And because you don't care you can't choose. You're forced to pay what THEY want you to pay, to put up with the quality and support THEY want to provide you.
Intel is the first but not always the best. Their move to 90nm for instance led to Prescott processors hotter than the sun. Granted their move to 65nm was a lot smoother, being first isn't always the goal.
Frankly, even given where I work I'm happy that Intel is turning out good [or at least better] hardware. We all win when technology is getting neatoer.
And for your information: Processors like the x86 series are not byte addressable. They usually load a cacheline and the processor fetches the byte you want.
So instead of doing
movb (%rax),%bl
You'd have to do
movq (%rax),%rbx andq $255,%rbx
Worse yet, if you want the [say] 5th byte of a 64-bit word...
You think we're still living in the IBM PC era where all PCs were the same and "supporting it" was easy because nobody really had huge deviations. Nothing is standard today.
You say my hardware doesn't work in Linux, fix Linux. I say fix your hardware.
I don't see how blaming Linux for the problems that lie directly in your hardware is going to fix anything. If you hate that your hardware doesn't work in Linux, buy hardware from companies which respect your rights. DLINK makes fine network cards that work in Linux, CMIPCI sound cards work flawlessly, Hauppage makes good TV cards, Nvidia based GFX cards are well supported, so on and so forth.
There is no reason you can't build a PC [and on occasion find a laptop] that is fully Linux compatible. That is, other than the fact you won't spend the time to research it and demand that it works with YOUR OS.
Um, the laptop processors were the same as the desktop in the 754 pin category. And in the Socket A they were ALSO compatible with the desktops.
And the move from K7 -> K8 brought about 90nm SOI which greatly reduced power and heat issues. A similarly clocked K8 would easily run 10-15C cooler than a K7 at idle.
AMD has plans to move to 65nm and 45nm. I won't say when [cuz that's secret and frankly I don't remember anyways]. They're just not rolling out a completely new core every other month to avoid wasting time supporting really short lived products.
Conroe seems like a decent design. Until I build a box with one I can't really say. If all the hype is true though it's a good competitor to AMD K8, not a replacement, certainly in a lot of server oriented computing tasks.
My GFX, TV, Sound and Network cards all work in Linux. I educated myself and bought specific stuff [that you can order from any wholesaler] and have been smooth sailing for years.
Oh right, you have to know two things about computers. Shit sorry. I forgot it's hip to be a techno-ignorant-opiniated bastard. if you're going to embrace technology it may pay off to know a thing or two about it. Otherwise, you're at the whim of whatever your corporate masters decide.
No HD in Vista32. BUY BUY BUY Buy new hardware, more more more, buy a new OS, just to play a movie!
I dunno about you, but I don't like the feeling of being owned. I mean it's bad enough I'm legally the property of Canada... hehehe.
Well educate people then. If you see someone bitch about Linux tell them to bitch about the shitty hardware. Put the blame where it ACTUALLY BELONGS.
I mean no amount of bitching at LKML will make Broadcom opensource their drivers. So instead of being a pissy little bitch, actually do something about it. Grassroots man!
See how I'm talking here on/. about educating you? Do this sort of shit to people elsewhere and in real life.
None of your comments have anything to do with GNU/Linux [I say GNU because many things like cameras are dealt with in the userspace].
Linux can handle playing your DVD. The MPAA just doesn't want you to.
Linux can handle your USB camera. The manufacturer just doesn't want you to.
Linux can handle that MP3 player.... etc.
Linux can even handle that wireless connection. Just the manufacturers don't care to let you.
so on so forth.
In Gentoo [of all OSes] it can consistently detect and setup my Intel Wireless device upon bootup [part of udev/coldplug]. If I program an AP in it [in the config] it will associate and setup the IP stack. If not it will try the strongest one and DHCP.
Too fucking bad that your Broadcom device doesn't have drivers. Intel funded their drivers, where the hell is Broadcom with theirs? Enough people use Linux to warrant making them.
Truth be told, the thing that is holding you back from using GNU/Linux is all those devices you take for granted. If the manufacturers gave two shits about your rights to use the device as YOU see fit they'd address the fact that there is a sizable Linux/BSD community and open up some drivers.
That's cute, except one problem. You cannot license public domain software. Therefore, your warranty is moot. You're either liable or you're not.
Implementing a patented algorithm is not a civil offense. Making it available is. However, you could argue [and I dunno how effective this would be] that a public domain implementation is not endorsed by the author. For example, suppose the patent is not valid in Japan. Then clearly the presense of the software does not indicate intent since there are places you can use it legally.
For me, I just don't care. To the best of my ability I don't think I violate any patents and even if I did I wouldn't care.
As for OSS, if an OSS project puts out a novel idea, that idea instantly becomes prior art and nobody can patent it. There is no way for them to be "locked out" of that field. But if somebody else thought of it first, why should OSS be allowed to get a free ride?
Because often the OSS implementor does not know about the patent. Try doing a search for every idea in the Linux kernel. Can't be done. It isn't feasible. So basically what you're telling me is I can sit down and implement something with zero knowledge of your patent, but then you can halt my work because you thought of the idea first [or were the first to publicly claim it]?
You wouldn't claim the colour or other non-important facets of the invention as a claim. If a knockoff adds two legs though chances are your claims include "chair has four legs." so it's a clear valid knock-off.
Software patents though usually read something like "...using a method of notifying a user..." or "...receives a message..." or "...stores data..." and are totally vague and non-specific. Method of notifying a user? That could mean through sound, an alert dialog, print out, beeper, vibration, etc.
Not that there are no valid [I hesitate to use the word good] software patents. Just that there are many really shitty ones. They're vague by design though. To snare as many victims as they can. Which is why us hippies hate them.
Mmm a lot of time would have to pass where some missing link would later become common knowledge.
To use the LZW cases there really isn't any missing link. Basically LZ78 is a dictionary matching algo for compression where it builds up strings of previously seen data and then replaces them with indexes. LZW is a modification where you preload the table with all byte values removing the need for escape sequences [since there are no literals anymore]. LZW is an obvious adaptation to LZ78 and any serious practitioner would have thought about it.
I'm not trying to dismiss all patents outright. Merely I'm suggesting that the vast majority of patents out there are not legitimate and should be tossed out.
For instance, Certicom has patents on a Crypto ALU where they have your basic ALU with larger registers and they do polynomial math. That to be is a no brainer. ALUs already existed prior to their patent and crypto algorithms [e.g. bignum stuff] already existed. Throwing the two together is obvious. Yet they hold a patent on it. Now it should be a patent on specific ALU dimensions and configuration but history has shown that patent holders will try to be vague as possible so they can mold their "specific" invention onto whatever competitors come up with.
While timing does have some significance it doesn't mean it was any less obvious. There is also the need for such a solution. If you never had to solve it before, it doesn't mean the solution isn't something anyone else couldn't come up with.
Ignoring the fact that patents cover very specific implementations of ideas (e.g. this is why you have to claim subroutines as parts of the invention)....
Most software algorithms [and in many cases hardware] are just evolutions of previous algorithms. Take LZW. Take any one of the millions held collectively by Apple, IBM, Microsoft and the like. Patents which are truly original and non-obvious are the exception not the rule. And given that OSS developers don't have the money to invest in patents it's just a way to lock them out.
It isn't that OSS people aren't creative and coming up with new and original ideas. It's just faced against an opposition with an unlimited budget and who takes advantage of the government it seems very counter-culture. So literally you have OSS developers scratching their heads saying "this has a patent?" and then having to do without.
Though occasionally this helps out. Look at PNG. Because Unisys are a bunch of asshat patent trolls they tried to pull LZW from under our feet. Then a group got together and invented PNG. Boom a better replacement without patents. Similarly with Vorbis.
But in a lot of cases this isn't possible. If I [were evil and had a] patent a SSA tree optimizer I'd FUD all GCC developers out of using it. Working around it may be possible but the solution may very well be less optimal.
I really think we'll see patent reform before that point. The real question is how long will that take to start?
Most large enough companies have a "portfolio" of totally abusive patents which they shouldn't have been granted. It's only a matter of time before all stores are taco bell if you catch my drift.
Think about it this way though, if some part-time OSS developer can stumble on the same idea how non-obvious is it?
There is a difference between "not done yet" and "non-obvious". The non-obvious leap is something where even if you knew about the problem you wouldn't have likely found the solution.
Take LZW for example, it's a rather straightforward addition to LZ78. In my books any competent comp.sci student would try that too. It's obvious. By the same token, it wasn't obvious how to get MP3 like efficiency 20 years ago in encoding audio [given that the state of the art back then was ADPCM, CELP and u-Law].
Most OSS developers couldn't implement an MP3 codec [hence the lack of options in choosing such a lib] but most could implement LZW, specially when given LZ78.
I'm not that interested. Just saying if you post an article about something visual you should have a visual.
It's like writing a 1000 word story on "that weird sound heard over in Orange county" then not actually having the clip of audio. In this day and age I think we can at least put a screenshot up of a new UI.
What's with stories that talk about something clearly visual and then not have screenshots or photos?
It's a story about a new UI... and there isn't a screenshot on the first page of the article [... we'll ignore the "split a 1000 word article into 300 pages to make ad revenue...]
Almost as bad as the stories linked to by fark... "Car rolled 98 times and then burst into flames" and the only picture is of some old dude taking viagra or whatever ad the "news" site has...
At the point where you legitmately need 800 processors... you're probably big enough to borrow the money to build it. Most small million dollar firms I know of run off of simple dual core AMD64s. At the point where you need a 50 million dollar cluster... you're probably big enough to buy one.
It sounds all nice in practice to buy "cheap time" but it's cheap for a reason
1. You don't own the boxes 2. It's not leased, they could go down at any moment 3. All your data is remote 4. You're subject to your network and their network and all the networks in between 5. You don't own the boxes so you can't add hardware, change key software components, etc...
There are a lot of downsides to it and running a business from it is stupid.
tom@bigbox ~ $ cat/proc/cpuinfo | grep "model name" model name : AMD Opteron(tm) Processor 885 model name : AMD Opteron(tm) Processor 885 model name : AMD Opteron(tm) Processor 885 model name : AMD Opteron(tm) Processor 885
1.75GB of ram? tom@bigbox ~ $ free total used free shared buffers cached Mem: 6104840 700028 5404812 0 62092 399356
Slashdot Mirror
And because you don't care you can't choose. You're forced to pay what THEY want you to pay, to put up with the quality and support THEY want to provide you.
Congrats, you're a good consumer.
Tom
Intel is the first but not always the best. Their move to 90nm for instance led to Prescott processors hotter than the sun. Granted their move to 65nm was a lot smoother, being first isn't always the goal.
Frankly, even given where I work I'm happy that Intel is turning out good [or at least better] hardware. We all win when technology is getting neatoer.
Tom
Newsflash: A lot of data is still 8 bits wide.
And for your information: Processors like the x86 series are not byte addressable. They usually load a cacheline and the processor fetches the byte you want.
So instead of doing
movb (%rax),%bl
You'd have to do
movq (%rax),%rbx
andq $255,%rbx
Worse yet, if you want the [say] 5th byte of a 64-bit word...
movq (%rax),%rbx
shrq $40,%rbx
andq $255,%rbx
That's clearly a winning idea!
Tom
Well I don't get you people.
You think we're still living in the IBM PC era where all PCs were the same and "supporting it" was easy because nobody really had huge deviations. Nothing is standard today.
You say my hardware doesn't work in Linux, fix Linux. I say fix your hardware.
I don't see how blaming Linux for the problems that lie directly in your hardware is going to fix anything. If you hate that your hardware doesn't work in Linux, buy hardware from companies which respect your rights. DLINK makes fine network cards that work in Linux, CMIPCI sound cards work flawlessly, Hauppage makes good TV cards, Nvidia based GFX cards are well supported, so on and so forth.
There is no reason you can't build a PC [and on occasion find a laptop] that is fully Linux compatible. That is, other than the fact you won't spend the time to research it and demand that it works with YOUR OS.
Tom
Why does support OSS mean you have to write the drivers yourself?
Tom
Um, the laptop processors were the same as the desktop in the 754 pin category. And in the Socket A they were ALSO compatible with the desktops.
And the move from K7 -> K8 brought about 90nm SOI which greatly reduced power and heat issues. A similarly clocked K8 would easily run 10-15C cooler than a K7 at idle.
AMD has plans to move to 65nm and 45nm. I won't say when [cuz that's secret and frankly I don't remember anyways]. They're just not rolling out a completely new core every other month to avoid wasting time supporting really short lived products.
Conroe seems like a decent design. Until I build a box with one I can't really say. If all the hype is true though it's a good competitor to AMD K8, not a replacement, certainly in a lot of server oriented computing tasks.
Tom
Hint: Stop buying monopolyware.
... hehehe.
My GFX, TV, Sound and Network cards all work in Linux. I educated myself and bought specific stuff [that you can order from any wholesaler] and have been smooth sailing for years.
Oh right, you have to know two things about computers. Shit sorry. I forgot it's hip to be a techno-ignorant-opiniated bastard. if you're going to embrace technology it may pay off to know a thing or two about it. Otherwise, you're at the whim of whatever your corporate masters decide.
No HD in Vista32. BUY BUY BUY Buy new hardware, more more more, buy a new OS, just to play a movie!
I dunno about you, but I don't like the feeling of being owned. I mean it's bad enough I'm legally the property of Canada
Tom
Well educate people then. If you see someone bitch about Linux tell them to bitch about the shitty hardware. Put the blame where it ACTUALLY BELONGS.
/. about educating you? Do this sort of shit to people elsewhere and in real life.
I mean no amount of bitching at LKML will make Broadcom opensource their drivers. So instead of being a pissy little bitch, actually do something about it. Grassroots man!
See how I'm talking here on
Tom
Imagine this, people like GNU/Linux for what it is, stop buying crap hardware that works only in WinTel and the hardware folk either adjust or die.
That is of course, if people have the courage of their convictions.
Tom
N.B. I game on a PS2/Xbox. My PC is for work/media and runs Gentoo Linux. I don't compromise with a Win32 boot. I have integrity.
None of your comments have anything to do with GNU/Linux [I say GNU because many things like cameras are dealt with in the userspace].
... etc.
Linux can handle playing your DVD. The MPAA just doesn't want you to.
Linux can handle your USB camera. The manufacturer just doesn't want you to.
Linux can handle that MP3 player.
Linux can even handle that wireless connection. Just the manufacturers don't care to let you.
so on so forth.
In Gentoo [of all OSes] it can consistently detect and setup my Intel Wireless device upon bootup [part of udev/coldplug]. If I program an AP in it [in the config] it will associate and setup the IP stack. If not it will try the strongest one and DHCP.
Too fucking bad that your Broadcom device doesn't have drivers. Intel funded their drivers, where the hell is Broadcom with theirs? Enough people use Linux to warrant making them.
Truth be told, the thing that is holding you back from using GNU/Linux is all those devices you take for granted. If the manufacturers gave two shits about your rights to use the device as YOU see fit they'd address the fact that there is a sizable Linux/BSD community and open up some drivers.
Tom
was OSS about marketshare?
If you build it, they will come.
Tom
That's cute, except one problem. You cannot license public domain software. Therefore, your warranty is moot. You're either liable or you're not.
Implementing a patented algorithm is not a civil offense. Making it available is. However, you could argue [and I dunno how effective this would be] that a public domain implementation is not endorsed by the author. For example, suppose the patent is not valid in Japan. Then clearly the presense of the software does not indicate intent since there are places you can use it legally.
For me, I just don't care. To the best of my ability I don't think I violate any patents and even if I did I wouldn't care.
Tom
GCC is free software right?
So are the hundreds of other tools?
Skippy.
Tom
Which is why the patent system will collapse in on itself eventually. That or all companies will merge. Take your pick.
Tom
As for OSS, if an OSS project puts out a novel idea, that idea instantly becomes prior art and nobody can patent it. There is no way for them to be "locked out" of that field. But if somebody else thought of it first, why should OSS be allowed to get a free ride?
Because often the OSS implementor does not know about the patent. Try doing a search for every idea in the Linux kernel. Can't be done. It isn't feasible. So basically what you're telling me is I can sit down and implement something with zero knowledge of your patent, but then you can halt my work because you thought of the idea first [or were the first to publicly claim it]?
Tom
You wouldn't claim the colour or other non-important facets of the invention as a claim. If a knockoff adds two legs though chances are your claims include "chair has four legs." so it's a clear valid knock-off.
Software patents though usually read something like "...using a method of notifying a user..." or "...receives a message..." or "...stores data..." and are totally vague and non-specific. Method of notifying a user? That could mean through sound, an alert dialog, print out, beeper, vibration, etc.
Not that there are no valid [I hesitate to use the word good] software patents. Just that there are many really shitty ones. They're vague by design though. To snare as many victims as they can. Which is why us hippies hate them.
Tom
Mmm a lot of time would have to pass where some missing link would later become common knowledge.
To use the LZW cases there really isn't any missing link. Basically LZ78 is a dictionary matching algo for compression where it builds up strings of previously seen data and then replaces them with indexes. LZW is a modification where you preload the table with all byte values removing the need for escape sequences [since there are no literals anymore]. LZW is an obvious adaptation to LZ78 and any serious practitioner would have thought about it.
I'm not trying to dismiss all patents outright. Merely I'm suggesting that the vast majority of patents out there are not legitimate and should be tossed out.
For instance, Certicom has patents on a Crypto ALU where they have your basic ALU with larger registers and they do polynomial math. That to be is a no brainer. ALUs already existed prior to their patent and crypto algorithms [e.g. bignum stuff] already existed. Throwing the two together is obvious. Yet they hold a patent on it. Now it should be a patent on specific ALU dimensions and configuration but history has shown that patent holders will try to be vague as possible so they can mold their "specific" invention onto whatever competitors come up with.
While timing does have some significance it doesn't mean it was any less obvious. There is also the need for such a solution. If you never had to solve it before, it doesn't mean the solution isn't something anyone else couldn't come up with.
Tom
Ignoring the fact that patents cover very specific implementations of ideas (e.g. this is why you have to claim subroutines as parts of the invention)....
Most software algorithms [and in many cases hardware] are just evolutions of previous algorithms. Take LZW. Take any one of the millions held collectively by Apple, IBM, Microsoft and the like. Patents which are truly original and non-obvious are the exception not the rule. And given that OSS developers don't have the money to invest in patents it's just a way to lock them out.
It isn't that OSS people aren't creative and coming up with new and original ideas. It's just faced against an opposition with an unlimited budget and who takes advantage of the government it seems very counter-culture. So literally you have OSS developers scratching their heads saying "this has a patent?" and then having to do without.
Though occasionally this helps out. Look at PNG. Because Unisys are a bunch of asshat patent trolls they tried to pull LZW from under our feet. Then a group got together and invented PNG. Boom a better replacement without patents. Similarly with Vorbis.
But in a lot of cases this isn't possible. If I [were evil and had a] patent a SSA tree optimizer I'd FUD all GCC developers out of using it. Working around it may be possible but the solution may very well be less optimal.
Tom
I really think we'll see patent reform before that point. The real question is how long will that take to start?
Most large enough companies have a "portfolio" of totally abusive patents which they shouldn't have been granted. It's only a matter of time before all stores are taco bell if you catch my drift.
Tom
Two words... "public domain".
There is no license. So the users are entirely responsible themselves for the use of the software. It also means they can benefit the most.
Tom
Think about it this way though, if some part-time OSS developer can stumble on the same idea how non-obvious is it?
There is a difference between "not done yet" and "non-obvious". The non-obvious leap is something where even if you knew about the problem you wouldn't have likely found the solution.
Take LZW for example, it's a rather straightforward addition to LZ78. In my books any competent comp.sci student would try that too. It's obvious. By the same token, it wasn't obvious how to get MP3 like efficiency 20 years ago in encoding audio [given that the state of the art back then was ADPCM, CELP and u-Law].
Most OSS developers couldn't implement an MP3 codec [hence the lack of options in choosing such a lib] but most could implement LZW, specially when given LZ78.
Tom
I'm not that interested. Just saying if you post an article about something visual you should have a visual.
It's like writing a 1000 word story on "that weird sound heard over in Orange county" then not actually having the clip of audio. In this day and age I think we can at least put a screenshot up of a new UI.
Tom
What's with stories that talk about something clearly visual and then not have screenshots or photos?
... and there isn't a screenshot on the first page of the article [... we'll ignore the "split a 1000 word article into 300 pages to make ad revenue ...]
... "Car rolled 98 times and then burst into flames" and the only picture is of some old dude taking viagra or whatever ad the "news" site has ...
It's a story about a new UI
Almost as bad as the stories linked to by fark
Tom
At the point where you legitmately need 800 processors ... you're probably big enough to borrow the money to build it. Most small million dollar firms I know of run off of simple dual core AMD64s. At the point where you need a 50 million dollar cluster ... you're probably big enough to buy one.
It sounds all nice in practice to buy "cheap time" but it's cheap for a reason
1. You don't own the boxes
2. It's not leased, they could go down at any moment
3. All your data is remote
4. You're subject to your network and their network and all the networks in between
5. You don't own the boxes so you can't add hardware, change key software components, etc...
There are a lot of downsides to it and running a business from it is stupid.
Tom
1.7Ghz Xeon?
/proc/cpuinfo | grep "model name"
:-)
tom@bigbox ~ $ cat
model name : AMD Opteron(tm) Processor 885
model name : AMD Opteron(tm) Processor 885
model name : AMD Opteron(tm) Processor 885
model name : AMD Opteron(tm) Processor 885
1.75GB of ram?
tom@bigbox ~ $ free
total used free shared buffers cached
Mem: 6104840 700028 5404812 0 62092 399356
No thank you.
Tom