is a totally new paradigm, and its benifets to them (if indeed OSS has any benefits to them) will need some time to digest.
Look at the new features! If this were MS code, it would be worth at least a century or two jump in the version number! The cool ones are:
Bug fixes: Yea, those.
Render Extension: The render extensions and additional stuff added to x11perf, xft, and xlib to support it.
Compositing code for Render is complete, but a lot of stuff (polygons, image scaling, seperate alpha, see the summery) are still unimplemented.
Updates to nv for GeForce2: Hah! BeOS had GeForce2 support before X!
It says that Render uses XAA for acceleration, and acceleration on the MGA chip is already implemented. 2Qs
1) If it uses XAA, why does it only accelerate on MGA?
2) Does this mean that it becomes a Render vs 3D choice for NVIDIA users? As far as I can see, the NVIDIA drivers don't support the Render extensions. Or am I just confused.
that use up loads of colors
>>>>>>>
Good god! Please don't tell me you're using 256 colors. We should set up a fund to pay for new (post '80s) graphics cards for people still using pallatized modes.
Not true at all. An OS that *can* be configured is different from an OS that *must* be configured. There is no problem in allow the user the option to change the configuration, as long as the configuration program is logically laid out, well documented (preferably in-program documentation), scalable to the user's level, and totally unnecessary. For an example of the former type of OS, take a look at Linux. For an example of the latter, take a look at BeOS or QNX RtP.
simple users can modify the OS" the main mantra behind Linux????
>>>>>>>>>>>>>>
Actually, the new distros are going AGAINST that idea. By creating a held-together-by-thumbtacks-and-bubblegum miasma of proprietory perl scripts and config files and package layouts and dependency files, ad nauseum, they are making it impossible to tweek the system beyond the "out-of-box" configuration. Instead of making the underlying system more elegant and logical, they are simply adding incoherent and non-uniform tools on top of a incoherent and non-uniform system. If a standard GUI interface was created, for say, the Slackware config files, something that would configure everything from sound to X to the kernel, then you'd end up with an elegant, modifable system. (Well, after they clean out/etc) Currently, however, you have tools, but none of them make sense. You go to KDE to look at the configuration, LinuxConf to deal with some config files, Xconfigurator to config X, XConfig to configure the kernel, sndconfig to configure sound, etc, etc. And when the tools leave something out, you have to hand-edit a set of underlying config files that have asnine layouts and are more suitable for editing programatically.
No.
Nautilus is essentially feature compatible with IE6. In other words, its a tuning of what has been done (better) before. OS X, on the other hand has three things going for it...
A) It looks orgasmically good. And its themable!
B) Quartz wipes the floor with X (except remote display)
C) Mac users would never use anything resembling IE.
Slashdot Mirror
Your mobo/CPU combo costs $100. For $150 (including S/H) you can get a 700MHz Thunderbird Athlon +mobo from PriceWatch. That $50 difference really is just noise, and the performance gain by going to an Athlon is staggering.
A) 3D performance == fp performance for all practical considerations.
B) Why the hell cares about 2D performance? That's more of a graphics card issue than anything else! As for MMX, it's essentially a 2D-only instruction set. Nobody uses integer 3D engines anymore.
Why is this worth discussing? /. users run x86.
/. users like cheap.
/. users think that excessive cooling requirements are a deal breaker for a CPU (god knows why ;)
A) It's an x86 chip. I'd bet that a full 90+% of
B) It's cheap. I'd bet a full 90+5 of
C) It doesn't need active cooling. A full 90+% of
D) Its runs Linux.
'nuff said.
Thirteen megs? I find that hard to believe. BeOS uses libc, and it takes up less than a meg.
People said the same thing about C and ASM. Its still not true, well optimized ASM is still faster than well optimized C. Given equal optimization, Java will always be slower than C. However, you get to the point where it really doesn't matter because the underlying machine is so powerful that the couple of percent difference is neglegible. I dislike Java for different reasons, mainly that compile once run everywhere is a load of bull, and most of Java's core (ie not stuff like the class libraries) can be replicated with good portable objective C code. (Does such a thing exist?) I think the POSIX way is absolutely the correct way to go. Most new OSs these days support POSIX, to some extent, and porting between implementations is a piece of cake. If POSIX was expanded to include GUI features, cross platform issues porbably wouldn't be a Java benifet aymore.
Haha, calm down.
:)
>>>>>>>>>
Arg.
Congratulations, another selling point for BeOS! NOT! Adding support to the nv driver (which existed before
the BeOS drivers) was as simple as adding the PCI id's of the Geforce2 (MX) cards.
>>>>>>>>>>>>>>..
But the fact remains that for a few weeks, OSS hardcores unwilling to use the NVIDIA-supplied drivers were able to use their cards in BeOS rather than Linux! Wait... the BeOS drivers are closed to... damn it, I'm confusing myself...
You're blowing smoke right?
>>>>>>>>
Don't think so. It seems that I'm correct, XAA provides a common interface for hardware acceleration. You missed my point: If XAA automatically delegates functions to hardware drivers, than any driver that implements XAA hardware acceleration will accelerate any app that uses XAA. You answered my question indirectly though, I didn't know that XRender added new functionality to XAA itself that new cards have to implement.
OT: Meanwhile Be is ready for a buyout! With a market cap of just above $30M they are ripe for the picking.
And to think JLG wanted $400M for a company that apparantly is only good at making PC demos
>>>>>>>>>>>>
RedHat is only good at making server OSs whats you're point? I mean there's no point fighting over it, Microsoft is still king.
Actually marketing and selling their stuff is foreign to the current management. Yep, as a former be fan myself it's really sad to see them struggle like this.
>>>>>>>>>>>>>
If you've read BeNews lately, you'll find out that they're doing pretty well for themselves. If their deals pan out, then they should become competitive.
But hey, nice demos and c00l technology simply aren't enough to survive.
>>>>>>>>>
Coming from a Linux user? What, besides c001 technology (and free stuff) IS Linux?
First, let me say that I'm so glad somebody else agrees with the fact that Linux distros are getting bloated. However, I've got a few points...
A) This is an example of the "good type of freedom" that multiple distros offers. Each one can tailored to different segments of the market. However, I think the main problem is that there is no "minimalist" distro aside from Stampede. (There's Slack, but they have a more stability rather than performance slant.) And of course, Stampede isn't exactly ready for prime time.
B) Everyone using seriously using Linux has broad-band access. While its fine for some distros to offer huge software bundles, I really think it makes more sense for the distro makers to concentrate on the core system, and leave software collections to $4 cheapbytes CDs. It would also stop the bastardization of software by distro makers. Suse and Mandrake have uglized versions of KDE, and I have not found a way to get rid of the decorations and menus.
C) I'd hesitate over all his statements about "preconfigured." The problem with many distros today isn't that stuff doesn't work "out of box" but that any changes in configuration require the user to go "against the grain" of the preconfigured system, which makes configuration harder. For example, I keep mentioning Mandrake and its inane SysV wrappers. While it may make it easier to work things out of box, it just complicates things when the user needs to do something themselves. Distro makers should not try to fix things with kludges on top (its obscene the number of kludges contained in Linux, particularly in distros like Mandrake, RedHat, and Suse) but reform the core system to make it cleaner and more coherant. Here's where the LSB could really help. If they made a good, clean core organization (without being weighed down by stupid *NIX legacy issues) while Linux is still small enough, and I am certain that distro makers would adopt it.
A) Why obther supporting ancient hardware. Mono cards? Good god! I thought palletized cards were as bad as it got! XFree86 4 is supposed to be the "second coming" if you will of X. There is no point in supporting ancient hardware. Either upgrade or stick to 3.3.6.
B) Blanking: Yes, fix it...please
C) Especially for the compile phase. The compilation instructions consist of the standard X 6.4 docs. That's just silly.
A) Who's still running PseudoColor visuals that actually want anti-aliasing? BeOS here works fine with 256 colors, anti-aliasing and all.
B) That's probably a good idea.
C) Who cares? As long as it is cross-machine compatible, I'm guessing the XFree folks have enough clout to change the extensions and have the industry follow.
I serious doubt you want only your Qt fonts anti-aliased and your Motif fonts un-anti-aliased. If it were lower than the toolkit level, you could have a config program that specified whether you wanted anti-aliasing or not.
That explains why nobody codes directly to xlib. And to tell the truth, the X protocol is just a dumb idea. Yes, you heard it here first (NOT) There is a reason for having some abstraction at the lowest levels: it allows apps to take automatic advantage of new features. The idea of putting a half-finished windowing environment (X) on top of UNIX is the whole reason you have mismatched applications, non-uniform configurations (wouldn't it be great if X display configs would work for every application?) and all the other problems X has. The sheer fact that every other windowing system has the sense to include some layer of abstractions for apps to code to should give you an idea that the X designers are alone in their thinking that a windowing system should implement *no* user-level features.
I would really enjoy it if XRender were as good as it hast he capability to be. Of course, now I know exactly what is going to happen. I'm going to be staring at Netscape's un-anti-aliased fonts until v7, XplayMidi will never give me anti-aliased fonts, and I'm going to have to eventually deal with 3 different config formats for my anti-aliased truetype fonts. (one for each toolkit)
If it were automated, (eg. pusing the reject-already submitted button gives you a certain messege) takes no time at all. And unless the guys going through the messeges aren't doing what we think they're doing (randomly posting messegse, which would account for many of the errors made recently) pushing an extra button really wouldn't increase the time factor.
Let me clear that up again. I'm not talking about the Slackware tgz's, but the general, distro-independent Linux i386 binaries.
No, I meant the binary TGZ packages. They're never up to date. The only thing in the /tgz directory is source code.
Those "fucks" don't support the D-Link 530TX for the same reason those "fucks" at XFree86 aren't doing DRI drivers for the Sis630. Limit time, limited resources, limited demand.
And those "fucks" have greast SBLive! support. If you have a particular problem, just ask for help.
I'm sure that if you ever went on a userboard and asked, you'd get tons of people trying to help. (Be users groups are very friendly.) However, that's a moot point. If you don't like it don't use it. If it doesn't support your hardware, then certainly don't use it. If you are into programming, and want to help out, then go ahead to BeUnited and join a project. If you're starting a programming project, and you want cross platform support, then program with non-X OSs in mind. (Use Qt or GTK or something cross-platform.)
Exactly what I said. The NVIDIA OpenGL ICD doesn't work without the NVIDIA X driver. Meaning that you have to choose between the NVIDIA nvidia driver (3D) or the XFree86 nv driver (Xrender)
Eh? GeForce2 support was supported before the NVidia drivers were released for BeOS, but who cares, X support is much better than BeOS (*cough* OpenGL 1.1.2 through GLX
*cough*)
>>>>>>>>>>>>>>
Yea yea, shut up. I forgot about the NVIDIA drivers. Well, at least BeOS had support before the XFree guys did.
1) If it uses XAA, why does it only accelerate on MGA?
Perhaps because XAA falls back to software function if the hardware is not there (or not implemented yet), much like BeOS.
>>>>>>>>>>>>>.
Every OS falls back on software, but XAA (X Acceleration Architecture) is (by definition) not a software rendrer.
Ah. Typical /. iDot. NVIDIA has "a clue."
.x releases, much less .0.x ones.
A) They can't because some of the stuff is proprietory code.
B) They can't because an ICD isn't just a driver, its a whole freaking OpenGL ICD. OpenGL ICDs are expensive and time consuming to develop, and every other consumer manufacturer is having troube with theirs. Given the fact that NVIDIA's ICD is totally kick ass, why would they give chip makers like ATI an advantage? I'd wager that if ATI's drivers were as good as NVIDIA's, then the Radeon would be at least 20-30% faster. Also, the Matrox G400MAX would have put a serious dent in RivaTNT2 Ultra sales had NVIDIA's ICD been open. Ideally, what NVIDIA would do is split the drive into three parts. An OSS kernel driver, and OSS X driver, and a closed OpenGL driver. That way they could keep the code closed, yet be able to implement extensions like this. Also, I don't think they really planned on this. Who thought that 4.0.2 would include such a innovative component? Stuff like this just dosn't get added in
C) They're a business. Get over it. Right now, you taking your Viper 550 out and putting it in our router probably costs them less than giving ATI free code.
D) Whining and insulting is no way to get what you want. That's why people consider the OSS market a dangerous proposition. If you want OSS drivers ask nicely, help them through it. It is a totally new paradigm, and its benifets to them (if indeed OSS has any benefits to them) will need some time to digest.
Look at the new features! If this were MS code, it would be worth at least a century or two jump in the version number! The cool ones are:
Bug fixes: Yea, those.
Render Extension: The render extensions and additional stuff added to x11perf, xft, and xlib to support it.
Compositing code for Render is complete, but a lot of stuff (polygons, image scaling, seperate alpha, see the summery) are still unimplemented.
Updates to nv for GeForce2: Hah! BeOS had GeForce2 support before X!
xf86cfg: A new, graphical configuration utility.
And much much more!
Here is the link.
It says that Render uses XAA for acceleration, and acceleration on the MGA chip is already implemented. 2Qs
1) If it uses XAA, why does it only accelerate on MGA?
2) Does this mean that it becomes a Render vs 3D choice for NVIDIA users? As far as I can see, the NVIDIA drivers don't support the Render extensions. Or am I just confused.
Wow! So you agree!
that use up loads of colors
>>>>>>>
Good god! Please don't tell me you're using 256 colors. We should set up a fund to pay for new (post '80s) graphics cards for people still using pallatized modes.
Not true at all. An OS that *can* be configured is different from an OS that *must* be configured. There is no problem in allow the user the option to change the configuration, as long as the configuration program is logically laid out, well documented (preferably in-program documentation), scalable to the user's level, and totally unnecessary. For an example of the former type of OS, take a look at Linux. For an example of the latter, take a look at BeOS or QNX RtP.
simple users can modify the OS" the main mantra behind Linux????
/etc) Currently, however, you have tools, but none of them make sense. You go to KDE to look at the configuration, LinuxConf to deal with some config files, Xconfigurator to config X, XConfig to configure the kernel, sndconfig to configure sound, etc, etc. And when the tools leave something out, you have to hand-edit a set of underlying config files that have asnine layouts and are more suitable for editing programatically.
>>>>>>>>>>>>>>
Actually, the new distros are going AGAINST that idea. By creating a held-together-by-thumbtacks-and-bubblegum miasma of proprietory perl scripts and config files and package layouts and dependency files, ad nauseum, they are making it impossible to tweek the system beyond the "out-of-box" configuration. Instead of making the underlying system more elegant and logical, they are simply adding incoherent and non-uniform tools on top of a incoherent and non-uniform system. If a standard GUI interface was created, for say, the Slackware config files, something that would configure everything from sound to X to the kernel, then you'd end up with an elegant, modifable system. (Well, after they clean out
No.
Nautilus is essentially feature compatible with IE6. In other words, its a tuning of what has been done (better) before. OS X, on the other hand has three things going for it...
A) It looks orgasmically good. And its themable!
B) Quartz wipes the floor with X (except remote display)
C) Mac users would never use anything resembling IE.