Don't be silly! CORDIC is a floating-point algorithm.
Don't be silly! CORDIC is an algorithm. Other than being something binary, the representation you use has no relevance at all (beyond the fact that the table of secant constants needs to be represented in the format of the result).
In fact, CORDIC is primarily useful when you don't have floating point hardware. If you have FP hardware (at least multipliers), you're frequently better off implementing one of the typical power series. In particular, CORDIC gives one more bit for each iteration, while a Taylor series (for only the most obvious example) doubles the number of bits with each term. If you only a few bits (e.g. 8 or 10) CORDIC works very well -- but if you have FP hardware and want something like the usual 64 bit result, 5 terms of the Taylor series will get you there faster and a Chebychev polynomial is generally faster still. Of course, back in the '50s when CORDIC was invented, the number of gates to do fast FP multiplication just wasn't available, so CORDIC ruled...
I did give enough pointers to answer the question.
At that point it's not a guess any more -- it's just a job of reverse engineering some ancient, obsolete source code. Unless you had far more interesting tricks than anything you've mentioned here, there's not much point or fun in that.
Dictionaries present their words based on popular usage. Because insure is popularly used in place of ensure, it has come to have a similar meaning. Although the original meanings may have varied as stated by other/.ers, I'm fairly certain they are nearly interchangeable at this point.
Probably this thread is mostly forgotten by now, but just in case anybody's still paying attention, here's what the Oxford English Dictionary has to say about the situation:
The form INSURE is properly a mere variant of ensure, and still occasionally appears in all the surviving senses. In general usage, however, it is now limited to the financial sense (with reference to 'insurance' of life or property), in which the form ensure is wholly obsolete.
So, at one time, the two were entirely synonymous, and now they're noticeably different. Nonetheless, although it's no longer in general usage, substituting "insure" for "ensure" doesn't necessarily change the meaning.
I'm afraid you're missing the point of anti-trust suits.
To the US government, it's more a case of: look, there are two companies making billions of dollars, and they hardly contribute anything to lobbying or campaigns! It must be a case worthy of an anti-trust suit!
Seriously, consider Microsoft's political contributions before and after they started being hit with anti-trust suits. It sure looks to me like their practices being officially approved was based more on contributions than on any actual change in how they do business.
There are simply too many different possibilities for there to be one reasonable guess. You can use shifts and adds, but that's rarely faster than the hardware, and at that time pipelining usually wasn't very useful. Table lookups of various sorts were fairly common, frequently with some bit-masking or -shifting to keep the table size reasonable.
- How did it compute sine and cosines?
CORDIC?
Since you haven't even specified what hardware it was for, attempting to guess at the low-level hardware trick(s) you used would be pretty pointless.
I believe Alpha Waves was the first video game on personal computers with full-screen "real" 3D (i.e. not scaled sprites), and I also believe it was the first 3D game with two simultaneous players sharing a screen. Would anybody have data to confirm or infirm that?
"infirm" ? Do you mean "deny", perhaps?
It wasn't the first game on a personal computer with real 3D -- I wrote some real 3D code that ran on a C64 around 1982 or '83. OTOH, the "ran" there is an exaggeration -- "walked" would be closer, and "stumbled" would probably be the right word. The code was all right, but there's only so much you can do on an 8-bit, 1 MHz processor. A couple of us played it enough to say it worked, but that's about it.
Oh, it would also depend a bit on what you're talking about as 3D -- this was wireframes only. I wrote some lighting code, but never used it except to create a couple of static images (one frame per night...)
It starts by taking a guess at the right answer, and then improving the guess until it's accurate enough to use.
The first step depends heavily on the fact that a floating point number on a computer is represented as a significand (aka mantissa) and an exponent (a power of two). For the moment, consider taking just the square root of X instead of its inverse. You could separate out the exponent part of the floating point number, divide it by two, and then put the result back together with the original significand, and have a reasonable starting point.
From there, you could improve your guesses to get a better approximation. The simplest version of that would be like a high-low game -- you split the difference between the current guess and the previous guess, and then add or subtract that depending on whether your previous guess was high or low. Eventually, you'll get arbitrarily close to the correct answer.
This can take quite a few iterations to get to the right answer though. To improve that, Newton-Raphson looks at the curve of the function you're working with, and projects a line tangent to the curve at the point of the current guess. Where that line crosses the origin gives you the next guess. That's probably a lot easier to understand from picture.
In this case, we're looking for the inverse square root, which changes the curve, but not the basic idea. As a general rule, the closer your first guess, the fewer iterations you need to get some particular level of accuracy. That's the point of the:
i = 0x5f3759df - (i>>1);
While the originator of this constant is unknown, and some of it is rather obscure, the basic idea of most of it is fairly simple: we start by shifting the original number right a bit. This divides both the mantissa and the exponent part by two, with the possibility that IF the exponent was odd, it shifts a bit from the exponent into the mantissa. The subtraction from the magic number then does a couple of things. For one thing, if a bit from the exponent was shifted into the mantissa, it removes it. The rest of the subtraction is trickier. If memory serves, it's based on the harmonic mean of the difference between sqrt(x) and (x/2) for every possible floating point number of the size you're using.
This is where the fact that it's 1/sqrt(x) instead of sqrt(x) means a lot: 1/sqrt(x) is a curve, but it's a fairly flat curve -- much flatter than sqrt(x). The result is that we can approximate a point on the curve fairly accurately with a line. In this case, it's really two lines, which gets it a bit closer still.
From there, the number has had a bit of extra tweaking done -- it doesn't actually give the most accurate first guess, but its errors are often enough in the opposite direction from those you get in the Newton-Raphson iteration steps that it gives slightly more accurate final results.
Maybe MS is somehow is benefiting from the endless cycles of MSIE-based spyware, viruses, and general security problems. If not, then it (and we) would be much better off if MS should drop MSIE completely.
MS really wasn't lying when they said IE was now part of the OS (or at least the shell). For example, if you open "My Computer" and type something like "www.microsoft.com" into its address bar, you'll get essentially the same result as if you had started by opening a window that openly stated it was IE.
Much of what a typical user sees as "Windows" is really IE. You might easily be right that MS and their customers would be better off without it, but MS would have to do a fair amount of work on a new shell to get rid of it entirely, at least from XP. I'm not sure about the shell in Vista, but I doubt it's changed drastically in this respect.
Of course, the merits of Direct-X are continually questioned by OpenGL fans, but there is no doubt that creating a standard is innovative.
The problem with that is that DirectX wasn't originated by Microsoft either. Direct3D started out as RenderMorphics RealityLab 2.0 under a new label. IOW, the "standard" already existed, and was already on the market before Microsoft acquired it.
A few other parts of DirectX might have been innovative, but when you get down to it, most of them were really just going back to the old way of doing things. Windows provided an interface to hardware at a level of abstraction that made it difficult to do anything but static graphics decently. Therefore, at the time, most attempts at animation and such were done outside of Windows. DirectX (as implied by the name) was intended as little more than a way to bypass the cruft they'd added on, and allow more direct access to the hardware, just about like if Windows wasn't there at all.
The other parts of DirectX provided decent access to sound (had been abstracted to the point that the primary capability was playing.wav files), joysticks (previously had to masquerade as a mouse to be supported by Windows at all) and so on. This wasn't innovation -- it was simply removing some of the most obvious and troublesome limitations in Windows.
Of course, since then, it's been updated quite a bit. Face reality though: most of the updates have been driven far more by Intel, ATI and nVidia than by Microsoft. They design hardware, and then DirectX is updated to provide access to the capabilities it provides (or will provide when it's finished -- it's a lot faster to describe the capability than implement it efficiently). It's open to argument that (perhaps) the geometry shaders added in DirectX 10 really are a Microsoft innovation -- though I suspect if you look hard, you'll find that nVidia and/or ATI really brought that to the table as well.
C# was not even their invention, and.NET is just a library around it, which is a pure copy of the Java innovation.
"The Java innovation" ? What was that?
Those who think Java was an innovation should do a bit of reading about the UCSD P-System. The idea of a virtual machine wasn't new then, but few (if any) had previously built it into a full-blown platform like the P-System.
Absolutely correct, and a point I should have made myself. However, many inventors do exactly the opposite of what you describe -- many try and push through as much stuff as they can find, because once an examiner has seen a reference, the presumption, should litigation arise down the road, is that the examiner looked at the reference (whether he actually did or not), and therefore the presumption is that the reference is NOT an invalidating reference.
Oh, make no mistake, I think most inventors do a fine job of disclosing everything of which they're aware. It does seem to me, however, that the system is almost surprisingly open to abuse.
BTW start going along with what your lawyers ask you to say, you'll make more money!
You might be right -- but right now, I make enough to pay my bills, and I still have my self-respect, and I value that more than a bigger house, newer car, etc. I don't mean to sound particularly noble or anything, because I don't think I really am -- in fact, acting as an expert witness is comparatively unusual for me, so I doubt it'd make a lot of difference either way. If I never act as an expert witness again, I doubt it'll affect my income too drastically.
Well, there is the affirmative duty for the patentee to disclose anything that might be material to patentability.
Not quite. They're obliged to disclose anything of which they're aware -- but they're not required to do any searching to find anything of which they weren't already aware. At least from what I've seen, most cite things they can't reasonably avoid (e.g. their own earlier patents, papers they've written, etc.) but darned little else. I've seen an "inventor" claim that he'd never read the data sheet for a part he used (he was just such a genius, he knew the pinout without looking...). The data sheet referenced an app note that almost directly disclosed his "invention"...
but considering the penalties for not disclosing material art (loss of the patent, maybe loss of a job to the patent agent/attorney is the Office of Enrollement and Discipline finds their failure to disclose egregious enough), I think that most patentees to a better than average job of disclosure.
They can only lose much of anything if it can be shown that they really were aware of the material. If most do a better job than average, either there's a lot of skew in that average, or else you're using "average" a bit differently than I learned it in statistics class (or maybe my memory's going bad after all these years -- though I still very clearly remember a girl who always showed up in some of the tightest jeans I've ever seen...)
What would YOU do if someone was paying you $300 or $400 per hour to say something like that!
Having previously been an expert witness, and having declined the umm...honor in at least one other case, I think I can say with reasonable certainty that I'd decline in a case like that. Come to think of it, nobody's asked me to be an expert witness for a while though...
Once a patent is issued, it is presumed valid -- which means, it is presumed to be patentable subject matter (35 U.S.C. 101), novel (35 U.S.C. 102), nonobvious (35 U.S.C. 103), and enabled (35 U.S.C. 112). It is up to the person challenging the validity to show why a patent isn't patentable, isn't novel, is obvious, or is not enabled.
IMO, this is one of the real sticking points: right now, when somebody applies for a patent, they have to declare that they think it's novel. The patent examiner then looks to see if it look obvious (to them). The standard, at this point, however is only that the patent applicant has to show that the "preponderance of the evidence" is in their favor. Oh, and patent examiners virtually never look at anything except older patents and applications.
When a patent is challenged, however, that changes: the challenger now has to show "clear and convincing evidence" that the patent is really obvious after all. This means, from a legal viewpoint, that the patent is supposed to be considered valid, even if there's a fairly strong reason to believe otherwise.
To add to that, nearly every patent office (definitely including the US PTO) gets part or all of its funding from patents that it issues. Worse still, examiners are rated (partly) in terms of the number of applications they examine. Neither of these really encourages patent examiners to do be as thorough as possible and reject all patent applications that really should be (though, in fairness, given their workload, I'm frequently impressed at how good of a job patent examiners do anyway).
To show that an invention is "obvious," you need essentially need to show one or more pieces of prior art that, in combination, "anticipate" or disclose the invention seeking to be patented, and you must show some "motivation to combine" the art into a single invention.
The motivation to combine must be a bit more than "motivation" as most people would think of it. Most of us would think of things like "it saves money" or "it makes the product smaller/lighter/more dependable" -- but when you're trying prove a patent is obvious, you're typically looking at something like combining two research articles, and the motivation to combine needs to be either a direct reference from one article to the other, or a third article that references both the ones you're looking at.
The standard for measurement of many things in patents is the "person of ordinary skill in the art" (POSITA). It seems to me that the POSITA has been reduced to something that bears essentially no resemblance to any real person at all. Instead, it's an almost purely mythical construct, with total awareness of all existing products and publications, and the ability to understand them, but essentially no ability to do any thinking at all. In particular, they can't draw even the most obvious conclusions unless something has been published to "suggest" (i.e. state) that conclusion for them. Even though they're theoretically aware of all papers, they're not able to put any two of them together without specific bibliographic references.
I know of at least one professor from Stanford who argued that putting a current limiting resistor in line with an LED was beyond the abilities of a person of ordinary skill in the art! If memory serves, he even won the case...
It's not really a dupe. You see, last time the "news" was that news.com was running a story about the case. This time, the "news" is that embedded.com is running a story about the case. The case itself is hardly news -- it's been around since April of 2005. Well, the real case is still older, but that's when they appealed to the Supreme Court.
Next week, it'll be news again -- and with a lot of luck they might find a link that's actually informative. Then again, that page has been around since June, so it's hardly news. Maybe coverage of the recent arguments would really be better. Oh wait...that's only a blog entry, undoubtedly written by some left wing nutcase. Maybe by next week, we can get a highly informed piece from a legitimate news service instead.
The CEO of the RIAA claims: "We celebrate advances in technology and recognize the importance of finding new ways to deliver content." At the same time, the Chairman and CEO of Universal Music Group claims MP3 players are: "just repositories for stolen music, and they all know it."
It seems fairly clear that neither the RIAA nor their member companies really understand technology well at all. They try to stagnate music distribution in the mid-80's, and they disregard how easy it is to compare and contrast their messages to make the lies obvious.
Both stories showing up on the same day is just a nice bonus. It's a bit like the old Dilbert where the PHB is holding a meeting:
Item 1: record profits -- executives will be getting bonuses.
Item 2: financial difficulties -- layoffs of technical staff.
What is the issue is is this a natural process, a man-made process or a combination?
Why is that the issue? Are we looking to assign blame?
The point is to figure out what (if anything) should be done based on what's happening. If global warming is caused primarily by mankind releasing carbon dioxide into the atmostphere, then it would behoove us to reduce our carbon dioxide emissions to control global warming. OTOH, if global warming is mostly a result (for example) the earth receiving more energy from the sun, then reducing carbon dioxide emissions will have only minimal effect.
It should also be pointed out that while most people generally take for granted that nearly everything we do releases carbon dioxide and therefore leads to global warming, that's a bit one-sided. Just for example, burning coal not only releases carbon dioxide but also generally releases at least some sulfur. Sulfur in the atmosphere causes global cooling (i.e. it does more to reflect energy from the sun back out of the atmosphere than to trap energy from the sun in the atmosphere). If you compare sulfur release into the atmosphere to the global temperature, you get a fairly close (inverse) correlation. That would support an argument that most of the arguments over global warming have things backwards: rather than currently causing global warming, the real situation could be that we (mankind) were mostly responsible for the mini-iceage by burning sulfur-bearing coal. In the last few decades we've reduced coal usage and (particularly) reduced emissions when we do use it (e.g. most power plants now have equipment to remove sulfur from their emissions). What we're seeing as global warming is really just the earth recovering back to about where it would have been if we hadn't been causing global cooling for centuries.
Likewise, many people (including some here) have suggested that planting trees as an obvious cure. Simple solutions to complex problems sound nice, but (as in this case) things are rarely as simple as they initially appear. Much of the loss of rain-forest that's often cited is largely illusory. While it's true that what's classed as rain-forest has been reduced (somewhat) the losses are often at least partially offset in terms of overall wooded land. Taking a look at the UN's FAO data we see that while the forest land in South America dropped by about.4% annually between 1990 and 2005, when/if we take the other wooded area into account, the loss is really about.2% annually (they don't provide totals/percentages for the other wooded areas -- you have to cut-n-paste into a spreadsheet yourself to get those).
Though what would be really nifty is to figure out a way of using an electrochemical effect to suppress corrosion... how hard could it be to nickle-plate a "disposable" razor? Hm.
Pure nickel is quite soft, so even though it would give you corrosion resistance, it wouldn't hold an edge well at all. Doing it well requires a bit more work. Fortunately, we have this cool system that lets people like you and me learn about ways people have already invented for dealing with problems like this. For example, you could read through one of Gillette's patents on the subject.
Oh wait: actually showing that a patent could be a useful resource doesn't quite fit with the slashdot "all patents are evil" groupthink. Forget I mentioned it!:-)
AMD dropped the prices when core2duo was announced dummyboy!
Of course AMD dropped their prices. In case you hadn't noticed, prices on CPUs have dropped on a regular basis ever since CPUs became commodity products. The news would have been if they were so far ahead of Intel on the price/performance curve that they didn't drop their prices.
Who would have been stupid enough to buy any of AMD processors had they not dropped the prices?? You??
What does the price on an AMD long before these Intel processors were available have to do with anything? Your statement is as pointless as saying the Intel 8088 was a terrible deal in 1982, because an AMD 3800+ costs half as much and runs a gazillion times as fast. Of course it does, but it just wasn't available in 1982!
I can buy matching up clock speeds as a possibility, but even assuming it's true, why does it make sense? If I'm buying a CPU, I'm concerned with what I'm going to get for my money. Anybody who still hasn't noticed that clock speed means nearly nothing needs to find a different line of work.
Quite a few people seem to have missed what seems to be a pretty obvious problem: the choices they've made as to what Intel processor to compare to what AMD processor just don't make sense. Look at the price table:
In every case, the Intel processor more expensive than the AMD to which they compare it. The Intel E6700 is over 60% more expensive than the AMD 5000+ they consider comparable. The Intel E6300 is not only more expensive than the AMD 3800+, but also more expensive than AMD's next step up, the 4200+.
Given their prices, the E6300 should obviously be compared to the 4200+ rather than to the 3800+. Looking at this particular pairing, rather than the nearly clean sweep for Intel, they each win some and lose some. If you simply count wins, the Intel wins more than the AMD -- but to mean much, you need to look at what they win at, not just how many different benchmarks they win. Just for example, PCMark05 goes 3:1 in favor of the E6300 -- but quite frankly, none of PCMark05 really means a thing.
Unless money is no object to you, the two lines look pretty closely matched. In video encoding and rendering tasks, Intel wins quite easily. In the ScienceMark scores, AMD wins pretty easily. Elsewhere, a lot are really too close to call based on the data provided. There are a number of cases in which each wins by less than 2%. It's impossible to say for sure without knowing things like the standard deviations on these scores, but there's a pretty fair chance they have no statistical significance at all.
Some people clearly just have too much time on their hands! Writing a serious paper about movie monsters is like thinking some silly reference to the Simpsons or Futurama is really funny.
Wow, look at that Karma go down the drain -- it's like after I finished cooking pasta tonight, only it took a second or two for the water to disappear after that!:-)
Note to self: next time right after last call, just think instead of posting!
Yep, for C++ at least Visual Studio's IDE peaked around version 6.
I'd argue the real peak was a bit earlier -- some of the earlier versions had a rather nice facility for organizing snippets of code that was inexcplicably removed in 6. Nonetheless, I'd agree that 6 was the last version before it all went complete to hades.
Although the standards compliance of the compiler has become much better in more recent versions [... ]
This is fairly easy (if somewhat expensive) to fix: Intel's compiler has substantially better conformance, and still installs with older versions of Visual Studio.
...just try looking in the help for a standard C++ library function in VS2005, with no.Net options or languages selected in the filters, and see how much.Net-related crap comes up).
For that matter, what would it take to get rid of all the WinCE garbage? I find that an even more complete PITA, personally.
Don't even get me started on Intellisense, and the various ways they've screwed up what should have been a really useful feature.
How about if, instead, I suggest you try Visual Assist X from Whole Tomato Software? You're right: IntelliSense was a fine idea, but Microsoft thoroughly screwed up the implementation. VA/X works naturally enough that I didn't even realize the degree to which I'd come to depend on it until I recently built up a new computer and tried to work without it for a while -- then the reality became quite obvious. It may easily be the single most nicely executed piece of software I've seen in quite a while.
And no, I'm not anything but a (highly) satisfied user -- though I've been a user long enough that the version history on their web site no longer covers the version I was using up until a couple of weeks ago...
For those who feel obliged to ask: no, it's not open source, it's not free (in either sense, though IMO, their pricing is quite reasonable) and it doesn't run (on) Linux. Sorry -- sometimes life is like that.
Actually the x86 takes the address, adds the segment base (which is zero most of the time in modern systems) and translates the result with the paging unit into a physical address.
Yup -- I accidentally swapped the two descriptions.
This seems to be the same as your description of what MULTICS expects.
For better or worse, it's not; MULTICS does expect the opposite of what the x86 provides -- I just accidentally swapped which was which when I desribed them.
That's simply incorrect. PL/I was chosen as the implementation language for MULTICS well before the first line of code was written. It was never written in assembly language. If you'd like to know some facts, consider reading a bit about the history of MULTICS.
The idea of the first portable operating system escaping the editors of this article is unforgivable.
Oddly enough, this is mostly true. Even though MULTICS was written in a high level language from the beginning, it wasn't very portable. It required a fairly heavy duty memory-management unit that most of the machines at the time simply didn't provide. It was a bit like a current x86 in protected mode, but in reverse. The x86 takes a virtual address and translates with with the paging unit to a linear address, then the segmentation unit (theoretically) does another translation on that to give a physical address. MULTICS required an MMU that took a segment-style address and translated it to a linear address, then a paging unit that translated that to a paged address.
Very few memory management units (then or now) provide that capability, and without it, MULTICS is pretty much dead in the water.
Slashdot Mirror
Don't be silly! CORDIC is an algorithm. Other than being something binary, the representation you use has no relevance at all (beyond the fact that the table of secant constants needs to be represented in the format of the result).
In fact, CORDIC is primarily useful when you don't have floating point hardware. If you have FP hardware (at least multipliers), you're frequently better off implementing one of the typical power series. In particular, CORDIC gives one more bit for each iteration, while a Taylor series (for only the most obvious example) doubles the number of bits with each term. If you only a few bits (e.g. 8 or 10) CORDIC works very well -- but if you have FP hardware and want something like the usual 64 bit result, 5 terms of the Taylor series will get you there faster and a Chebychev polynomial is generally faster still. Of course, back in the '50s when CORDIC was invented, the number of gates to do fast FP multiplication just wasn't available, so CORDIC ruled...
At that point it's not a guess any more -- it's just a job of reverse engineering some ancient, obsolete source code. Unless you had far more interesting tricks than anything you've mentioned here, there's not much point or fun in that.
Probably this thread is mostly forgotten by now, but just in case anybody's still paying attention, here's what the Oxford English Dictionary has to say about the situation:
The form INSURE is properly a mere variant of ensure, and still occasionally appears in all the surviving senses. In general usage, however, it is now limited to the financial sense (with reference to 'insurance' of life or property), in which the form ensure is wholly obsolete.So, at one time, the two were entirely synonymous, and now they're noticeably different. Nonetheless, although it's no longer in general usage, substituting "insure" for "ensure" doesn't necessarily change the meaning.
I'm afraid you're missing the point of anti-trust suits.
To the US government, it's more a case of: look, there are two companies making billions of dollars, and they hardly contribute anything to lobbying or campaigns! It must be a case worthy of an anti-trust suit!
Seriously, consider Microsoft's political contributions before and after they started being hit with anti-trust suits. It sure looks to me like their practices being officially approved was based more on contributions than on any actual change in how they do business.
There are simply too many different possibilities for there to be one reasonable guess. You can use shifts and adds, but that's rarely faster than the hardware, and at that time pipelining usually wasn't very useful. Table lookups of various sorts were fairly common, frequently with some bit-masking or -shifting to keep the table size reasonable.
CORDIC?
Since you haven't even specified what hardware it was for, attempting to guess at the low-level hardware trick(s) you used would be pretty pointless.
"infirm" ? Do you mean "deny", perhaps?
It wasn't the first game on a personal computer with real 3D -- I wrote some real 3D code that ran on a C64 around 1982 or '83. OTOH, the "ran" there is an exaggeration -- "walked" would be closer, and "stumbled" would probably be the right word. The code was all right, but there's only so much you can do on an 8-bit, 1 MHz processor. A couple of us played it enough to say it worked, but that's about it.
Oh, it would also depend a bit on what you're talking about as 3D -- this was wireframes only. I wrote some lighting code, but never used it except to create a couple of static images (one frame per night...)
It starts by taking a guess at the right answer, and then improving the guess until it's accurate enough to use.
The first step depends heavily on the fact that a floating point number on a computer is represented as a significand (aka mantissa) and an exponent (a power of two). For the moment, consider taking just the square root of X instead of its inverse. You could separate out the exponent part of the floating point number, divide it by two, and then put the result back together with the original significand, and have a reasonable starting point.
From there, you could improve your guesses to get a better approximation. The simplest version of that would be like a high-low game -- you split the difference between the current guess and the previous guess, and then add or subtract that depending on whether your previous guess was high or low. Eventually, you'll get arbitrarily close to the correct answer.
This can take quite a few iterations to get to the right answer though. To improve that, Newton-Raphson looks at the curve of the function you're working with, and projects a line tangent to the curve at the point of the current guess. Where that line crosses the origin gives you the next guess. That's probably a lot easier to understand from picture.
In this case, we're looking for the inverse square root, which changes the curve, but not the basic idea. As a general rule, the closer your first guess, the fewer iterations you need to get some particular level of accuracy. That's the point of the:
While the originator of this constant is unknown, and some of it is rather obscure, the basic idea of most of it is fairly simple: we start by shifting the original number right a bit. This divides both the mantissa and the exponent part by two, with the possibility that IF the exponent was odd, it shifts a bit from the exponent into the mantissa. The subtraction from the magic number then does a couple of things. For one thing, if a bit from the exponent was shifted into the mantissa, it removes it. The rest of the subtraction is trickier. If memory serves, it's based on the harmonic mean of the difference between sqrt(x) and (x/2) for every possible floating point number of the size you're using.
This is where the fact that it's 1/sqrt(x) instead of sqrt(x) means a lot: 1/sqrt(x) is a curve, but it's a fairly flat curve -- much flatter than sqrt(x). The result is that we can approximate a point on the curve fairly accurately with a line. In this case, it's really two lines, which gets it a bit closer still.
From there, the number has had a bit of extra tweaking done -- it doesn't actually give the most accurate first guess, but its errors are often enough in the opposite direction from those you get in the Newton-Raphson iteration steps that it gives slightly more accurate final results.
MS really wasn't lying when they said IE was now part of the OS (or at least the shell). For example, if you open "My Computer" and type something like "www.microsoft.com" into its address bar, you'll get essentially the same result as if you had started by opening a window that openly stated it was IE.
Much of what a typical user sees as "Windows" is really IE. You might easily be right that MS and their customers would be better off without it, but MS would have to do a fair amount of work on a new shell to get rid of it entirely, at least from XP. I'm not sure about the shell in Vista, but I doubt it's changed drastically in this respect.
The problem with that is that DirectX wasn't originated by Microsoft either. Direct3D started out as RenderMorphics RealityLab 2.0 under a new label. IOW, the "standard" already existed, and was already on the market before Microsoft acquired it.
A few other parts of DirectX might have been innovative, but when you get down to it, most of them were really just going back to the old way of doing things. Windows provided an interface to hardware at a level of abstraction that made it difficult to do anything but static graphics decently. Therefore, at the time, most attempts at animation and such were done outside of Windows. DirectX (as implied by the name) was intended as little more than a way to bypass the cruft they'd added on, and allow more direct access to the hardware, just about like if Windows wasn't there at all.
The other parts of DirectX provided decent access to sound (had been abstracted to the point that the primary capability was playing .wav files), joysticks (previously had to masquerade as a mouse to be supported by Windows at all) and so on. This wasn't innovation -- it was simply removing some of the most obvious and troublesome limitations in Windows.
Of course, since then, it's been updated quite a bit. Face reality though: most of the updates have been driven far more by Intel, ATI and nVidia than by Microsoft. They design hardware, and then DirectX is updated to provide access to the capabilities it provides (or will provide when it's finished -- it's a lot faster to describe the capability than implement it efficiently). It's open to argument that (perhaps) the geometry shaders added in DirectX 10 really are a Microsoft innovation -- though I suspect if you look hard, you'll find that nVidia and/or ATI really brought that to the table as well.
"The Java innovation" ? What was that?
Those who think Java was an innovation should do a bit of reading about the UCSD P-System. The idea of a virtual machine wasn't new then, but few (if any) had previously built it into a full-blown platform like the P-System.
Oh, make no mistake, I think most inventors do a fine job of disclosing everything of which they're aware. It does seem to me, however, that the system is almost surprisingly open to abuse.
You might be right -- but right now, I make enough to pay my bills, and I still have my self-respect, and I value that more than a bigger house, newer car, etc. I don't mean to sound particularly noble or anything, because I don't think I really am -- in fact, acting as an expert witness is comparatively unusual for me, so I doubt it'd make a lot of difference either way. If I never act as an expert witness again, I doubt it'll affect my income too drastically.
Not quite. They're obliged to disclose anything of which they're aware -- but they're not required to do any searching to find anything of which they weren't already aware. At least from what I've seen, most cite things they can't reasonably avoid (e.g. their own earlier patents, papers they've written, etc.) but darned little else. I've seen an "inventor" claim that he'd never read the data sheet for a part he used (he was just such a genius, he knew the pinout without looking...). The data sheet referenced an app note that almost directly disclosed his "invention"...
They can only lose much of anything if it can be shown that they really were aware of the material. If most do a better job than average, either there's a lot of skew in that average, or else you're using "average" a bit differently than I learned it in statistics class (or maybe my memory's going bad after all these years -- though I still very clearly remember a girl who always showed up in some of the tightest jeans I've ever seen...)
Having previously been an expert witness, and having declined the umm...honor in at least one other case, I think I can say with reasonable certainty that I'd decline in a case like that. Come to think of it, nobody's asked me to be an expert witness for a while though...
IMO, this is one of the real sticking points: right now, when somebody applies for a patent, they have to declare that they think it's novel. The patent examiner then looks to see if it look obvious (to them). The standard, at this point, however is only that the patent applicant has to show that the "preponderance of the evidence" is in their favor. Oh, and patent examiners virtually never look at anything except older patents and applications.
When a patent is challenged, however, that changes: the challenger now has to show "clear and convincing evidence" that the patent is really obvious after all. This means, from a legal viewpoint, that the patent is supposed to be considered valid, even if there's a fairly strong reason to believe otherwise.
To add to that, nearly every patent office (definitely including the US PTO) gets part or all of its funding from patents that it issues. Worse still, examiners are rated (partly) in terms of the number of applications they examine. Neither of these really encourages patent examiners to do be as thorough as possible and reject all patent applications that really should be (though, in fairness, given their workload, I'm frequently impressed at how good of a job patent examiners do anyway).
The motivation to combine must be a bit more than "motivation" as most people would think of it. Most of us would think of things like "it saves money" or "it makes the product smaller/lighter/more dependable" -- but when you're trying prove a patent is obvious, you're typically looking at something like combining two research articles, and the motivation to combine needs to be either a direct reference from one article to the other, or a third article that references both the ones you're looking at.
The standard for measurement of many things in patents is the "person of ordinary skill in the art" (POSITA). It seems to me that the POSITA has been reduced to something that bears essentially no resemblance to any real person at all. Instead, it's an almost purely mythical construct, with total awareness of all existing products and publications, and the ability to understand them, but essentially no ability to do any thinking at all. In particular, they can't draw even the most obvious conclusions unless something has been published to "suggest" (i.e. state) that conclusion for them. Even though they're theoretically aware of all papers, they're not able to put any two of them together without specific bibliographic references.
I know of at least one professor from Stanford who argued that putting a current limiting resistor in line with an LED was beyond the abilities of a person of ordinary skill in the art! If memory serves, he even won the case...
It's not really a dupe. You see, last time the "news" was that news.com was running a story about the case. This time, the "news" is that embedded.com is running a story about the case. The case itself is hardly news -- it's been around since April of 2005. Well, the real case is still older, but that's when they appealed to the Supreme Court.
Next week, it'll be news again -- and with a lot of luck they might find a link that's actually informative. Then again, that page has been around since June, so it's hardly news. Maybe coverage of the recent arguments would really be better. Oh wait...that's only a blog entry, undoubtedly written by some left wing nutcase. Maybe by next week, we can get a highly informed piece from a legitimate news service instead.
What, me sarcastic? Nevvvvverrrrrr!
The CEO of the RIAA claims: "We celebrate advances in technology and recognize the importance of finding new ways to deliver content." At the same time, the Chairman and CEO of Universal Music Group claims MP3 players are: "just repositories for stolen music, and they all know it."
It seems fairly clear that neither the RIAA nor their member companies really understand technology well at all. They try to stagnate music distribution in the mid-80's, and they disregard how easy it is to compare and contrast their messages to make the lies obvious.
Both stories showing up on the same day is just a nice bonus. It's a bit like the old Dilbert where the PHB is holding a meeting:
Item 1: record profits -- executives will be getting bonuses.
Item 2: financial difficulties -- layoffs of technical staff.
The point is to figure out what (if anything) should be done based on what's happening. If global warming is caused primarily by mankind releasing carbon dioxide into the atmostphere, then it would behoove us to reduce our carbon dioxide emissions to control global warming. OTOH, if global warming is mostly a result (for example) the earth receiving more energy from the sun, then reducing carbon dioxide emissions will have only minimal effect.
It should also be pointed out that while most people generally take for granted that nearly everything we do releases carbon dioxide and therefore leads to global warming, that's a bit one-sided. Just for example, burning coal not only releases carbon dioxide but also generally releases at least some sulfur. Sulfur in the atmosphere causes global cooling (i.e. it does more to reflect energy from the sun back out of the atmosphere than to trap energy from the sun in the atmosphere). If you compare sulfur release into the atmosphere to the global temperature, you get a fairly close (inverse) correlation. That would support an argument that most of the arguments over global warming have things backwards: rather than currently causing global warming, the real situation could be that we (mankind) were mostly responsible for the mini-iceage by burning sulfur-bearing coal. In the last few decades we've reduced coal usage and (particularly) reduced emissions when we do use it (e.g. most power plants now have equipment to remove sulfur from their emissions). What we're seeing as global warming is really just the earth recovering back to about where it would have been if we hadn't been causing global cooling for centuries.
Likewise, many people (including some here) have suggested that planting trees as an obvious cure. Simple solutions to complex problems sound nice, but (as in this case) things are rarely as simple as they initially appear. Much of the loss of rain-forest that's often cited is largely illusory. While it's true that what's classed as rain-forest has been reduced (somewhat) the losses are often at least partially offset in terms of overall wooded land. Taking a look at the UN's FAO data we see that while the forest land in South America dropped by about .4% annually between 1990 and 2005, when/if we take the other wooded area into account, the loss is really about .2% annually (they don't provide totals/percentages for the other wooded areas -- you have to cut-n-paste into a spreadsheet yourself to get those).
There are also problems with the idea itself. For one, large plantations of trees cause some environmental problems themselves. For another, depending on the latitude at which they're grown, trees can actually contribute to global warming.
Neither the problem, nor its cause, nor "the" solution is nearly as clear or certain as many would have you believe.
Nor any mention of people who release their software to the public domain, the really, honestly and truly good guys! :-)
Pure nickel is quite soft, so even though it would give you corrosion resistance, it wouldn't hold an edge well at all. Doing it well requires a bit more work. Fortunately, we have this cool system that lets people like you and me learn about ways people have already invented for dealing with problems like this. For example, you could read through one of Gillette's patents on the subject.
Oh wait: actually showing that a patent could be a useful resource doesn't quite fit with the slashdot "all patents are evil" groupthink. Forget I mentioned it! :-)
Of course AMD dropped their prices. In case you hadn't noticed, prices on CPUs have dropped on a regular basis ever since CPUs became commodity products. The news would have been if they were so far ahead of Intel on the price/performance curve that they didn't drop their prices.
What does the price on an AMD long before these Intel processors were available have to do with anything? Your statement is as pointless as saying the Intel 8088 was a terrible deal in 1982, because an AMD 3800+ costs half as much and runs a gazillion times as fast. Of course it does, but it just wasn't available in 1982!
I can buy matching up clock speeds as a possibility, but even assuming it's true, why does it make sense? If I'm buying a CPU, I'm concerned with what I'm going to get for my money. Anybody who still hasn't noticed that clock speed means nearly nothing needs to find a different line of work.
Look again at the first graph on the second page you've linked.
Given their prices, the E6300 should obviously be compared to the 4200+ rather than to the 3800+. Looking at this particular pairing, rather than the nearly clean sweep for Intel, they each win some and lose some. If you simply count wins, the Intel wins more than the AMD -- but to mean much, you need to look at what they win at, not just how many different benchmarks they win. Just for example, PCMark05 goes 3:1 in favor of the E6300 -- but quite frankly, none of PCMark05 really means a thing.
Unless money is no object to you, the two lines look pretty closely matched. In video encoding and rendering tasks, Intel wins quite easily. In the ScienceMark scores, AMD wins pretty easily. Elsewhere, a lot are really too close to call based on the data provided. There are a number of cases in which each wins by less than 2%. It's impossible to say for sure without knowing things like the standard deviations on these scores, but there's a pretty fair chance they have no statistical significance at all.
Some people clearly just have too much time on their hands! Writing a serious paper about movie monsters is like thinking some silly reference to the Simpsons or Futurama is really funny.
:-)
Wow, look at that Karma go down the drain -- it's like after I finished cooking pasta tonight, only it took a second or two for the water to disappear after that!
Note to self: next time right after last call, just think instead of posting!
I'd argue the real peak was a bit earlier -- some of the earlier versions had a rather nice facility for organizing snippets of code that was inexcplicably removed in 6. Nonetheless, I'd agree that 6 was the last version before it all went complete to hades.
This is fairly easy (if somewhat expensive) to fix: Intel's compiler has substantially better conformance, and still installs with older versions of Visual Studio.
For that matter, what would it take to get rid of all the WinCE garbage? I find that an even more complete PITA, personally.
How about if, instead, I suggest you try Visual Assist X from Whole Tomato Software? You're right: IntelliSense was a fine idea, but Microsoft thoroughly screwed up the implementation. VA/X works naturally enough that I didn't even realize the degree to which I'd come to depend on it until I recently built up a new computer and tried to work without it for a while -- then the reality became quite obvious. It may easily be the single most nicely executed piece of software I've seen in quite a while.
And no, I'm not anything but a (highly) satisfied user -- though I've been a user long enough that the version history on their web site no longer covers the version I was using up until a couple of weeks ago...
For those who feel obliged to ask: no, it's not open source, it's not free (in either sense, though IMO, their pricing is quite reasonable) and it doesn't run (on) Linux. Sorry -- sometimes life is like that.
Yup -- I accidentally swapped the two descriptions.
For better or worse, it's not; MULTICS does expect the opposite of what the x86 provides -- I just accidentally swapped which was which when I desribed them.
IBM VM/370 came out in 1972. I can't say I remember MVS at all well though -- back when I did mainframe stuff, it was mostly on Control Data machines.
That's simply incorrect. PL/I was chosen as the implementation language for MULTICS well before the first line of code was written. It was never written in assembly language. If you'd like to know some facts, consider reading a bit about the history of MULTICS.
Oddly enough, this is mostly true. Even though MULTICS was written in a high level language from the beginning, it wasn't very portable. It required a fairly heavy duty memory-management unit that most of the machines at the time simply didn't provide. It was a bit like a current x86 in protected mode, but in reverse. The x86 takes a virtual address and translates with with the paging unit to a linear address, then the segmentation unit (theoretically) does another translation on that to give a physical address. MULTICS required an MMU that took a segment-style address and translated it to a linear address, then a paging unit that translated that to a paged address.
Very few memory management units (then or now) provide that capability, and without it, MULTICS is pretty much dead in the water.