Learning on your own **and** learning as part of a formal degree program is probably the best. Most purely self taught tend to have gaps in their knowledge.
These so called gaps in knowledge, are mostly a reflection of the formal training you've received.
Everyone has gaps in knowledge, it's just that formal education uses checklists in the form of curriculums.
It's actually quite damaging, since it fosters people who go around telling others what they NEED to know, while oblivious to their own shortcomings that weren't part of their training.
Not really. To use your analogy, a pilot will use checklists to make sure something important is not overlooked. Similarly such checklists of core classes make sure that some important basic concepts are not overlooked.
To be more clear. I was not referring to some piece of arcane knowledge when I referred to "gaps". I was referring to very practical useful material that was directly applicable to the problem at hand.
We could talk about the lack of social, life and work experience which comes as a result of spending most of your young adulthood behind a schoolbench.
We could also talk about being in a university environment surrounded by equipment you will not have access to in industry and being surrounded by peers who posses more talent and creativity than you will find in most work environments. And how you and these peers can go off and work on projects that interest or entertain you, projects that are not class assignments, and gain valuable self-taught lessons and experience. Again, to emphasize the overall theme of this thread... University trained and self-taught are not mutually exclusive. Oh, and by peers, I am referring to those who entered a CS program (or equivalent) with a genuine interest in the topic and interest in programming, not those who are merely there because a guidance councilor or parent told them CS was a good career path. Universities are what would be referred to as a hub, a concentration, of the former self-motivated talent.
Some people have the ability to develop extraordinary skills without formal training, and yet, most people can't no matter how much training they get.
My argument is that the former can go even farther when combining the self guided studies and interests with the formal. Self guided studies and experience, and practical work experience, help one evaluate the usefulness of the formal concepts, some concepts are quite applicable to the real world and others more applicable to academia. And in the other direction the formal training can provide a more capable foundation to guide and put into context one's self guided efforts.
learning by doing should be the most accurate way of learning, and, it works very well for areas where "doing" isn't associated with high cost.
Again, "doing" takes place both off campus and on campus.
All training costs time and money, why not pick the one with least associated cost?
Because the purely self-guided route often leads to a lesser result, a meaningful competitive disadvantage in capabilities.
I'll close with: I have seen both sides, I've been there and done that. I started out as a non-CS major, took an intro CS class and fell in love with it. I started doing lots of programming outside of class assignments. Dropped out and with friends literally developed software in a house we rented with computers spread throughout, bedrooms, living room, etc. Some of us lived there, some came and went during the day. Our garage was our manufacturing and shipping (the days when software came on floppy discs dropped into boxes with a manual - I so much prefer today's purely digital distribution channels). We did not get rich but we paid our bills, had fun and learned quite a lot. I returned to school as a CS major and earned a BS and MS while working "normal" programming jobs. I've lived both sides of this argument and I know others who have comparable combinations of the self-taught and formal training.
The two paths complement one another, each adds to the other.
Self taught and degree aren't mutally exclusive... Also university isn't just about learning a trade (that's trade school). It's about getting a rounded education in stuff you probably don't give a shit about...
I can't agree more. Learning on your own **and** learning as part of a formal degree program is probably the best. Most purely self taught tend to have gaps in their knowledge. They are just as smart, possessing the same raw talent and I have worked with many and would be happy to work with them again... but occasionally gaps are evident. There are classes in a degree program that a person has no interest in and they are unlikely to study on their own. However these "uninteresting" topics are sometimes important or may provide an unexpected solution or insight into something you are working on.
I have only met one person who is purely self taught, reads computer science textbooks or the equivalent, and reads such books covering a wide variety of topics comparable to what one sees in a traditional computer science program. When I was working on my degree I borrowed Knuth vol 1-3 from this person, these were not vanity books for a bookshelf, they were all obviously read.
Most people do not posses the discipline to do it on their own. They will benefit from a formal program that forces them to do things they would not otherwise do.
I teach at a university. I make the same for an online class as I make for one that is classroom based.
But once the lecture is recorded, the administration can hire anyone (even grad students) to teach (TA) the course. You're extraneous until they need an updated recording. Of course researchers would love that...
The recorded lecture is only part of the class. My micro and macro economic classes had recorded lectures that we could view at our convenience prior to class. Class time was then spent entirely on discussion. Discussion including being called on by the professor to explain some concept, discuss a concept in the context of current events, etc. My fellow students and I liked this format much better than more traditional classes even though it probably increased our workload.
Using class time for a canned lecture is a waste. However increasing discussion time yields a better educational experience and the professor is quite critical to this discussion process. Well, a good professor.
Yeah, I actually know that. My 1st and 2nd editions of K&R are well used. I have no idea why I referred to Ken Thompson. I guess I was thinking about C compilers, hacking them, Ken Thompson's paper and had a major brain fart connecting the language and the book. Its quite embarrassing.
No matter how few people actually read through the Linux kernel code, it's sufficiently open that blatant backdoors are not going to be inserted.
Open source suffers from quasi-religious stuff too, as you just demonstrated with your claim. Ken Thompson, of Bell Labs and Unix and C fame - the "K" in K&R, demonstrates the insufficiency of being able to read the source code. http://cm.bell-labs.com/who/ken/trust.html
The reality, as I understand it is that this is a gross oversimplification.
Yes and no. That $45/user figure was an average of course. The problem is that the influencers are not really known, and anyone can be an influencer to a degree. That is why facebook ads are deliverd to anyone fitting the desired profile. The profile can specify things like gender, age and location but not things like social connectivity and some likelihood of providing "micro-influence".
I think what you are describing is more hypothetical than reality. It may be a direction that facebook could conceivably head in but today facebook advertising is not that different from web advertising, paying for the bulk impressions or individual clicks of ordinary users, although possibly with the profile matching previously described.
No. Advertisers think a facebook user is worth $45/year. If advertisers do not see that $45/year translate into well over $45/year of additional sales then that $45 value will drop as fast as the stock price.
For an iPhone/iPad app I find that google ads are more effective and cost less. I've run a rotation of google-only, facebook-only, google-and-facebook, and no ads. Maybe facebook ads work for web and desktop but they do not seem to work for mobile.
Stupid, the average price/earnings ratio (P/E) of Nasdaq is about 21, Yahoo 17, Apple 16, Microsoft 15, see, Facebook is 105. You do not have to be an investor to see the company is hugely overvalued, the stock should be heading south.
The people who lost on facebook knew it was overvalued. However they expected it to become even more overvalued. They were thinking along the lines of "I'll buy at 30 and dump at 50, no way can it sustain that price." They were half right, facebook would have an unsustainable launch and bounce, they merely got the entry/exit points wrong. Had there been fewer shares it may have made it to 50.
Everyone trading knew there would be "irrational exuberance" and they thought they could make a quick buck off of it. They were not investing.
It's not about students applying to foxconn for a part time job. It's about the univeristy they work for ordering students who had no previous relations to foxconn to go work there and suspending classes as a result.
Its my understanding that this sort of stuff is not unprecedented, its how some colleges fund themselves to a degree. The following is decades old (80s) but I understand one prominent school was also a major aspirin manufacturer. Students were required to put in a shift at the aspirin plant. Its how the college kept the doors open. The funding from the government was insufficient, the students were no real potential source of revenue,... the factory was their solution. It was no secret, students knew this before they went there.
I want to stress that in the preceding that (1) the aspirin work was to literally keep the college in operation and that (2) the students knew this ahead of time. We are discussing a very different society with very different methods of organization. What seems strange to a westerner may not be so strange to locals.
That said, the Foxconn story seems a little different. However for those who did not read the article it does mention that students are expected to "experience working conditions" as part of their education. So there seems to be some degree of mitigation. It sounds as if these college educated likely-to-be future leaders and managers were supposed to spend some time learning what it is like to be a regular worker. It *may* be that the work was abrupt and unplanned, not that it wasn't supposed to occur in some form with a bit more planning and prior notice to the student. I think a little more info is necessary regarding what sort of work students are expected to perform while at the university.
Really. How does your reality conflict with the subtle type of errors that I discussed, the simple incorrect calculation rather than a crash? The type of error that goes by unnoticed. Recall the Intel FDIV bug? Overclocking errors can result in errors of that nature but different in their specifics on every CPU.
Well it could make the privacy situation even worse. There is no reason to believe that a node will not try to mine and monetize your data just like facebook, or try to censor some type of information they are hosting (due to local laws not a personal bias?),... As a person objects to one node's policy and moves to another they are increasing the number of 3rd parties that have their private info.
Basically there is no free lunch. There is a potential downside to hopping from one node to another.
Yes, over clocking can possibly have repercussions, so can getting out of bed in the morning.
You just don't (or do not want to) get it. Its not "can possibly have", its "usually does have".
All you are saying is that you are an apple fanboy and will discount anything that is at all modified from the way the one true jobs gave it to you.
Yeah, demonstrating that overclocking makes CPUs unreliable and shortens their life is Apple fanboyism. That is quite desperate. Given your desperation to cling to the notion that overclocking is good for everyone I suggest you rethink who is the fanboy in this discussion.
There is lots of inadequate testing, its the nature of the beast with respect to overclocking.
Again. The failure mode is not "works fine" vs "crashes". At some speed a particular CPU will experience failures. These failures will start out quite subtle, literally 2+2=5 sort of stuff and will progress in severity as higher speeds are attempted. Where the subtle failures begin, and where that line is crossed between subtle and crashing, will be different for every CPU. The prerequisites for these failures will differ. A certain sequence of instructions may be needed on one CPU, a different set of instructions on a different CPU. Maybe specific data patterns are required. And by different CPUs I am talking about two CPUs that came off of the same production line, not different makes and models.
Unless you have very sophisticated test apparatus, the sort of stuff only Intel, AMD, etc have, you really can't effectively test an overclocked CPU. All you can hope to achieve is to recognize the CPU that fails earlier and harder than its peers.
The false sense of security in overclocking is in part due to the way CPU vendors once filled orders. For simplicity lets say a vendor offered a CPU in slow, medium and fast versions. The production line only makes the fast version. CPUs that fail testing at fast are retested at medium and slow. Sometimes there would be more slow CPUs ordered than were available, so the company would sell CPUs that actually passed testing at medium or fast as slow CPUs. An overclocker might get lucky. He might buy a slow CPU that actually passed testing at fast. However the overclocker has no way of knowing if this is the case, or if the CPU sold as slow really did fail testing at fast. Again, that failure may only be recognizable by the special testing rig the vendor has, and not by some test suite an end user has.
... people have been doing this with desktop CPUs for years.
And experiencing the exact sort of problems I described. Sometimes aware, sometimes not.
We are talking about a phone here, not the computer in charge of launching the nukes. If it becomes unreliable clock it down, not a major risk.
Still its a hobbyists/hacker thing where unreliability is accepted to a degree. Which is why I don't count overclocking as a performance improving option for the public at large. Another reason I don't count overclocking is that it shortens the lifespan of the CPU. That is another long discussion and I assume I am boring you enough.:-)
Have you never overclocked a desktop CPU?
More importantly I've had experience identifying the subtle sort of problems I described.
The x86 CPUID instruction can return a vendor id string, "GenuineIntel" for Intel CPUs. I've seen an otherwise seemingly working CPU return a vendor id string with a one bit error when overclocked, misspelling the string. I've seen an otherwise seeming working CPU throw a general protection fault when loading an immediate value into a register when overclocked (if you haven't programmed in assembly language just trust me that the preceding is impossible in a working CPU), and dependent upon a certain sequence of instructions. I've seen basic math instructions (ADD, MUL, etc) return erroneous results, causing an array index calculation to go horribly wrong in a seemingly working overclocked CPU. By seemingly working in all these cases I mean the machine booted and appeared to run fine, appeared to run various apps and games fine, but one program or the other would consistently crash or misbehave when overclocked.
There is no testing for overclocking. Again, errors can be subtle, a simple incorrect answer and it can be dependent upon instruction sequences. The sequence of instructions can vary, the type of error can vary, the point at which an individual CPU begins experiencing such errors can vary,... overclocking makes devices unreliable.
No the droid 3 is not a successor either. It is also a locked down mediocre device. There is no motorola equivalent.
Apple is not months behind they are now at least a year behind. The point is a 3GS is slower than a D1.
I doubt the iPhone5 will have a 1920x1280 screen, but that would be great. The point I was making is the current iPhone screen is a relic and badly needs to be updated. The DPI should hopefully at least stay the same, but the device will be slightly larger.
The point is a 3GS is slower than a D1.
How is the 3GS slower, the D1 and 3GS have the same CPU and RAM? The D1 CPU is 8% slower and its running Java apps rather than native binaries but that is nitpicking. And no, I do not count overclocking. That is not the user experience that nearly all D1 users had/have. Plus overclocked CPUs, especially severely overclocked like the one described generally have errors. Note that not all overclocking induced errors are obvious, resulting in crashes, etc. Sometimes the CPU simply gives the wrong answer after a specific sequence of instructions. The sequence and errors are different for every overclocked CPU so its not something easily tested for.
I doubt the iPhone5 will have a 1920x1280 screen, but that would be great.
I was not suggesting that. That would seem pointless given that we are already at the limits of the human visual system. 3 to 3GS was a doubling but that would seem to be a one time thing. I'm just arguing that an increase to 720p seems unlikely since it is not a whole number multiple, a 30% taller screen would be needed to maintain pixels per inch and keep unstretched legacy apps just as readable. Yes it would be nice for new apps but I expect too many backwards compatibility problems. Besides, on a 3.5 inch screen can one really see a difference in a video rendered at 960x640 vs 1280x720?
Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
AFAIK and as far as I could find in searches, VMWare is not nor ever was Linux-based. If you have information to the contrary I would be interested.
We are discussing Linux systems not the Linux kernel in isolation. Vmware is a quite common thing to run on such systems and the problems associated with it stem from kernel design decisions (no stable kernel interface, expecting software to be recompiled against the new kernel).
> Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
That's been addressed. If vmware still screws this up then that's a problem being create by VMware Corp and not Linux.
If vmware requires no update, merely to be recompiled using the new kernel headers, then how is it not a Linux problem? Linux chose to not have a stable interface, to be free to redefine structures passed as parameters for example, thereby requiring that software that directly interfaces with the kernel be recompiled. Whether this development approach is right or wrong is irrelevant, the fact remains that it has a cost and it is due to a Linux decision.
Running to MacOS like an idiot is not going to insulate you from proprietary developers that behave badly.
Define "behave badly".
"Running to Mac OS X"? You realize many of these people have been using Linux since the 90s?
So some people try two *nix environments side by side, find their software runs under both but one is less of a hassle (and has other off topic advantages), and moving to the system that is less of a hassle is "idiotic" in your opinion? That the "correct" decision is to make a political/philosophical one rather than a technical one in your opinion? That is why it sounds like you are saying.
Yes, seriously, someday iPad apps will be developed on an iPad.
That day will be when the iPad plugs into a "docking station" and acts as the "cpu" and an external keyboard, mouse, display and storage (HDD, SSD, etc) connect to it through the docking station.
> as attested to by various long term Linux users in yesterday's article on the subject
Except that most of those "various" users were all called perpenso. Just because you've got a data point doesn't mean it applies to everybody.
Actually there were others making similar points and their posts and my post had various followups identifying long time Linux users who moved on as well. There was a common theme among many of these people. They were once desktop Linux users and now they just use Linux for the server sitting in a closet.
I got linux on desktop. It works perfectly. Seriously, what's the problem?
Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
These things add up and explain the many defections from desktop Linux to Mac OS X, as attested to by various long term Linux users in yesterday's article on the subject. The short story is that many Linux users merely wanted a *nix environment, they were not into the politics or crusade. That is desktop Linux's problem, its becoming a less interesting option for those who just want a *nix environment and don't want to join a social movement.
Apple had 10% market share with the Macintosh at one time. So they are still behind where they used to be.
The 7.5% figure is from tracking visits to some website.
I believe the folks who track new PC sales say Mac went from 3.5% (pre Intel CPU) to 10%. I used to have access to proprietary software usage info that reported similar numbers, usage as in the software automatically contacted a server that provided patches and notifications as the software starts up.
We are discussing google services which require a login, not cookies.
For those that care you use Firefox with privacy turned on. That's how I would access any Google service from a system not my own (and sometimes even from my own system).
How does private browsing prevent google from knowing who you are, again you **logged into** a google service.
Slashdot Mirror
Learning on your own **and** learning as part of a formal degree program is probably the best. Most purely self taught tend to have gaps in their knowledge.
These so called gaps in knowledge, are mostly a reflection of the formal training you've received. Everyone has gaps in knowledge, it's just that formal education uses checklists in the form of curriculums. It's actually quite damaging, since it fosters people who go around telling others what they NEED to know, while oblivious to their own shortcomings that weren't part of their training.
Not really. To use your analogy, a pilot will use checklists to make sure something important is not overlooked. Similarly such checklists of core classes make sure that some important basic concepts are not overlooked.
To be more clear. I was not referring to some piece of arcane knowledge when I referred to "gaps". I was referring to very practical useful material that was directly applicable to the problem at hand.
We could talk about the lack of social, life and work experience which comes as a result of spending most of your young adulthood behind a schoolbench.
We could also talk about being in a university environment surrounded by equipment you will not have access to in industry and being surrounded by peers who posses more talent and creativity than you will find in most work environments. And how you and these peers can go off and work on projects that interest or entertain you, projects that are not class assignments, and gain valuable self-taught lessons and experience. Again, to emphasize the overall theme of this thread ... University trained and self-taught are not mutually exclusive. Oh, and by peers, I am referring to those who entered a CS program (or equivalent) with a genuine interest in the topic and interest in programming, not those who are merely there because a guidance councilor or parent told them CS was a good career path. Universities are what would be referred to as a hub, a concentration, of the former self-motivated talent.
Some people have the ability to develop extraordinary skills without formal training, and yet, most people can't no matter how much training they get.
My argument is that the former can go even farther when combining the self guided studies and interests with the formal. Self guided studies and experience, and practical work experience, help one evaluate the usefulness of the formal concepts, some concepts are quite applicable to the real world and others more applicable to academia. And in the other direction the formal training can provide a more capable foundation to guide and put into context one's self guided efforts.
learning by doing should be the most accurate way of learning, and, it works very well for areas where "doing" isn't associated with high cost.
Again, "doing" takes place both off campus and on campus.
All training costs time and money, why not pick the one with least associated cost?
Because the purely self-guided route often leads to a lesser result, a meaningful competitive disadvantage in capabilities.
I'll close with: I have seen both sides, I've been there and done that. I started out as a non-CS major, took an intro CS class and fell in love with it. I started doing lots of programming outside of class assignments. Dropped out and with friends literally developed software in a house we rented with computers spread throughout, bedrooms, living room, etc. Some of us lived there, some came and went during the day. Our garage was our manufacturing and shipping (the days when software came on floppy discs dropped into boxes with a manual - I so much prefer today's purely digital distribution channels). We did not get rich but we paid our bills, had fun and learned quite a lot. I returned to school as a CS major and earned a BS and MS while working "normal" programming jobs. I've lived both sides of this argument and I know others who have comparable combinations of the self-taught and formal training.
The two paths complement one another, each adds to the other.
Self taught and degree aren't mutally exclusive ... Also university isn't just about learning a trade (that's trade school). It's about getting a rounded education in stuff you probably don't give a shit about ...
I can't agree more. Learning on your own **and** learning as part of a formal degree program is probably the best. Most purely self taught tend to have gaps in their knowledge. They are just as smart, possessing the same raw talent and I have worked with many and would be happy to work with them again ... but occasionally gaps are evident. There are classes in a degree program that a person has no interest in and they are unlikely to study on their own. However these "uninteresting" topics are sometimes important or may provide an unexpected solution or insight into something you are working on.
I have only met one person who is purely self taught, reads computer science textbooks or the equivalent, and reads such books covering a wide variety of topics comparable to what one sees in a traditional computer science program. When I was working on my degree I borrowed Knuth vol 1-3 from this person, these were not vanity books for a bookshelf, they were all obviously read.
Most people do not posses the discipline to do it on their own. They will benefit from a formal program that forces them to do things they would not otherwise do.
I teach at a university. I make the same for an online class as I make for one that is classroom based.
But once the lecture is recorded, the administration can hire anyone (even grad students) to teach (TA) the course. You're extraneous until they need an updated recording. Of course researchers would love that...
The recorded lecture is only part of the class. My micro and macro economic classes had recorded lectures that we could view at our convenience prior to class. Class time was then spent entirely on discussion. Discussion including being called on by the professor to explain some concept, discuss a concept in the context of current events, etc. My fellow students and I liked this format much better than more traditional classes even though it probably increased our workload.
Using class time for a canned lecture is a waste. However increasing discussion time yields a better educational experience and the professor is quite critical to this discussion process. Well, a good professor.
The "K" of K&R is wrong.
Yeah, I actually know that. My 1st and 2nd editions of K&R are well used. I have no idea why I referred to Ken Thompson. I guess I was thinking about C compilers, hacking them, Ken Thompson's paper and had a major brain fart connecting the language and the book. Its quite embarrassing.
No matter how few people actually read through the Linux kernel code, it's sufficiently open that blatant backdoors are not going to be inserted.
Open source suffers from quasi-religious stuff too, as you just demonstrated with your claim. Ken Thompson, of Bell Labs and Unix and C fame - the "K" in K&R, demonstrates the insufficiency of being able to read the source code.
http://cm.bell-labs.com/who/ken/trust.html
The reality, as I understand it is that this is a gross oversimplification.
Yes and no. That $45/user figure was an average of course. The problem is that the influencers are not really known, and anyone can be an influencer to a degree. That is why facebook ads are deliverd to anyone fitting the desired profile. The profile can specify things like gender, age and location but not things like social connectivity and some likelihood of providing "micro-influence".
I think what you are describing is more hypothetical than reality. It may be a direction that facebook could conceivably head in but today facebook advertising is not that different from web advertising, paying for the bulk impressions or individual clicks of ordinary users, although possibly with the profile matching previously described.
You are worth $45/year to facebook.
No. Advertisers think a facebook user is worth $45/year. If advertisers do not see that $45/year translate into well over $45/year of additional sales then that $45 value will drop as fast as the stock price.
... Facebook is a dead-end for advertisers ...
For an iPhone/iPad app I find that google ads are more effective and cost less. I've run a rotation of google-only, facebook-only, google-and-facebook, and no ads. Maybe facebook ads work for web and desktop but they do not seem to work for mobile.
Stupid, the average price/earnings ratio (P/E) of Nasdaq is about 21, Yahoo 17, Apple 16, Microsoft 15, see, Facebook is 105. You do not have to be an investor to see the company is hugely overvalued, the stock should be heading south.
The people who lost on facebook knew it was overvalued. However they expected it to become even more overvalued. They were thinking along the lines of "I'll buy at 30 and dump at 50, no way can it sustain that price." They were half right, facebook would have an unsustainable launch and bounce, they merely got the entry/exit points wrong. Had there been fewer shares it may have made it to 50.
Everyone trading knew there would be "irrational exuberance" and they thought they could make a quick buck off of it. They were not investing.
It's not about students applying to foxconn for a part time job. It's about the univeristy they work for ordering students who had no previous relations to foxconn to go work there and suspending classes as a result.
Its my understanding that this sort of stuff is not unprecedented, its how some colleges fund themselves to a degree. The following is decades old (80s) but I understand one prominent school was also a major aspirin manufacturer. Students were required to put in a shift at the aspirin plant. Its how the college kept the doors open. The funding from the government was insufficient, the students were no real potential source of revenue, ... the factory was their solution. It was no secret, students knew this before they went there.
I want to stress that in the preceding that (1) the aspirin work was to literally keep the college in operation and that (2) the students knew this ahead of time. We are discussing a very different society with very different methods of organization. What seems strange to a westerner may not be so strange to locals.
That said, the Foxconn story seems a little different. However for those who did not read the article it does mention that students are expected to "experience working conditions" as part of their education. So there seems to be some degree of mitigation. It sounds as if these college educated likely-to-be future leaders and managers were supposed to spend some time learning what it is like to be a regular worker. It *may* be that the work was abrupt and unplanned, not that it wasn't supposed to occur in some form with a bit more planning and prior notice to the student. I think a little more info is necessary regarding what sort of work students are expected to perform while at the university.
my reality conflicts with your claims.
Really. How does your reality conflict with the subtle type of errors that I discussed, the simple incorrect calculation rather than a crash? The type of error that goes by unnoticed. Recall the Intel FDIV bug? Overclocking errors can result in errors of that nature but different in their specifics on every CPU.
... why should I care? ...
Well it could make the privacy situation even worse. There is no reason to believe that a node will not try to mine and monetize your data just like facebook, or try to censor some type of information they are hosting (due to local laws not a personal bias?), ... As a person objects to one node's policy and moves to another they are increasing the number of 3rd parties that have their private info.
Basically there is no free lunch. There is a potential downside to hopping from one node to another.
So what? What does this have to do with anything?
You asked for my experience with overclocking.
Yes, over clocking can possibly have repercussions, so can getting out of bed in the morning.
You just don't (or do not want to) get it. Its not "can possibly have", its "usually does have".
All you are saying is that you are an apple fanboy and will discount anything that is at all modified from the way the one true jobs gave it to you.
Yeah, demonstrating that overclocking makes CPUs unreliable and shortens their life is Apple fanboyism. That is quite desperate. Given your desperation to cling to the notion that overclocking is good for everyone I suggest you rethink who is the fanboy in this discussion.
There is lots of testing, ...
There is lots of inadequate testing, its the nature of the beast with respect to overclocking.
Again. The failure mode is not "works fine" vs "crashes". At some speed a particular CPU will experience failures. These failures will start out quite subtle, literally 2+2=5 sort of stuff and will progress in severity as higher speeds are attempted. Where the subtle failures begin, and where that line is crossed between subtle and crashing, will be different for every CPU. The prerequisites for these failures will differ. A certain sequence of instructions may be needed on one CPU, a different set of instructions on a different CPU. Maybe specific data patterns are required. And by different CPUs I am talking about two CPUs that came off of the same production line, not different makes and models.
Unless you have very sophisticated test apparatus, the sort of stuff only Intel, AMD, etc have, you really can't effectively test an overclocked CPU. All you can hope to achieve is to recognize the CPU that fails earlier and harder than its peers.
The false sense of security in overclocking is in part due to the way CPU vendors once filled orders. For simplicity lets say a vendor offered a CPU in slow, medium and fast versions. The production line only makes the fast version. CPUs that fail testing at fast are retested at medium and slow. Sometimes there would be more slow CPUs ordered than were available, so the company would sell CPUs that actually passed testing at medium or fast as slow CPUs. An overclocker might get lucky. He might buy a slow CPU that actually passed testing at fast. However the overclocker has no way of knowing if this is the case, or if the CPU sold as slow really did fail testing at fast. Again, that failure may only be recognizable by the special testing rig the vendor has, and not by some test suite an end user has.
... people have been doing this with desktop CPUs for years.
And experiencing the exact sort of problems I described. Sometimes aware, sometimes not.
We are talking about a phone here, not the computer in charge of launching the nukes. If it becomes unreliable clock it down, not a major risk.
Still its a hobbyists/hacker thing where unreliability is accepted to a degree. Which is why I don't count overclocking as a performance improving option for the public at large. Another reason I don't count overclocking is that it shortens the lifespan of the CPU. That is another long discussion and I assume I am boring you enough. :-)
Have you never overclocked a desktop CPU?
More importantly I've had experience identifying the subtle sort of problems I described.
The x86 CPUID instruction can return a vendor id string, "GenuineIntel" for Intel CPUs. I've seen an otherwise seemingly working CPU return a vendor id string with a one bit error when overclocked, misspelling the string. I've seen an otherwise seeming working CPU throw a general protection fault when loading an immediate value into a register when overclocked (if you haven't programmed in assembly language just trust me that the preceding is impossible in a working CPU), and dependent upon a certain sequence of instructions. I've seen basic math instructions (ADD, MUL, etc) return erroneous results, causing an array index calculation to go horribly wrong in a seemingly working overclocked CPU. By seemingly working in all these cases I mean the machine booted and appeared to run fine, appeared to run various apps and games fine, but one program or the other would consistently crash or misbehave when overclocked.
There is no testing for overclocking. Again, errors can be subtle, a simple incorrect answer and it can be dependent upon instruction sequences. The sequence of instructions can vary, the type of error can vary, the point at which an individual CPU begins experiencing such errors can vary, ... overclocking makes devices unreliable.
No the droid 3 is not a successor either. It is also a locked down mediocre device. There is no motorola equivalent.
Apple is not months behind they are now at least a year behind. The point is a 3GS is slower than a D1.
I doubt the iPhone5 will have a 1920x1280 screen, but that would be great. The point I was making is the current iPhone screen is a relic and badly needs to be updated. The DPI should hopefully at least stay the same, but the device will be slightly larger.
The point is a 3GS is slower than a D1.
How is the 3GS slower, the D1 and 3GS have the same CPU and RAM? The D1 CPU is 8% slower and its running Java apps rather than native binaries but that is nitpicking. And no, I do not count overclocking. That is not the user experience that nearly all D1 users had/have. Plus overclocked CPUs, especially severely overclocked like the one described generally have errors. Note that not all overclocking induced errors are obvious, resulting in crashes, etc. Sometimes the CPU simply gives the wrong answer after a specific sequence of instructions. The sequence and errors are different for every overclocked CPU so its not something easily tested for.
I doubt the iPhone5 will have a 1920x1280 screen, but that would be great.
I was not suggesting that. That would seem pointless given that we are already at the limits of the human visual system. 3 to 3GS was a doubling but that would seem to be a one time thing. I'm just arguing that an increase to 720p seems unlikely since it is not a whole number multiple, a 30% taller screen would be needed to maintain pixels per inch and keep unstretched legacy apps just as readable. Yes it would be nice for new apps but I expect too many backwards compatibility problems. Besides, on a 3.5 inch screen can one really see a difference in a video rendered at 960x640 vs 1280x720?
Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
AFAIK and as far as I could find in searches, VMWare is not nor ever was Linux-based. If you have information to the contrary I would be interested.
We are discussing Linux systems not the Linux kernel in isolation. Vmware is a quite common thing to run on such systems and the problems associated with it stem from kernel design decisions (no stable kernel interface, expecting software to be recompiled against the new kernel).
> Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
That's been addressed. If vmware still screws this up then that's a problem being create by VMware Corp and not Linux.
If vmware requires no update, merely to be recompiled using the new kernel headers, then how is it not a Linux problem? Linux chose to not have a stable interface, to be free to redefine structures passed as parameters for example, thereby requiring that software that directly interfaces with the kernel be recompiled. Whether this development approach is right or wrong is irrelevant, the fact remains that it has a cost and it is due to a Linux decision.
Running to MacOS like an idiot is not going to insulate you from proprietary developers that behave badly.
Define "behave badly".
"Running to Mac OS X"? You realize many of these people have been using Linux since the 90s?
So some people try two *nix environments side by side, find their software runs under both but one is less of a hassle (and has other off topic advantages), and moving to the system that is less of a hassle is "idiotic" in your opinion? That the "correct" decision is to make a political/philosophical one rather than a technical one in your opinion? That is why it sounds like you are saying.
and the proprietary nVidia driver, and the proprietary $whatever, etc..
It seems to me your problems are not with Linux kernel itself, but with the proprietary extensions you add onto it.
No, its Linux itself. No update of the vmware code is necessary, vmware just needs to be rebuilt under the updated kernel.
No.
Yes, seriously, someday iPad apps will be developed on an iPad.
That day will be when the iPad plugs into a "docking station" and acts as the "cpu" and an external keyboard, mouse, display and storage (HDD, SSD, etc) connect to it through the docking station.
> as attested to by various long term Linux users in yesterday's article on the subject
Except that most of those "various" users were all called perpenso. Just because you've got a data point doesn't mean it applies to everybody.
Actually there were others making similar points and their posts and my post had various followups identifying long time Linux users who moved on as well. There was a common theme among many of these people. They were once desktop Linux users and now they just use Linux for the server sitting in a closet.
I got linux on desktop. It works perfectly. Seriously, what's the problem?
Well it is annoying to have to rebuild things when the kernel is updated, vmware comes to mind.
These things add up and explain the many defections from desktop Linux to Mac OS X, as attested to by various long term Linux users in yesterday's article on the subject. The short story is that many Linux users merely wanted a *nix environment, they were not into the politics or crusade. That is desktop Linux's problem, its becoming a less interesting option for those who just want a *nix environment and don't want to join a social movement.
Apple had 10% market share with the Macintosh at one time. So they are still behind where they used to be.
The 7.5% figure is from tracking visits to some website.
I believe the folks who track new PC sales say Mac went from 3.5% (pre Intel CPU) to 10%. I used to have access to proprietary software usage info that reported similar numbers, usage as in the software automatically contacted a server that provided patches and notifications as the software starts up.
We are discussing google services which require a login, not cookies.
For those that care you use Firefox with privacy turned on. That's how I would access any Google service from a system not my own (and sometimes even from my own system).
How does private browsing prevent google from knowing who you are, again you **logged into** a google service.
Are you logging on to personal websites at work?
I'm not but plenty of people do. Chat with someone who does tech support for a company and ask them what they find in browser histories.
If it's cookie based stuff, how would it follow you there?
We are discussing google services which require a login, not cookies.