Seriously, I might have misunderstood feminism, but I thought the point was to increase choices and achievement not force new equally restrictive constraints on women while also screwing over men to achieve it.
> Seriously, I might have misunderstood feminism, but I thought the point was to increase choices and achievement while also screwing over men to achieve it.
Only some things can be fixed via a ucode patch, others cannot. See AMD's TLB errata for an example of something that cannot. Other things can be fixed by disabling a feature, but disabling that feature might cost performance. Once again, see AMD's TLB errata for an example. Still other things can be worked around in the OS, sometimes for negligible performance loss, sometimes not. The Intel F00F bug was a perfect example of something that could be worked around in the OS with no performance loss, and the AMD TLB errata had an OS workaround too which incurred a small (1%?) performance loss. Other things have almost no workaround, and require Intel or AMD to recall silicon and give out new processors. Intel's Pentium FDIV bug was a good example of that. It depends entirely on what piece of the chip is at fault.
If something can be fixed in ucode for a negligible performance loss, or worked around in the OS for a negligible performance loss, that's the best-case scenario for Intel. In that case it's just a matter of getting BIOSes/OSes updated and patches rolled out to OEMs.
A sale of is IP is a sale just like any other item which give me full rights to do anything I want it.
Such as...
The courts have ruled on that again and again when it comes to the rights of reselling items.
Reselling is not "doing anything I want", it is an application of first-sale doctrine.
The only thing I am not legally allowed to do is redistribute a copy of the item in question, because every single piece of IP produced automatically gets assigned a copyright that prevents anyone but the owner of the copyright to do just that. That copyright can in turn be sold and traded, transfering the rights of copying the item for redistribution.
However, every specific copy is consider its own and can be resold.
Once again, you're specifically addressing resale, which is different than "doing anything I want".
Wrong, you can not sell rights to use IP. You can only sell the IP itself while maintaining the exclusive right to copy and sell more of it.
You're being pedantic with words. Maybe that's important in a court, but this is Slashdot, and no one here gets it unless you do it my way. We're saying the exact same thing, except you're using proper terminology and I'm putting it in terms that Slashdotters understand.
If the definition of "IP" is something that can be easily reproduced and distributed, you haven't "bought" IP if you are restricted from reproducing it. What have you bought? A can't-be-further-copied copy of the IP? That doesn't make much sense to people around here. What you've effectively done purchased the right to use the IP. Effectively. (Not actually, as you so nicely pointed out in your post). In your words, you've purchased a copy of the IP which can't be further copied because of copyright restrictions. In my words, you're not purchasing the IP, you're purchasing the right to use the IP.
The difference is, when you put it my way, Slashdotters can't use a bunch of fallacious arguments about theft and deprivation. Those same fallacious arguments don't work on your words either (because of copyright restrictions), but no one here buys into copyrights, so it's impossible to have dialogue about how to fix the actual problem.
The actual problem being how to allow IP creators to be compensated for their work.
I'm sick of people endlessly spewing this argument in these discussions.
With Intellectual Property (IP), there is no physical item. Any "sale" of IP is the sale of a right to use the IP, not a transfer of the IP itself. If a company wants to transfer ownership of the IP, a company is transferring the ability to sell rights to use the IP. Any argument based on "deprivation" is wrong and misleading because it's literally impossible.
Theft of IP is gaining use of the IP, without paying the owner of the IP (the only party who is able to sell rights to use the IP) for the use of that IP.
Right on. The CPU makers do have two ways to "throw more transistors in" which can help: more cache, and more cores. But there are problems with both of these techniques.
The problem with more cache is that it only scales so well in terms of performance, power, and area. With a larger cache, you get:
larger die size, which means fewer die per wafer which increases manufacturing cost
any particular die is now more likely it is to have a defect, lowering yields and thus further increasing cost
leakage power increases, so now your 64MB of cache burns 2-3W just for being there
bigger cache == longer latency to access it, so you get diminishing returns and eventually your cache will start to perform more like a mildly faster DRAM
Finally, more cache doesn't give you a linear speedup (meaning, doubling the cache doesn't give you 2x the performance), because it only helps the latency of loads and stores, it doesn't help you get through massive datasets like truly parallelizing your application and running it on more cores does.
Now what about more cores? More cores are similar in technique to "more pixel shaders" or "more vertex shaders" because they are directly helping with the problem that your CPUs are working on. But like Parent and Grandparent posts said, most times, the problems CPUs are working on are not easily parallelizable. Not only is it hard to write parallel code, dealing with concurrency and locking properly, but with lots of parallel problems there exists only so much inherent parallelism in the problem or algorithm itself. Even if you parallelized everything that could be parallelized, then the serial parts of your problem will start to dominate your computation time. See Amdahl's Law.
I think you need a history lesson. When AMD released the original Opteron in late 2003/early 2004, it had numerous desirable, innovative features that Intel's offerings at the time did not have:
64 bit support in the form of AMD64
An on-die memory controller
HyperTransport point-to-point interconnect for non-coherent I/O links and coherent socket-to-socket links
Better power numbers than P4
Better performance than P4
AMD has alleged that Intel used its monopoly position to exclude Opteron and other K8 derivatives (Athlon64) from major OEMs for 2 years, from its release in late 03 until sometime in 2005, when the antitrust allegations were filed. During that time, the problem was not poorer, "cloned" products which offered no advantages over Intel's. Basically, anyone who followed the x86 processor market during that time knew that Opteron/Athlon64 was better than P4, for a competitive price.
Linus was in school until 1996, during which time he was working on the first incarnations of Linux which you downloaded. Oh, and it turned into his Master's thesis too. Did you even both to read the part of the post where I said "started their major projects during college or grad school" ? And then as soon as he finished school, he got a job at Transmeta.
You have completely missed *my* point - that people can work on their hobby projects in school, or while holding a day job, but development can't continue like that forever. I guess I've been trolled by an AC. Oh well.
I don't know why you named those four people; at least three of those four have been or are currently being compensated for their most famous "free" projects.
Linus went to Transmeta in 1996 shortly after his Master's degree, and Transmeta paid him to work on Linux.
Ian Murdock founded Debian during college, then was a part time student and staff programmer at the University of Arizona before founding Progeny (and presumably getting VC funding for it). One thing Progeny did was produce a commercially-saleable derivative of Debian. Then after that he went to Sun.
Larry Wall was at JPL after grad school, and I'm sure he's made plenty of money off the Perl books he publishes through O'Reilly.
I don't know about Stallman; he's some sort of communo-socio-anarchist and may survive on ramen handouts from the local organic food store, so you might have me there.
A common thread among those people is that they all started their major projects during college or grad school and found financial backing as they were leaving academia. Or in Larry Wall's case, he had a day job at JPL while working on Perl. I think you'll admit that college/grad student life can't realistically go on forever. Eventually your parents will stop giving you money and/or the university will stop paying your room and board, and you'll have to find a "real job" to support yourself and your family. I think lots of people in the open-source community are employed by the likes of IBM, Red Hat, Oracle, OSDL, etc. for their work. No, I don't feel like finding more references.
The message might be that we need to fund more people in grad school to work on pet projects, or that Microsoft needs to fund them, but in general I agree with Mr. Gates - development on large-scale projects can't continue indefinitely without some sort of compensation.
x86, which is the classic CISC, is also a variable-length ISA. That means certain instructions take just a single byte to encode, compared with a fixed 4 bytes on the most common RISCs. This can be a factor in instruction cache size/effectiveness. Fewer bytes for instructions == more instructions fit in the ICache == ICache is more effective. I don't have any numbers, but I would expect the average instruction length on CISC to be in the 10s of % smaller than RISC. That means either greater performance, or lower power, or both. Perhaps enough of each to offset the greater power/die area required to decode these variable-length instructions.
It's not a big factor, but combined with the other points that have been made in this thread (about how CISCs translate to RISC internally, etc., the advantages of using a very-mature x86 OS/software stack), there are a number of reasons why an x86 embedded processor might be good.
Unsold cars at a car dealership are "property" too, but the dealership does not get taxed on this "property" if someone steals one of them.
Now, the problem with this analogy (as I'm sure would have been pointed out by 10 different people and modded +5 Insightful) is that with Intellectual Property, the original property owner is not deprived of anything if someone steals the IP, because IP costs $0 to replicate. Unlike a car.
However, the analogy isn't totally worthless. The point isn't the deprivation or lack thereof of said property holder - it's that someone else acquired something of value that the property holder owns, without compensating the property owner. That view applies to both cars and IP.
I wish people would stop analogizing Intellectual Property to physical things, because it inevitably leads to someone pointing out that the owner of the IP doesn't lose anything if someone "steals" the IP, and thus it's not theft, and thus it's not wrong.
The point is that someone obtained something of value for free, without compensating the creator of that value. Whether or not anyone was "deprived" of anything, or something was "lost", is completely immaterial.
You obviously didn't read the thread that GP linked to. Things that computers address are measured in powers of 2, like RAM. Measurements of things like network speed or CPU frequency can use base-10 because computers don't need to address every bit in a network transmission or every cycle in a CPU clock. Read this post from that thread if you're still confused.
Developers claim adding UI options increases dev time beyond what is desired, increases chances of bugs, and is not worth the effort. Project forks. New developers put in work and dev time to iron out new UI options and/or features that original developers refused to add.
At this point, the fork presumably has a decent number of features that users have been demanding of the original. Either users will move to the fork, or the fork's changes will merge back to the original project since 1) they have been vetted thoroughly enough in the fork to appease the original developers, and 2) the original developers see that the fork's features are desirable enough that it was, in fact, worth the extra dev time and effort to implement them.
Either way, users and fork maintainers can say "we told you so" to the original developers.
This story reminded me of 1984's Ministry of Truth, which regulary "edited" history to match the current political scene. Writing stuff in Wikipedia makes it true.
I think I was too generous when I said "a lot of easily-parellelizable problems out there". The rest of them (besides graphics and physics) are in the HPC community - protein folding, SETI, that sort of thing - the stuff people run on BlueGene. Not something your average, everyday user is going to experience, at least not in the near future.
So then, what's the real benefit of CUDA == GPGPU == general purpose processing on a GPU, if it only speeds up a certain class of general-purpose problems? It starts to become less general-purpose then, don't you think? Cheaper computing power to do protein folding or SETI, sure, but then? Maybe we'll find more problems that are helped by GPGPU, and nVdia and ATI will have the infrastructure in place to support them. But like I said, I'm a bit skeptical.
That's exactly why I'm skeptical of CUDA in general. Rendering graphics is one of those "embarrassingly parallel" operations, where for every frame rendered, each triangle, vertex, and pixel is independent of the others. I don't know much about physics processing, but I'm guessing it's the same. Simply throwing say, 4x more hardware at the problem gives you pretty much the ideal 4x speedup. Once you introduce random data dependencies into the problem, though, parallelization is much harder and you don't get the same speedups.
There are a lot of easily-parallelizable problems out there, which is great...but then again, there are a lot of not-so-easily-parallelizable problems out there too. I don't think CUDA would do any better than a general-purpose CPU on the latter class of problems.
1. In order to point out that the games must target various demographics.
So, to use the OP as a case study, math games target mostly boys?
Aside from the above, not every damn word a person says is meant with an agenda. You're reading a ridiculous amount into the statement. Most people do not carefully consider their every sentence and think of all possible interpretations of it, snipping out every bit of detail that is not immediately relevant, compressing it into some robotic minimal communication. Most people just say things or write things.
Possible, but given the fact that this community has many times debated the very thing which the OP alluded to, I'd consider it a remote possibility that the thought did not cross OP's mind.
I remember a game called Math Maze, and it's definitely not the same one described above, but it was a staple among the boys of my class. The girls preferred Mathville, for what it's worth. Maybe by throwing that one out there you won't accuse me of wild sexism. The one black kid (male) liked a third game that involved speed (the others did not) called Math Race. NOW I AM A RACIST! The East Asian kid (male) liked Math Maze too. I'm totally playing into the nerdy Asian stereotype. Societal progress has stopped entirely.
The purpose of your above paragraph is explicitly to discuss differing demographics, so of course it's fine to state differences. OP made a seemingly offhand remark, singling out a single demographic, in an unrelated paragraph.
The French teacher wanted us to play this text adventure in French, but within the first half hour somebody had worked out the minimal set of steps to win the game with full points, and distributed it to everyone so they could beat that quickly and go back to playing the games they wanted.
Would it have been so bad if, instead of "somebody", I said "I" in the last sentence? I mean, it's not really relevant.
You are recounting the story and speaking in the first person; of course it's fine to use "I". If you are an East Asian male, and instead you had said "I, an East Asian male", then yes, that's flamebait.
Perhaps my motivation is show how much smarter I am than all the other plebes.
called Assassin that some frats and/or dorms play at UIUC. A guy locked himself in his room to avoid an assassin, who then camped right outside his door, ready to strike the instant he opened the door. Supposedly, the guy really had to do #2, but rather than open the door and lose, he crapped out his 2nd floor window.
The number 1 reason why x86 processors burn this much power is the incredibly dense feature set they implement, a feature set that was designed with servers/workstations/desktops in mind, not.01W cell phones.
First and foremost, they implement a variable-length CISC instruction set, which complicates nearly everything in the front end of the machine (branch prediction to instruction decode), and some stuff in the back end too. Add on multiple operating modes, multiple paging modes implemented in hardware, advanced memory subsystems with hardware misalign support and relatively strong store ordering (vs. ARM at least), virtualization, SIMD, backward compatibility/legacy features. That's not to say that ARM doesn't have any of these, but from the little I know about it, they are pared down or specifically designed with an extremely low-power core in mind, not a server. And then, with an x86 you have to crank up the clock speed because of the pipeline depth.
That said, it can't hurt the x86 market to get TDP down an order of magnitude from ~50-100W to 5W.
Also, from the little I know about cell phones, there are typically two processors, a "baseband DSP" which handles the radio, and an "application processor" which runs the UI, apps, Java, etc. How much power does the baseband burn? With an x86, maybe it's possible to get rid of it entirely.
Sorry, infringing the copyright on a work that is given away for free precludes 4.
Nope. Ever heard of advertising revenue? How about non-free items included with the free ones? Ever hear of a company called Red Hat?
The nature of the copyrighted work is that it is trivial and has about a $20 value
That's about the value of a music CD of your favorite band. Now, someone remind me, what is the fine for illegally redistributing copyrighted material?
and in a Chinese court the case would be thrown out.
Both Intel and AMD used a shared cache design with a single connection to the system bus (FSB and HT, respectively).
In AMD's case, the shared cache sits on the other side of the fully-connected crossbar, which allows intra-core communication to happen without using HyperTransport at all. So yes, it's shared, but each core has its own "port" to it and can access it independently. Same deal for the Intel shared L2. The phrase "single connection to the system bus" is misleading because it implies a bottleneck where there (most times) isn't one.
In the case of AMD, it also means a shared memory controller.
The memory controller on Intel systems is shared as well. It's just sitting on a different chip, across the FSB.
In fact, once you move to multi-socket, AMD systems generally have as many memory controllers as there are sockets, and with NUMA optimizations in modern OS's, it's likely that a core will only ever need the memory controller which it's closest to. In Intel systems, all cores on all sockets still share a single memory controller.
Every time someone makes an analogy to printing presses and books, they get +5 Insightful. Here are the fundamental flaws with that analogy:
* Printing presses still produce physical things.
* People like that because they want to read books in physical form, not on a computer screen. Thus the physical item still has value to them.
* It still costs some money and some effort to duplicate an entire book on a copier. It's possible to scan it into electronic form, but see the previous bullet for why no one cares about that.
If people hated reading books in physical form and instead the electronic form was the medium of choice, I'm pretty sure you'd see similar things happening to the book industry. Fortunately for them, electronic books just plain suck. Also, I think it's pretty obvious that they physical form of music has absolutely no value to a consumer over the digital form, for 99% of consumers. In fact, the digital form might have more value because it can be easily transferred to a portable music player.
A better analogy is newspapers. People don't mind reading their news on a computer screen, even if it's basically the exact same content as a physical newspaper and there's no value-adding going on by virtue of it being an electronic medium (video clips, animations, interactivity). I'm pretty sure you're aware that physical newspapers have taken a hit in recent times because people can get the same information online for "free" (ad-supported). Of course, newspapers have been ad-supported for a long time, so instead of just disappearing altogether, the ads just went from physical form into online form. Physical-form music has *never* been ad-supported. Maybe all this means is that any physical-form music will be ad-supported. Hey, maybe I've just stumbled onto a brilliant business idea!
Slashdot Mirror
Seriously, I might have misunderstood feminism, but I thought the point was to increase choices and achievement not force new equally restrictive constraints on women while also screwing over men to achieve it.
> Seriously, I might have misunderstood feminism, but I thought the point was to increase choices and achievement while also screwing over men to achieve it.
There, fixed that for you. :)
Only some things can be fixed via a ucode patch, others cannot. See AMD's TLB errata for an example of something that cannot. Other things can be fixed by disabling a feature, but disabling that feature might cost performance. Once again, see AMD's TLB errata for an example. Still other things can be worked around in the OS, sometimes for negligible performance loss, sometimes not. The Intel F00F bug was a perfect example of something that could be worked around in the OS with no performance loss, and the AMD TLB errata had an OS workaround too which incurred a small (1%?) performance loss. Other things have almost no workaround, and require Intel or AMD to recall silicon and give out new processors. Intel's Pentium FDIV bug was a good example of that. It depends entirely on what piece of the chip is at fault.
If something can be fixed in ucode for a negligible performance loss, or worked around in the OS for a negligible performance loss, that's the best-case scenario for Intel. In that case it's just a matter of getting BIOSes/OSes updated and patches rolled out to OEMs.
A sale of is IP is a sale just like any other item which give me full rights to do anything I want it.
Such as...
The courts have ruled on that again and again when it comes to the rights of reselling items.
Reselling is not "doing anything I want", it is an application of first-sale doctrine.
The only thing I am not legally allowed to do is redistribute a copy of the item in question, because every single piece of IP produced automatically gets assigned a copyright that prevents anyone but the owner of the copyright to do just that. That copyright can in turn be sold and traded, transfering the rights of copying the item for redistribution.
However, every specific copy is consider its own and can be resold.
Once again, you're specifically addressing resale, which is different than "doing anything I want".
Wrong, you can not sell rights to use IP. You can only sell the IP itself while maintaining the exclusive right to copy and sell more of it.
You're being pedantic with words. Maybe that's important in a court, but this is Slashdot, and no one here gets it unless you do it my way. We're saying the exact same thing, except you're using proper terminology and I'm putting it in terms that Slashdotters understand.
If the definition of "IP" is something that can be easily reproduced and distributed, you haven't "bought" IP if you are restricted from reproducing it. What have you bought? A can't-be-further-copied copy of the IP? That doesn't make much sense to people around here. What you've effectively done purchased the right to use the IP. Effectively. (Not actually, as you so nicely pointed out in your post). In your words, you've purchased a copy of the IP which can't be further copied because of copyright restrictions. In my words, you're not purchasing the IP, you're purchasing the right to use the IP.
The difference is, when you put it my way, Slashdotters can't use a bunch of fallacious arguments about theft and deprivation. Those same fallacious arguments don't work on your words either (because of copyright restrictions), but no one here buys into copyrights, so it's impossible to have dialogue about how to fix the actual problem.
The actual problem being how to allow IP creators to be compensated for their work.
I'm sick of people endlessly spewing this argument in these discussions.
With Intellectual Property (IP), there is no physical item. Any "sale" of IP is the sale of a right to use the IP, not a transfer of the IP itself. If a company wants to transfer ownership of the IP, a company is transferring the ability to sell rights to use the IP. Any argument based on "deprivation" is wrong and misleading because it's literally impossible.
Theft of IP is gaining use of the IP, without paying the owner of the IP (the only party who is able to sell rights to use the IP) for the use of that IP.
The plural of "datum" is not "proof" either.
http://everything2.com/index.pl?node_id=163824
Right on. The CPU makers do have two ways to "throw more transistors in" which can help: more cache, and more cores. But there are problems with both of these techniques.
The problem with more cache is that it only scales so well in terms of performance, power, and area. With a larger cache, you get:
Finally, more cache doesn't give you a linear speedup (meaning, doubling the cache doesn't give you 2x the performance), because it only helps the latency of loads and stores, it doesn't help you get through massive datasets like truly parallelizing your application and running it on more cores does.
Now what about more cores? More cores are similar in technique to "more pixel shaders" or "more vertex shaders" because they are directly helping with the problem that your CPUs are working on. But like Parent and Grandparent posts said, most times, the problems CPUs are working on are not easily parallelizable. Not only is it hard to write parallel code, dealing with concurrency and locking properly, but with lots of parallel problems there exists only so much inherent parallelism in the problem or algorithm itself. Even if you parallelized everything that could be parallelized, then the serial parts of your problem will start to dominate your computation time. See Amdahl's Law.
I think you need a history lesson. When AMD released the original Opteron in late 2003/early 2004, it had numerous desirable, innovative features that Intel's offerings at the time did not have:
AMD has alleged that Intel used its monopoly position to exclude Opteron and other K8 derivatives (Athlon64) from major OEMs for 2 years, from its release in late 03 until sometime in 2005, when the antitrust allegations were filed. During that time, the problem was not poorer, "cloned" products which offered no advantages over Intel's. Basically, anyone who followed the x86 processor market during that time knew that Opteron/Athlon64 was better than P4, for a competitive price.
Linus was in school until 1996, during which time he was working on the first incarnations of Linux which you downloaded. Oh, and it turned into his Master's thesis too. Did you even both to read the part of the post where I said "started their major projects during college or grad school" ? And then as soon as he finished school, he got a job at Transmeta.
You have completely missed *my* point - that people can work on their hobby projects in school, or while holding a day job, but development can't continue like that forever. I guess I've been trolled by an AC. Oh well.
I don't know why you named those four people; at least three of those four have been or are currently being compensated for their most famous "free" projects.
A common thread among those people is that they all started their major projects during college or grad school and found financial backing as they were leaving academia. Or in Larry Wall's case, he had a day job at JPL while working on Perl. I think you'll admit that college/grad student life can't realistically go on forever. Eventually your parents will stop giving you money and/or the university will stop paying your room and board, and you'll have to find a "real job" to support yourself and your family. I think lots of people in the open-source community are employed by the likes of IBM, Red Hat, Oracle, OSDL, etc. for their work. No, I don't feel like finding more references.
The message might be that we need to fund more people in grad school to work on pet projects, or that Microsoft needs to fund them, but in general I agree with Mr. Gates - development on large-scale projects can't continue indefinitely without some sort of compensation.
x86, which is the classic CISC, is also a variable-length ISA. That means certain instructions take just a single byte to encode, compared with a fixed 4 bytes on the most common RISCs. This can be a factor in instruction cache size/effectiveness. Fewer bytes for instructions == more instructions fit in the ICache == ICache is more effective. I don't have any numbers, but I would expect the average instruction length on CISC to be in the 10s of % smaller than RISC. That means either greater performance, or lower power, or both. Perhaps enough of each to offset the greater power/die area required to decode these variable-length instructions.
It's not a big factor, but combined with the other points that have been made in this thread (about how CISCs translate to RISC internally, etc., the advantages of using a very-mature x86 OS/software stack), there are a number of reasons why an x86 embedded processor might be good.
Yes, and likewise if not for the U.S., Paris would have been renamed "Hitlerstadt" and we'd all be speaking German.
No. Here's an analogy:
Unsold cars at a car dealership are "property" too, but the dealership does not get taxed on this "property" if someone steals one of them.
Now, the problem with this analogy (as I'm sure would have been pointed out by 10 different people and modded +5 Insightful) is that with Intellectual Property, the original property owner is not deprived of anything if someone steals the IP, because IP costs $0 to replicate. Unlike a car.
However, the analogy isn't totally worthless. The point isn't the deprivation or lack thereof of said property holder - it's that someone else acquired something of value that the property holder owns, without compensating the property owner. That view applies to both cars and IP.
I wish people would stop analogizing Intellectual Property to physical things, because it inevitably leads to someone pointing out that the owner of the IP doesn't lose anything if someone "steals" the IP, and thus it's not theft, and thus it's not wrong.
The point is that someone obtained something of value for free, without compensating the creator of that value. Whether or not anyone was "deprived" of anything, or something was "lost", is completely immaterial.
You obviously didn't read the thread that GP linked to. Things that computers address are measured in powers of 2, like RAM. Measurements of things like network speed or CPU frequency can use base-10 because computers don't need to address every bit in a network transmission or every cycle in a CPU clock. Read this post from that thread if you're still confused.
Let me lay out a possible scenario.
Developers claim adding UI options increases dev time beyond what is desired, increases chances of bugs, and is not worth the effort. Project forks. New developers put in work and dev time to iron out new UI options and/or features that original developers refused to add.
At this point, the fork presumably has a decent number of features that users have been demanding of the original. Either users will move to the fork, or the fork's changes will merge back to the original project since 1) they have been vetted thoroughly enough in the fork to appease the original developers, and 2) the original developers see that the fork's features are desirable enough that it was, in fact, worth the extra dev time and effort to implement them.
Either way, users and fork maintainers can say "we told you so" to the original developers.
This story reminded me of 1984's Ministry of Truth, which regulary "edited" history to match the current political scene. Writing stuff in Wikipedia makes it true.
I think I was too generous when I said "a lot of easily-parellelizable problems out there". The rest of them (besides graphics and physics) are in the HPC community - protein folding, SETI, that sort of thing - the stuff people run on BlueGene. Not something your average, everyday user is going to experience, at least not in the near future.
So then, what's the real benefit of CUDA == GPGPU == general purpose processing on a GPU, if it only speeds up a certain class of general-purpose problems? It starts to become less general-purpose then, don't you think? Cheaper computing power to do protein folding or SETI, sure, but then? Maybe we'll find more problems that are helped by GPGPU, and nVdia and ATI will have the infrastructure in place to support them. But like I said, I'm a bit skeptical.
That's exactly why I'm skeptical of CUDA in general. Rendering graphics is one of those "embarrassingly parallel" operations, where for every frame rendered, each triangle, vertex, and pixel is independent of the others. I don't know much about physics processing, but I'm guessing it's the same. Simply throwing say, 4x more hardware at the problem gives you pretty much the ideal 4x speedup. Once you introduce random data dependencies into the problem, though, parallelization is much harder and you don't get the same speedups.
There are a lot of easily-parallelizable problems out there, which is great...but then again, there are a lot of not-so-easily-parallelizable problems out there too. I don't think CUDA would do any better than a general-purpose CPU on the latter class of problems.
So, to use the OP as a case study, math games target mostly boys?
Aside from the above, not every damn word a person says is meant with an agenda. You're reading a ridiculous amount into the statement. Most people do not carefully consider their every sentence and think of all possible interpretations of it, snipping out every bit of detail that is not immediately relevant, compressing it into some robotic minimal communication. Most people just say things or write things.Possible, but given the fact that this community has many times debated the very thing which the OP alluded to, I'd consider it a remote possibility that the thought did not cross OP's mind.
I remember a game called Math Maze, and it's definitely not the same one described above, but it was a staple among the boys of my class. The girls preferred Mathville, for what it's worth. Maybe by throwing that one out there you won't accuse me of wild sexism. The one black kid (male) liked a third game that involved speed (the others did not) called Math Race. NOW I AM A RACIST! The East Asian kid (male) liked Math Maze too. I'm totally playing into the nerdy Asian stereotype. Societal progress has stopped entirely.The purpose of your above paragraph is explicitly to discuss differing demographics, so of course it's fine to state differences. OP made a seemingly offhand remark, singling out a single demographic, in an unrelated paragraph.
The French teacher wanted us to play this text adventure in French, but within the first half hour somebody had worked out the minimal set of steps to win the game with full points, and distributed it to everyone so they could beat that quickly and go back to playing the games they wanted.
Would it have been so bad if, instead of "somebody", I said "I" in the last sentence? I mean, it's not really relevant.
You are recounting the story and speaking in the first person; of course it's fine to use "I". If you are an East Asian male, and instead you had said "I, an East Asian male", then yes, that's flamebait.
Perhaps my motivation is show how much smarter I am than all the other plebes.Given that this is Slashdot, that's likely.
Why should the GP mention the sex of those who were competing at all, other than to backdoor a debate about girls and math?
called Assassin that some frats and/or dorms play at UIUC. A guy locked himself in his room to avoid an assassin, who then camped right outside his door, ready to strike the instant he opened the door. Supposedly, the guy really had to do #2, but rather than open the door and lose, he crapped out his 2nd floor window.
The number 1 reason why x86 processors burn this much power is the incredibly dense feature set they implement, a feature set that was designed with servers/workstations/desktops in mind, not .01W cell phones.
First and foremost, they implement a variable-length CISC instruction set, which complicates nearly everything in the front end of the machine (branch prediction to instruction decode), and some stuff in the back end too. Add on multiple operating modes, multiple paging modes implemented in hardware, advanced memory subsystems with hardware misalign support and relatively strong store ordering (vs. ARM at least), virtualization, SIMD, backward compatibility/legacy features. That's not to say that ARM doesn't have any of these, but from the little I know about it, they are pared down or specifically designed with an extremely low-power core in mind, not a server. And then, with an x86 you have to crank up the clock speed because of the pipeline depth.
That said, it can't hurt the x86 market to get TDP down an order of magnitude from ~50-100W to 5W.
Also, from the little I know about cell phones, there are typically two processors, a "baseband DSP" which handles the radio, and an "application processor" which runs the UI, apps, Java, etc. How much power does the baseband burn? With an x86, maybe it's possible to get rid of it entirely.
Nope. Ever heard of advertising revenue? How about non-free items included with the free ones? Ever hear of a company called Red Hat?
The nature of the copyrighted work is that it is trivial and has about a $20 valueThat's about the value of a music CD of your favorite band. Now, someone remind me, what is the fine for illegally redistributing copyrighted material?
and in a Chinese court the case would be thrown out.Would it?
In AMD's case, the shared cache sits on the other side of the fully-connected crossbar, which allows intra-core communication to happen without using HyperTransport at all. So yes, it's shared, but each core has its own "port" to it and can access it independently. Same deal for the Intel shared L2. The phrase "single connection to the system bus" is misleading because it implies a bottleneck where there (most times) isn't one.
In the case of AMD, it also means a shared memory controller.The memory controller on Intel systems is shared as well. It's just sitting on a different chip, across the FSB.
In fact, once you move to multi-socket, AMD systems generally have as many memory controllers as there are sockets, and with NUMA optimizations in modern OS's, it's likely that a core will only ever need the memory controller which it's closest to. In Intel systems, all cores on all sockets still share a single memory controller.
Every time someone makes an analogy to printing presses and books, they get +5 Insightful. Here are the fundamental flaws with that analogy:
* Printing presses still produce physical things.
* People like that because they want to read books in physical form, not on a computer screen. Thus the physical item still has value to them.
* It still costs some money and some effort to duplicate an entire book on a copier. It's possible to scan it into electronic form, but see the previous bullet for why no one cares about that.
If people hated reading books in physical form and instead the electronic form was the medium of choice, I'm pretty sure you'd see similar things happening to the book industry. Fortunately for them, electronic books just plain suck. Also, I think it's pretty obvious that they physical form of music has absolutely no value to a consumer over the digital form, for 99% of consumers. In fact, the digital form might have more value because it can be easily transferred to a portable music player.
A better analogy is newspapers. People don't mind reading their news on a computer screen, even if it's basically the exact same content as a physical newspaper and there's no value-adding going on by virtue of it being an electronic medium (video clips, animations, interactivity). I'm pretty sure you're aware that physical newspapers have taken a hit in recent times because people can get the same information online for "free" (ad-supported). Of course, newspapers have been ad-supported for a long time, so instead of just disappearing altogether, the ads just went from physical form into online form. Physical-form music has *never* been ad-supported. Maybe all this means is that any physical-form music will be ad-supported. Hey, maybe I've just stumbled onto a brilliant business idea!