No, sorry, the Athlon core had no significant similarity to any DEC Alpha design. You're conflating the fact that some people left DEC and joined AMD into the cores being the same. As with any other tech industry, people change jobs in the semiconductor industry all the time, but the designs don't walk out the door with them unless they commit a crime! The only time that actually happens legitimately is when the change of employer is due to an acquisition or merger, and that wasn't how ex-DEC people ended up at AMD.
Only if you've never heard of a license agreement. The amount of cross-licensing that happens between microprocessor vendors is fairly prolific.
The only Alpha technology in the original 1999 Athlon was the frontside bus. If I recall correctly, it was licensed to AMD for free because DEC (or whichever post-DEC entity owned Alpha at that point) still harbored dreams of riding Windows NT for Alpha into the PC space, and had troubles getting anybody to build off the shelf systems. Being FSB compatible with the #2 x86 processor vendor was a great way to get their foot in the door of the PC market.
First of all, the bus is not any kind of small detail having no effect on processor architecture. It was a significant contributing factor to the Athlon being substantially faster than the Pentium III.
But then you look at the rest of it. The multiprocessor configuration is very similar, in part because of the shared bus. The Athlon and then-contemporary Alphas both have 64KB+64KB L1 instruction and data cache, exclusive 2-way associative. In later iterations (AMD64), they adopted even more Alpha characteristics, like the on-die memory controller. You can hardly say that they have "no significant similarity."
Who said you can only "optimize" an old design? You can do whatever you like with it, you don't have to stick to just minor tweaks to extract a few more MHz.
Which is what they've already done to get it from 500MHz to up near 4GHz over the course of more than a decade. But at some point you've done all you can with small and medium changes and you have to make big changes, which is what they're doing now.
I'll put it this way. Intel cleansheeted the Pentium IV, but that didn't work out so well. When they went back to basing new designs off the venerable Pentium Pro lineage, they started stomping AMD into the ground.
Core 2 was catch up. It was only somewhat faster than K8 and is about the same speed as K10. The subsequent improvements have been in large part adopting the things they learned from the Pentium 4 into the Core architecture. Sandy Bridge looks more like a modified Pentium 4 with a shorter pipeline than it does a Pentium Pro.
You seem to be making some nonsense argument about how you get from point A to point B, as if replacing pieces of an architecture one at a time gets you to a different place than replacing several things at once. The fact of the matter is that they needed to make some significant changes and they did. That is very hard to get right on the first try, so now they've got some bugs to work out -- why are you so sure that they can't?
Pretty sure it isn't actually a true "clean slate". There's lots of signs of bits & pieces carried over from K8 to Bulldozer and modified to fit the new system design. Which is entirely sensible, because reinventing almost everything from scratch ala Pentium IV is too much work for not enough reward.
Of course you don't have to reinvent the wheel. That isn't what I'm saying. What I'm saying is that they couldn't just keep making small changes and expect to stay ahead like they had been for the previous decade. They had to do a major update. There was no viable option that says "keep existing architecture with minimal changes."
Have you ever heard this parable? If you ask a coach to observe two runners who each make the same time around the tra
The i5-2500K is compared to the FX-8150 because the FX-8150 is $25 more expensive, and when AMD wins in a specific benchmark its only a small advantage going to AMD while when the Intel wins a specific benchmark its usually a significant advantage. The numbers speak for themselves [anandtech.com]
Which tells you that you probably ought not to buy the more expensive eight-thread version unless you have a workload that actually has eight threads.
Trying to compare to the FX-6200 is sort of laughable thing to suggest. I dont get how you claim that the FX-6200 is significantly cheaper or better performing at anything since you cant even buy one yet, and nobody has benchmarked one.
They seem to be on sale. Moreover, you can pretty well predict how they'll perform: Somewhat better than the FX-8150 on anything with three threads or less (owing to higher clock speed) and somewhat better than 3/4ths as well as the FX-8150 on anything with many threads (owing to higher clock speed but only 3/4ths as many cores).
That puts it in the same ballpark as the 2500k (at least to the extent that the FX-8150 is) but at a noticeably lower price point.
That's why they make circles. You put them in the one that doesn't let them see anything (or only lets them see things things you want them to see), and they just assume you don't have very much to say.
With Google's new anti-privacy policy, they are explicitly reserving the right to take all that information and do whatever they want with it internally.
Can you explain what it is that causes such alarm about the new privacy policy (other than that the WSJ keeps writing negative stories about Google)?
As far as I can tell there are two primary privacy questions when someone is using a web service:
(1) What data do they collect? This is important because eve if they don't intend to do anything objectionable, they could still be forced to turn it over to a totalitarian government or someone could break in and steal it and you would then see the problems like you see in (2). The trouble with this is that it's a losing battle -- everybody collects everything they can get their hands on, so unless you want to be like Richard Stallman and read the internet by having people print it out on paper for you, there isn't much you can do. Plus, especially in the case of the totalitarian government, there is no way to actually verify that they aren't being forced to collect more data than they say they are. More importantly, this isn't the thing that changed in the new privacy policy, so if you were OK with this before then nothing has changed.
Which leads to: (2) What do they use it for? The things to be worried about are that they provide it to insurance companies who use it to raise your rates or deny coverage, or provide it to governments who use it to silence dissidents, or that they're just loose with distributing it to anyone and it ends up in the hands of insurance companies and governments. But according to the privacy policy, that isn't what they're doing. They're using it internally, obviously to target ads. If you refrain from reading the ads, the effect on you is inconsequential. If you read them, you see ads that are somewhat more relevant... which doesn't seem like a particular cause for outrage.
So where is the path that leads to something bad happening? What terrible end result is enabled by the new privacy policy that was not possible under the old one?
It seems to me the reason for the real name policy is to make it so that people who know each other in meatspace can find one another. So what they could do is, collect your real name and then let you supply an alias which is the only thing any third party is ever allowed to see. The exception would be if someone searches for your real name, in which case the opposite would happen: It would show that you're a user and allow the person who did the search to send you a message or a request, but it wouldn't tell them your alias or let them see anything you've posted until after you've authorized them to do so. (And then you do so selectively by using circles etc.)
Is anyone from Google listening? I've got a feature request.
I wouldnt say that i5 buyers are howling high-end PC gamers.. the i5-2500K is cheaper and performs better on most benchmarks (including multi-threaded) than the FX-8150.
This is clearly the mid-range market with prices between $225 and $250, not the high end gaming market. AMD has some absolutely incredible budget offerings (the A8 series APU's for instance) that Intel simply cannot touch, but don't kid yourself about the mid-range market.. where Intel is highly competitive even when its not winning outright.
I don't think my point was to argue that AMD's offerings are superior to Intel's, but rather that they're competitive. And people are continuously comparing the i5 2500k to the FX-8150 because the comparison makes Intel look good, but try comparing it to something else, like the FX-6200. Because of the way modules work (i.e. they're faster per thread if you only run one thread), I would expect the FX-6200 to match or beat the 2500k at plenty of common workloads: If you're only running one thread (or really up to three threads) then the 2500k can't use all its cores and meanwhile each thread has an entire module to itself and a 500MHz clock speed advantage on the FX-6200, so you wouldn't expect the 2500k to win until you add the fourth thread, and then its advantage would fall off pretty quickly if you add a fifty and sixth thread that allow the FX-6200 to use two threads on all three modules. The 2500k might still be faster with six threads (though it would be pretty close), but then you consider that the FX-6200 is significantly cheaper.
What I'm saying is that it's a close decision, which means AMD is competitive, which means they'll be selling plenty of chips.
You are implying that only PC enthusiasts are interested in single-thread performance. How come that Intel invests so much in improving it, then, and does not lose its market share to AMD? This is also at odds with AMD's improvements in Turbo Core in Bulldozer.
That's kind of a silly question, isn't it? Why would they lose market share as a result of higher single-thread performance?
Intel is aiming for the broadest possible market. They want the subset of people who care about maximizing single thread performance in addition to the subset who care about maximizing multi-thread performance and the (by far largest) subset who don't really care about record-breaking performance very much at all and just want the best value for money.
AMD has concluded that they can't afford to match Intel's R&D budget in order to target every market segment and can get by with only the latter two. And they're not wrong -- they sell as many processors as they can produce.
I am afraid that either most of desktop workloads remain essentially single-threaded (e.g. compilers are single-threaded if you want to perform global (whole program) optimizations as opposed to trivial - but suboptimal - splitting the program to multiple compile units) or it is peak single threaded performance that affects the feeling of "snappiness" that customers want to pay money for.
You are apparently under the impression that CPU performance is still a major factor in "snappiness" in modern computers. The fact of the matter is that any K10, Bulldozer or Core processor in the vast majority of desktops spends 99% of its time idle. Moreover, the user without an unlimited budget will find that they experience better performance for the same price by combining an AMD processor with a Raptor or SSD than they would have using a more expensive Core processor with a slower disk. AMD can succeed perfectly well without beating Intel's fastest CPU on single thread performance as long as they can beat several of its slower CPUs at the same or lower price point.
The example of compiling stuff is particularly artificial because a) the vast majority of users don't ever compile anything (or, really, do anything CPU intensive whatsoever) and b) if you actually care about waiting for compiles to finish then you can turn off the most expensive performance optimizations until you get to deployment for beta testing because in the rare event that you discover it causes a major problem you can in the worst case just turn it back off.
Do the math here, if Bulldozer's cores were 60-80% of Intel's then their 8 core chip should perform 120-160% of Intel's quad core chips in multithreaded performance.
Only if by "do the math" you mean "ignore the math." Bulldozer modules are neither a complete pair of cores nor a single core with hyperthreading, remember.
If you run a single thread on a module, it doesn't have to share the FPU or caches with any other threads and will have higher performance -- hence 60-80% of Intel's on single threaded workloads. If you run two threads on the same module than they share some things but not everything as HT would, so instead of having total performance go up by a pittance as it does with HT, it goes up substantially by adding another thread (even though performance per thread goes down somewhat).
It's actually a pretty slick design if you think about it. If you have a multi-threaded workload then you get better-than-HT scaling and if you have a single-threaded workload you get an extra half a core worth of resources that can be applied to that thread. Clock-for-clock a BD module is pretty competitive with a Westmere-EP core on threaded workloads.
Bulldozer is AMD's Pentium IV
It doesn't really look like the Pentium 4 at all. The problem with the P4 was that the pipeline was way, way too long. Prescott had a more than 30 stage pipeline in order to achieve much higher clock speeds, and then it didn't achieve much higher clock speeds.
Bulldozer isn't about that. The pipeline is about the same length as Sandy Bridge (they're both a little more than half that of Prescott). About the only thing BD has in common with the P4 is that they both run hot. But the P4 ran hot because they were sacrificing everything trying to hit 4GHz on 90nm process technology. Notwithstanding the demos showing BD at 8GHz on liquid helium, AMD isn't chasing GHz here. The heat issues may be attributable to this being the first generation. It remains to be seen whether they can address them in the future, but there is no sense in counting them out this early in the design evolution.
Now, here's the puzzling part: they want to use bulldozer, the failure, as the new core for the A series, the success. I hope they find a way to fix it, otherwise my next rig will have an Intel for the first time in ten years.
I think the people calling bulldozer a failure have the wrong expectations. The core used in the existing A series is a direct descendant of the original Athlon from 1999, which itself was very similar to (and designed by the same people as) the DEC Alpha introduced in 1992, predating even the Pentium Pro. Suffice it to say that there isn't a lot of optimizing left to be done on the design.
Bulldozer is a clean slate. The current implementation has some obvious shortcomings, not least of which that the cache architecture is lame. (The L1 is too small and the L2 latency is too high. They might actually do pretty well to make a smaller, lower latency, non-exclusive L2 and use the extra transistors for a bigger L3 or even an L4.) But that's not a bad thing. It's something they can fix and make future generations faster than the current generation. Which is the problem with the old K10 -- there are no easy little changes left to be made to make it substantially faster than it is now.
The other part of the problem is that people want Bulldozer to be something it's not. It isn't designed for first in class single thread performance. It's designed to have adequate single thread performance while reducing the number of transistors per core so that you can have a lot of cores. It's designed for the server market, in other words. And to a lesser extent the workstation market. They designed something that would let them compete in the space that has the highest margins. So now all the high-end gamers who only care about single thread performance are howling at the moon because AMD concluded it couldn't compete with Intel in that sector and stopped trying.
What you have to realize is that it isn't that the design is flawed. It's that you aren't the target market. They could have built something that achieved 90-100% of Intel's best on single threads instead of 60-80% by doubling the number of transistors per thread and halving the number of threads and cores, but think about who would buy that. PC enthusiasts who comprise about 0% of the market. It wouldn't sell in the server market because the performance per core * number of cores would be lower. It wouldn't sell in the budget market because it would require too many transistors per thread and therefore cost too much to manufacture.
Instead, with Bulldozer they can use more modules and sell to the server market or anyone else with threaded software and then and use fewer modules in combination with a GPU and sell to the budget market and the midrange gaming market, and leave the six dozen howling high-end PC gamers to Intel.
Most of the time it has nothing to do with a desire to cheat, steal, or harm another person. It could just be ignorance and bad driving habits. You could be a great driver and habitually speed 5-10 mph above the limit. That is not, in of itself, a sign of sociopathic behavior.
Not only that, where are they getting their numbers? If it's the number of traffic tickets they could just as easily be measuring the propensity of cops to pull over people who drive more expensive cars.
I don't know, it seems to me it would be easier to get a message to a human pilot than a machine that can't understand, for example, skywriting containing an abort code, or some nonstandard method of conveying a message in morse code like sending a few hundred kilowatts through an arc gap which might be discernible even on a jammed frequency.
Secondly, I would expect military grade equipment to be fail-secure. That is, even if they did gain physical access, it would brick itself rather than allowing someone to make changes.
That's the trouble with military grade equipment. Which way is secure? You can't have a system where the enemy can cause your bombers to disarm themselves because they received invalid input from enemy transmitters.
I don't think the problem is flying it without wireless. You put a map in the thing's computer and it can pretty well get there on some combination of inertial guidance, gyroscopes and whatever else instruments you give it.
The problem is stopping it if the crisis is averted between when you launch and when the target is reached and someone wanting to cause WWIII starts jamming the wireless.
You pay capital gains tax on your "profit" either way.
Theoretically. But the point is that it changes when you pay. If you get a dividend, you pay taxes now. If the stock price goes up, you pay taxes when you sell the stock, which may be fifty years from now. Or never, if you pass it on to your kids. And in the meantime the would-be tax money is growing in the market.
One issue is that IIRC dividends are paid out after the corporate income tax, so some feel they are taxed twice.
That more or less applies to stock appreciation as well. If the value of the company goes up because it's holding more assets, the profits that paid for those assets were taxed to the corporation. And if you sell all your shares at the higher price then you pay capital gains tax on all the gains.
Change that and dividends are the only way because.... Your reliance on stock splits implies the company will grow exponentially and indefinitely - something that is not mathematically possible.
Using mathematical absolutes is not productive. A company can't grow exponentially and indefinitely, but it can grow exponentially for two hundred years or more.
In addition to that, "growth" and "stock price appreciation" are not necessarily related, and in ways that can make this wrong:
Your percent ownership in the company decreases (exponentially) every time you sell that 5 percent of your holdings.
Let's suppose that instead of issuing dividends, a company engages in stock buybacks. So the company starts out with a million shares and suppose you own a thousand of them. Over the course of ten years, the company buys back half a million shares and you sell five hundred of your thousand. You still own the same percentage of the company as you ever did, the company is still in the same financial condition as it would be if it had issued dividends instead of buying back stock, but you ended up deferring your taxes for a lot longer than you would have with dividends.
Then the company does a stock split, so it ends up with a million outstanding shares again and you end up with a thousand of those million again. (Of course, if someone keeps selling stock in order to get money, they're continually reducing their total basis and eventually they can no longer defer a substantial amount of taxes this way, but they can still defer the taxes indefinitely if they're willing to just hold all the shares without selling them.)
They probably have all the rights that Righthaven had.
And look at Righthaven today.
"As of November 2011, the company's assets are subject to confiscation by the US Marshals Service due to expired debts from legal fees to a successful defendant.[6] In January 2012, its domain name, righthaven.com, was sold at auction to an undisclosed purchaser to help satisfy its debts.[7]" h/t Wikipedia
Until there's a reasonable alternative, people aren't going to "stop posting new videos".
That would almost be a good point if it weren't for the fact that Hollywood keeps suing them all into bankruptcy, which pretty well leads back to them being the bad guys.
This story, about Rumblefish claiming to represent the owner of the rights to a birdsong makes me sick. I think the target of our anger should be Rumblefish entirely. They're the bullies. They're the bad guys.
I think everyone agrees that they're the bad guys. But they're not the only bad guys. Somebody else insisted that the system be set up in a way that allows them to do what they're doing.
That said, I agree with your general sentiment. It's past time that little piggies in brick houses stop taunting the big bad wolf before we get out the wrecking ball and have us some bacon.
While I agree that the price won't ever fall below the value of a company broken down and liquidated, I think that your assumption that stock price is in some way tied to company value is inherently wrong. If there's no dividend, then there's really -no- connection between a company's performance and stock price. It's 100% based on pure supply and demand.
Nonsense. Never mind the liquidation value, they're worth at least the going concern value to potential merger partners.
More importantly, you can't call it a ponzi scheme without throwing dividend-issuing companies into the same boat. If you can somehow continue issuing dividends perpetually, you can grow the stock price in exactly the same way merely by throwing the would-be dividends onto a giant money pile that keeps growing and consequently grows the value of the company even if the value of the company's non-liquid assets remains the same.
The fact of the matter is that a company can grow at the same rate as the overall economy pretty much indefinitely. But not all of them do. (And you can argue that the "overall economy" also can't continue to grow forever, but it seems that is a bridge we cross when we get to it.)
you're theoretically correct, but in practice, most (non-institutional) investors don't have the technical chops and the market knowledge/connections to judge valuations that well, so they are essentially betting (i.e., they're potential suckers - i'd guess that even professional investors often gamble based on intuition rather than analysis).
Of course they are. And for that reason they regularly get taken for a ride. But the market price for anything that even resembles a blue chip stock is set almost entirely by the institutional investors. They have by far the most capital behind them. Which isn't at all to say that their evaluations are perfect, but rather to say this: If you think you can do better than they can, you're probably wrong. (Though if you're right you'll make a mint.)
Of course I'm also sure that none of this has anything to do with the fact that YouTube gets a cut of those ad proceeds. And that a small user posting original content would probably opt not to insert ads, such that YouTube would be then getting a cut of zero.
I kind of doubt that's it. I mean sure, they get a cut of the ads, but YouTube actually has to care about what YouTube users think. There is, after all, no ad revenue if people stop posting new videos.
It seems to me more likely that the entertainment industry is behind it. Recall that YouTube has been trying to get Hollywood to let it compete with Hulu. Of course, Hollywood is run by tyrannical despots who claim to own everything anyone has ever created, so this sort of behavior is right up their alley. And, Hollywood has never cared one lick what the users think about it.
I mean think about it: Who is more likely to demand something user hostile? A company that makes its living based on users liking its service better than those of competitors, or a cartel that makes its money by filing lawsuits and screwing artists out of contractually agreed royalties?
I think the interesting thing is the demographics. Facebook was once the place to be because it started off as only college students, so everyone on there was respectable. Now G+ has got a bunch of professionals posting about software development and politics and Facebook is full of people's moms playing farmville.
If you buy some stock at $100 per share, and every year you get (say) $4 in dividends ($1 per share per quarter), you'll be earning 4% interest—this is better than just about all savings accounts, and quite a few bonds at the moment. If you bought stocks in a company at $100/share with no dividend, your money isn't exactly doing much for you until (a) the stock goes up in price, and (b) you sell it. At least with a dividend-paying stock you'll be earning cash regardless of the price of the stock.
The thing you're missing is that the dividend comes directly out of the stock price. If a company issues 4% of the value of the company as a dividend, the stock price immediately goes down by 4% because the company has 4% fewer assets (namely the cash just issued as a dividend). All the "profits" you got as dividends come straight out of the money you get back when you sell the stock. The fact that the stock hasn't (necessarily) gone down is totally irrelevant -- if the company had either held the cash or used it to buy back its own stock instead of issuing a dividend, the share price would be that much higher than it is after the dividend. TANSTAAFL. On top of that, you pay tax on dividends immediately, whereas you can defer paying taxes on stock appreciation until you actually sell the shares.
The problem is you're playing the greater fool game. If you intend to make money by selling a stock at a later date, then what you're saying is that you think you can find someone willing to buy AFTER you got the value out of it. Your buying strategy is also then based on inside information, "intuition" or some other divine insight, otherwise people would have already known and driven the price up. Another downside is that you have to actually SELL your assets to get any money from them.
That's not how it works. The stock price of a company has at least some relationship to the value of the company -- certainly it will almost never fall substantially below the liquidation value, because if it ever falls below it, there are plenty of vultures who will swoop in and buy it for just under that amount and then liquidate the assets at a profit.
So if you buy a stock for $100 and it goes up to $120, that is usually because the company is now worth more money. They might have more money in the bank, or their products might be more popular in the market and the expected future profits are higher, etc. The person who buys it is not (necessarily) a sucker -- they're buying something ostensibly worth $120 for $120.
What you're probably referring to is the efficient market hypothesis, which basically says that the risk-adjusted return for any stock is the same, because if it wasn't then market participants would sell stocks with lower risk-adjusted returns (thus lowering their price) and buy stocks with higher ones (thus raising their price), until the risk-adjusted return for both stocks is the same. (The ironic thing about the efficient market hypothesis is that it only works if market participants assume it isn't true -- because if people assume it's true it encourages people to be lazy in evaluating bargains and then it ceases to be true; but since they generally don't assume it's true, it tends to be true. Another way of saying this is "people who trade stocks on the basis of the efficient market hypothesis are suckers.")
But that doesn't have anything to do with whether someone who buys at a higher price is getting a better or worse deal. Stocks with higher volatility (like a lot of tech stocks, because fortunes change overnight in this industry) tend to have higher returns when they're doing well, because the probability is higher that you'll see losses than you will if you buy e.g. Walmart. In other words, the returns can be higher with Apple because the risk is higher which makes the risk-adjusted expected returns the same for both Apple and Walmart (and, if you buy the efficient market hypothesis, all other securities). There is no reason to expect that you'll do better buying Apple vs. Walmart, regardless of their past performance or anything of that nature ("past performance is no guarantee of future results"), because if there was any reason to expect that then the respective prices of those stocks would almost immediately change to reflect it and it would case to be the case.
The point being that if you buy a stock and then later sell it for a higher price, neither you nor the buyer are necessarily suckers. You just have different valuations of the value of the stock -- and the difference may be very small. If you think it isn't worth more than $100 and someone else thinks it's worth $100.20, you aren't having some great existential disagreement about the future of the company. You're disagreeing about whether the future risk-adjusted returns will be two tenths of a percent higher or lower and comparing that favorably or disfavorably with the risk-adjusted returns for other investments (which will be in the same range).
The old school way which is now gaining popularity again is to buy companies that pay dividends. This gives you a return without finding a fool to sell to or cashing out. It's how it's supposed to work. If the stock goes down, it's not even relevant unless they cut the dividend (there is eventually a c
Slashdot Mirror
No, sorry, the Athlon core had no significant similarity to any DEC Alpha design. You're conflating the fact that some people left DEC and joined AMD into the cores being the same. As with any other tech industry, people change jobs in the semiconductor industry all the time, but the designs don't walk out the door with them unless they commit a crime! The only time that actually happens legitimately is when the change of employer is due to an acquisition or merger, and that wasn't how ex-DEC people ended up at AMD.
Only if you've never heard of a license agreement. The amount of cross-licensing that happens between microprocessor vendors is fairly prolific.
The only Alpha technology in the original 1999 Athlon was the frontside bus. If I recall correctly, it was licensed to AMD for free because DEC (or whichever post-DEC entity owned Alpha at that point) still harbored dreams of riding Windows NT for Alpha into the PC space, and had troubles getting anybody to build off the shelf systems. Being FSB compatible with the #2 x86 processor vendor was a great way to get their foot in the door of the PC market.
First of all, the bus is not any kind of small detail having no effect on processor architecture. It was a significant contributing factor to the Athlon being substantially faster than the Pentium III.
But then you look at the rest of it. The multiprocessor configuration is very similar, in part because of the shared bus. The Athlon and then-contemporary Alphas both have 64KB+64KB L1 instruction and data cache, exclusive 2-way associative. In later iterations (AMD64), they adopted even more Alpha characteristics, like the on-die memory controller. You can hardly say that they have "no significant similarity."
Who said you can only "optimize" an old design? You can do whatever you like with it, you don't have to stick to just minor tweaks to extract a few more MHz.
Which is what they've already done to get it from 500MHz to up near 4GHz over the course of more than a decade. But at some point you've done all you can with small and medium changes and you have to make big changes, which is what they're doing now.
I'll put it this way. Intel cleansheeted the Pentium IV, but that didn't work out so well. When they went back to basing new designs off the venerable Pentium Pro lineage, they started stomping AMD into the ground.
Core 2 was catch up. It was only somewhat faster than K8 and is about the same speed as K10. The subsequent improvements have been in large part adopting the things they learned from the Pentium 4 into the Core architecture. Sandy Bridge looks more like a modified Pentium 4 with a shorter pipeline than it does a Pentium Pro.
You seem to be making some nonsense argument about how you get from point A to point B, as if replacing pieces of an architecture one at a time gets you to a different place than replacing several things at once. The fact of the matter is that they needed to make some significant changes and they did. That is very hard to get right on the first try, so now they've got some bugs to work out -- why are you so sure that they can't?
Pretty sure it isn't actually a true "clean slate". There's lots of signs of bits & pieces carried over from K8 to Bulldozer and modified to fit the new system design. Which is entirely sensible, because reinventing almost everything from scratch ala Pentium IV is too much work for not enough reward.
Of course you don't have to reinvent the wheel. That isn't what I'm saying. What I'm saying is that they couldn't just keep making small changes and expect to stay ahead like they had been for the previous decade. They had to do a major update. There was no viable option that says "keep existing architecture with minimal changes."
Have you ever heard this parable? If you ask a coach to observe two runners who each make the same time around the tra
The i5-2500K is compared to the FX-8150 because the FX-8150 is $25 more expensive, and when AMD wins in a specific benchmark its only a small advantage going to AMD while when the Intel wins a specific benchmark its usually a significant advantage. The numbers speak for themselves [anandtech.com]
Which tells you that you probably ought not to buy the more expensive eight-thread version unless you have a workload that actually has eight threads.
Trying to compare to the FX-6200 is sort of laughable thing to suggest. I dont get how you claim that the FX-6200 is significantly cheaper or better performing at anything since you cant even buy one yet, and nobody has benchmarked one.
They seem to be on sale. Moreover, you can pretty well predict how they'll perform: Somewhat better than the FX-8150 on anything with three threads or less (owing to higher clock speed) and somewhat better than 3/4ths as well as the FX-8150 on anything with many threads (owing to higher clock speed but only 3/4ths as many cores).
That puts it in the same ballpark as the 2500k (at least to the extent that the FX-8150 is) but at a noticeably lower price point.
Then I would assume you know what a meme is.
That's why they make circles. You put them in the one that doesn't let them see anything (or only lets them see things things you want them to see), and they just assume you don't have very much to say.
With Google's new anti-privacy policy, they are explicitly reserving the right to take all that information and do whatever they want with it internally.
Can you explain what it is that causes such alarm about the new privacy policy (other than that the WSJ keeps writing negative stories about Google)?
As far as I can tell there are two primary privacy questions when someone is using a web service:
(1) What data do they collect? This is important because eve if they don't intend to do anything objectionable, they could still be forced to turn it over to a totalitarian government or someone could break in and steal it and you would then see the problems like you see in (2). The trouble with this is that it's a losing battle -- everybody collects everything they can get their hands on, so unless you want to be like Richard Stallman and read the internet by having people print it out on paper for you, there isn't much you can do. Plus, especially in the case of the totalitarian government, there is no way to actually verify that they aren't being forced to collect more data than they say they are. More importantly, this isn't the thing that changed in the new privacy policy, so if you were OK with this before then nothing has changed.
Which leads to:
(2) What do they use it for? The things to be worried about are that they provide it to insurance companies who use it to raise your rates or deny coverage, or provide it to governments who use it to silence dissidents, or that they're just loose with distributing it to anyone and it ends up in the hands of insurance companies and governments. But according to the privacy policy, that isn't what they're doing. They're using it internally, obviously to target ads. If you refrain from reading the ads, the effect on you is inconsequential. If you read them, you see ads that are somewhat more relevant... which doesn't seem like a particular cause for outrage.
So where is the path that leads to something bad happening? What terrible end result is enabled by the new privacy policy that was not possible under the old one?
I've just had a thought.
It seems to me the reason for the real name policy is to make it so that people who know each other in meatspace can find one another. So what they could do is, collect your real name and then let you supply an alias which is the only thing any third party is ever allowed to see. The exception would be if someone searches for your real name, in which case the opposite would happen: It would show that you're a user and allow the person who did the search to send you a message or a request, but it wouldn't tell them your alias or let them see anything you've posted until after you've authorized them to do so. (And then you do so selectively by using circles etc.)
Is anyone from Google listening? I've got a feature request.
You must be new here.
I wouldnt say that i5 buyers are howling high-end PC gamers.. the i5-2500K is cheaper and performs better on most benchmarks (including multi-threaded) than the FX-8150.
This is clearly the mid-range market with prices between $225 and $250, not the high end gaming market. AMD has some absolutely incredible budget offerings (the A8 series APU's for instance) that Intel simply cannot touch, but don't kid yourself about the mid-range market.. where Intel is highly competitive even when its not winning outright.
I don't think my point was to argue that AMD's offerings are superior to Intel's, but rather that they're competitive. And people are continuously comparing the i5 2500k to the FX-8150 because the comparison makes Intel look good, but try comparing it to something else, like the FX-6200. Because of the way modules work (i.e. they're faster per thread if you only run one thread), I would expect the FX-6200 to match or beat the 2500k at plenty of common workloads: If you're only running one thread (or really up to three threads) then the 2500k can't use all its cores and meanwhile each thread has an entire module to itself and a 500MHz clock speed advantage on the FX-6200, so you wouldn't expect the 2500k to win until you add the fourth thread, and then its advantage would fall off pretty quickly if you add a fifty and sixth thread that allow the FX-6200 to use two threads on all three modules. The 2500k might still be faster with six threads (though it would be pretty close), but then you consider that the FX-6200 is significantly cheaper.
What I'm saying is that it's a close decision, which means AMD is competitive, which means they'll be selling plenty of chips.
You are implying that only PC enthusiasts are interested in single-thread performance. How come that Intel invests so much in improving it, then, and does not lose its market share to AMD? This is also at odds with AMD's improvements in Turbo Core in Bulldozer.
That's kind of a silly question, isn't it? Why would they lose market share as a result of higher single-thread performance?
Intel is aiming for the broadest possible market. They want the subset of people who care about maximizing single thread performance in addition to the subset who care about maximizing multi-thread performance and the (by far largest) subset who don't really care about record-breaking performance very much at all and just want the best value for money.
AMD has concluded that they can't afford to match Intel's R&D budget in order to target every market segment and can get by with only the latter two. And they're not wrong -- they sell as many processors as they can produce.
I am afraid that either most of desktop workloads remain essentially single-threaded (e.g. compilers are single-threaded if you want to perform global (whole program) optimizations as opposed to trivial - but suboptimal - splitting the program to multiple compile units) or it is peak single threaded performance that affects the feeling of "snappiness" that customers want to pay money for.
You are apparently under the impression that CPU performance is still a major factor in "snappiness" in modern computers. The fact of the matter is that any K10, Bulldozer or Core processor in the vast majority of desktops spends 99% of its time idle. Moreover, the user without an unlimited budget will find that they experience better performance for the same price by combining an AMD processor with a Raptor or SSD than they would have using a more expensive Core processor with a slower disk. AMD can succeed perfectly well without beating Intel's fastest CPU on single thread performance as long as they can beat several of its slower CPUs at the same or lower price point.
The example of compiling stuff is particularly artificial because a) the vast majority of users don't ever compile anything (or, really, do anything CPU intensive whatsoever) and b) if you actually care about waiting for compiles to finish then you can turn off the most expensive performance optimizations until you get to deployment for beta testing because in the rare event that you discover it causes a major problem you can in the worst case just turn it back off.
Do the math here, if Bulldozer's cores were 60-80% of Intel's then their 8 core chip should perform 120-160% of Intel's quad core chips in multithreaded performance.
Only if by "do the math" you mean "ignore the math." Bulldozer modules are neither a complete pair of cores nor a single core with hyperthreading, remember.
If you run a single thread on a module, it doesn't have to share the FPU or caches with any other threads and will have higher performance -- hence 60-80% of Intel's on single threaded workloads. If you run two threads on the same module than they share some things but not everything as HT would, so instead of having total performance go up by a pittance as it does with HT, it goes up substantially by adding another thread (even though performance per thread goes down somewhat).
It's actually a pretty slick design if you think about it. If you have a multi-threaded workload then you get better-than-HT scaling and if you have a single-threaded workload you get an extra half a core worth of resources that can be applied to that thread. Clock-for-clock a BD module is pretty competitive with a Westmere-EP core on threaded workloads.
Bulldozer is AMD's Pentium IV
It doesn't really look like the Pentium 4 at all. The problem with the P4 was that the pipeline was way, way too long. Prescott had a more than 30 stage pipeline in order to achieve much higher clock speeds, and then it didn't achieve much higher clock speeds.
Bulldozer isn't about that. The pipeline is about the same length as Sandy Bridge (they're both a little more than half that of Prescott). About the only thing BD has in common with the P4 is that they both run hot. But the P4 ran hot because they were sacrificing everything trying to hit 4GHz on 90nm process technology. Notwithstanding the demos showing BD at 8GHz on liquid helium, AMD isn't chasing GHz here. The heat issues may be attributable to this being the first generation. It remains to be seen whether they can address them in the future, but there is no sense in counting them out this early in the design evolution.
Hasn't AMD already shipped 4GHz+ processors?
Now, here's the puzzling part: they want to use bulldozer, the failure, as the new core for the A series, the success. I hope they find a way to fix it, otherwise my next rig will have an Intel for the first time in ten years.
I think the people calling bulldozer a failure have the wrong expectations. The core used in the existing A series is a direct descendant of the original Athlon from 1999, which itself was very similar to (and designed by the same people as) the DEC Alpha introduced in 1992, predating even the Pentium Pro. Suffice it to say that there isn't a lot of optimizing left to be done on the design.
Bulldozer is a clean slate. The current implementation has some obvious shortcomings, not least of which that the cache architecture is lame. (The L1 is too small and the L2 latency is too high. They might actually do pretty well to make a smaller, lower latency, non-exclusive L2 and use the extra transistors for a bigger L3 or even an L4.) But that's not a bad thing. It's something they can fix and make future generations faster than the current generation. Which is the problem with the old K10 -- there are no easy little changes left to be made to make it substantially faster than it is now.
The other part of the problem is that people want Bulldozer to be something it's not. It isn't designed for first in class single thread performance. It's designed to have adequate single thread performance while reducing the number of transistors per core so that you can have a lot of cores. It's designed for the server market, in other words. And to a lesser extent the workstation market. They designed something that would let them compete in the space that has the highest margins. So now all the high-end gamers who only care about single thread performance are howling at the moon because AMD concluded it couldn't compete with Intel in that sector and stopped trying.
What you have to realize is that it isn't that the design is flawed. It's that you aren't the target market. They could have built something that achieved 90-100% of Intel's best on single threads instead of 60-80% by doubling the number of transistors per thread and halving the number of threads and cores, but think about who would buy that. PC enthusiasts who comprise about 0% of the market. It wouldn't sell in the server market because the performance per core * number of cores would be lower. It wouldn't sell in the budget market because it would require too many transistors per thread and therefore cost too much to manufacture.
Instead, with Bulldozer they can use more modules and sell to the server market or anyone else with threaded software and then and use fewer modules in combination with a GPU and sell to the budget market and the midrange gaming market, and leave the six dozen howling high-end PC gamers to Intel.
Most of the time it has nothing to do with a desire to cheat, steal, or harm another person. It could just be ignorance and bad driving habits. You could be a great driver and habitually speed 5-10 mph above the limit. That is not, in of itself, a sign of sociopathic behavior.
Not only that, where are they getting their numbers? If it's the number of traffic tickets they could just as easily be measuring the propensity of cops to pull over people who drive more expensive cars.
I don't know, it seems to me it would be easier to get a message to a human pilot than a machine that can't understand, for example, skywriting containing an abort code, or some nonstandard method of conveying a message in morse code like sending a few hundred kilowatts through an arc gap which might be discernible even on a jammed frequency.
Secondly, I would expect military grade equipment to be fail-secure. That is, even if they did gain physical access, it would brick itself rather than allowing someone to make changes.
That's the trouble with military grade equipment. Which way is secure? You can't have a system where the enemy can cause your bombers to disarm themselves because they received invalid input from enemy transmitters.
I don't think the problem is flying it without wireless. You put a map in the thing's computer and it can pretty well get there on some combination of inertial guidance, gyroscopes and whatever else instruments you give it.
The problem is stopping it if the crisis is averted between when you launch and when the target is reached and someone wanting to cause WWIII starts jamming the wireless.
You pay capital gains tax on your "profit" either way.
Theoretically. But the point is that it changes when you pay. If you get a dividend, you pay taxes now. If the stock price goes up, you pay taxes when you sell the stock, which may be fifty years from now. Or never, if you pass it on to your kids. And in the meantime the would-be tax money is growing in the market.
One issue is that IIRC dividends are paid out after the corporate income tax, so some feel they are taxed twice.
That more or less applies to stock appreciation as well. If the value of the company goes up because it's holding more assets, the profits that paid for those assets were taxed to the corporation. And if you sell all your shares at the higher price then you pay capital gains tax on all the gains.
Change that and dividends are the only way because.... Your reliance on stock splits implies the company will grow exponentially and indefinitely - something that is not mathematically possible.
Using mathematical absolutes is not productive. A company can't grow exponentially and indefinitely, but it can grow exponentially for two hundred years or more.
In addition to that, "growth" and "stock price appreciation" are not necessarily related, and in ways that can make this wrong:
Your percent ownership in the company decreases (exponentially) every time you sell that 5 percent of your holdings.
Let's suppose that instead of issuing dividends, a company engages in stock buybacks. So the company starts out with a million shares and suppose you own a thousand of them. Over the course of ten years, the company buys back half a million shares and you sell five hundred of your thousand. You still own the same percentage of the company as you ever did, the company is still in the same financial condition as it would be if it had issued dividends instead of buying back stock, but you ended up deferring your taxes for a lot longer than you would have with dividends.
Then the company does a stock split, so it ends up with a million outstanding shares again and you end up with a thousand of those million again. (Of course, if someone keeps selling stock in order to get money, they're continually reducing their total basis and eventually they can no longer defer a substantial amount of taxes this way, but they can still defer the taxes indefinitely if they're willing to just hold all the shares without selling them.)
And look at Righthaven today.
"As of November 2011, the company's assets are subject to confiscation by the US Marshals Service due to expired debts from legal fees to a successful defendant.[6] In January 2012, its domain name, righthaven.com, was sold at auction to an undisclosed purchaser to help satisfy its debts.[7]" h/t Wikipedia
It gets better.
Until there's a reasonable alternative, people aren't going to "stop posting new videos".
That would almost be a good point if it weren't for the fact that Hollywood keeps
suing them all into bankruptcy, which pretty well leads back to them being the bad guys.
This story, about Rumblefish claiming to represent the owner of the rights to a birdsong makes me sick. I think the target of our anger should be Rumblefish entirely. They're the bullies. They're the bad guys.
I think everyone agrees that they're the bad guys. But they're not the only bad guys. Somebody else insisted that the system be set up in a way that allows them to do what they're doing.
That said, I agree with your general sentiment. It's past time that little piggies in brick houses stop taunting the big bad wolf before we get out the wrecking ball and have us some bacon.
While I agree that the price won't ever fall below the value of a company broken down and liquidated, I think that your assumption that stock price is in some way tied to company value is inherently wrong. If there's no dividend, then there's really -no- connection between a company's performance and stock price. It's 100% based on pure supply and demand.
Nonsense. Never mind the liquidation value, they're worth at least the going concern value to potential merger partners.
More importantly, you can't call it a ponzi scheme without throwing dividend-issuing companies into the same boat. If you can somehow continue issuing dividends perpetually, you can grow the stock price in exactly the same way merely by throwing the would-be dividends onto a giant money pile that keeps growing and consequently grows the value of the company even if the value of the company's non-liquid assets remains the same.
The fact of the matter is that a company can grow at the same rate as the overall economy pretty much indefinitely. But not all of them do. (And you can argue that the "overall economy" also can't continue to grow forever, but it seems that is a bridge we cross when we get to it.)
you're theoretically correct, but in practice, most (non-institutional) investors don't have the technical chops and the market knowledge/connections to judge valuations that well, so they are essentially betting (i.e., they're potential suckers - i'd guess that even professional investors often gamble based on intuition rather than analysis).
Of course they are. And for that reason they regularly get taken for a ride. But the market price for anything that even resembles a blue chip stock is set almost entirely by the institutional investors. They have by far the most capital behind them. Which isn't at all to say that their evaluations are perfect, but rather to say this: If you think you can do better than they can, you're probably wrong. (Though if you're right you'll make a mint.)
Of course I'm also sure that none of this has anything to do with the fact that YouTube gets a cut of those ad proceeds. And that a small user posting original content would probably opt not to insert ads, such that YouTube would be then getting a cut of zero.
I kind of doubt that's it. I mean sure, they get a cut of the ads, but YouTube actually has to care about what YouTube users think. There is, after all, no ad revenue if people stop posting new videos.
It seems to me more likely that the entertainment industry is behind it. Recall that YouTube has been trying to get Hollywood to let it compete with Hulu. Of course, Hollywood is run by tyrannical despots who claim to own everything anyone has ever created, so this sort of behavior is right up their alley. And, Hollywood has never cared one lick what the users think about it.
I mean think about it: Who is more likely to demand something user hostile? A company that makes its living based on users liking its service better than those of competitors, or a cartel that makes its money by filing lawsuits and screwing artists out of contractually agreed royalties?
I think the interesting thing is the demographics. Facebook was once the place to be because it started off as only college students, so everyone on there was respectable. Now G+ has got a bunch of professionals posting about software development and politics and Facebook is full of people's moms playing farmville.
If you buy some stock at $100 per share, and every year you get (say) $4 in dividends ($1 per share per quarter), you'll be earning 4% interest—this is better than just about all savings accounts, and quite a few bonds at the moment. If you bought stocks in a company at $100/share with no dividend, your money isn't exactly doing much for you until (a) the stock goes up in price, and (b) you sell it. At least with a dividend-paying stock you'll be earning cash regardless of the price of the stock.
The thing you're missing is that the dividend comes directly out of the stock price. If a company issues 4% of the value of the company as a dividend, the stock price immediately goes down by 4% because the company has 4% fewer assets (namely the cash just issued as a dividend). All the "profits" you got as dividends come straight out of the money you get back when you sell the stock. The fact that the stock hasn't (necessarily) gone down is totally irrelevant -- if the company had either held the cash or used it to buy back its own stock instead of issuing a dividend, the share price would be that much higher than it is after the dividend. TANSTAAFL. On top of that, you pay tax on dividends immediately, whereas you can defer paying taxes on stock appreciation until you actually sell the shares.
The problem is you're playing the greater fool game. If you intend to make money by selling a stock at a later date, then what you're saying is that you think you can find someone willing to buy AFTER you got the value out of it. Your buying strategy is also then based on inside information, "intuition" or some other divine insight, otherwise people would have already known and driven the price up. Another downside is that you have to actually SELL your assets to get any money from them.
That's not how it works. The stock price of a company has at least some relationship to the value of the company -- certainly it will almost never fall substantially below the liquidation value, because if it ever falls below it, there are plenty of vultures who will swoop in and buy it for just under that amount and then liquidate the assets at a profit.
So if you buy a stock for $100 and it goes up to $120, that is usually because the company is now worth more money. They might have more money in the bank, or their products might be more popular in the market and the expected future profits are higher, etc. The person who buys it is not (necessarily) a sucker -- they're buying something ostensibly worth $120 for $120.
What you're probably referring to is the efficient market hypothesis, which basically says that the risk-adjusted return for any stock is the same, because if it wasn't then market participants would sell stocks with lower risk-adjusted returns (thus lowering their price) and buy stocks with higher ones (thus raising their price), until the risk-adjusted return for both stocks is the same. (The ironic thing about the efficient market hypothesis is that it only works if market participants assume it isn't true -- because if people assume it's true it encourages people to be lazy in evaluating bargains and then it ceases to be true; but since they generally don't assume it's true, it tends to be true. Another way of saying this is "people who trade stocks on the basis of the efficient market hypothesis are suckers.")
But that doesn't have anything to do with whether someone who buys at a higher price is getting a better or worse deal. Stocks with higher volatility (like a lot of tech stocks, because fortunes change overnight in this industry) tend to have higher returns when they're doing well, because the probability is higher that you'll see losses than you will if you buy e.g. Walmart. In other words, the returns can be higher with Apple because the risk is higher which makes the risk-adjusted expected returns the same for both Apple and Walmart (and, if you buy the efficient market hypothesis, all other securities). There is no reason to expect that you'll do better buying Apple vs. Walmart, regardless of their past performance or anything of that nature ("past performance is no guarantee of future results"), because if there was any reason to expect that then the respective prices of those stocks would almost immediately change to reflect it and it would case to be the case.
The point being that if you buy a stock and then later sell it for a higher price, neither you nor the buyer are necessarily suckers. You just have different valuations of the value of the stock -- and the difference may be very small. If you think it isn't worth more than $100 and someone else thinks it's worth $100.20, you aren't having some great existential disagreement about the future of the company. You're disagreeing about whether the future risk-adjusted returns will be two tenths of a percent higher or lower and comparing that favorably or disfavorably with the risk-adjusted returns for other investments (which will be in the same range).
The old school way which is now gaining popularity again is to buy companies that pay dividends. This gives you a return without finding a fool to sell to or cashing out. It's how it's supposed to work. If the stock goes down, it's not even relevant unless they cut the dividend (there is eventually a c