There are not many problems these days that cannot be parallelized and split up to be run on a large number of off the shelf hardware. It is much easier to grow a Beowulf Cluster to add performance than redesigning to eke out every bit of capability of top-of-the-line hardware. Much easier also, to redesign your problem so that it can take advantage of parallelism. I agree that this was probably a boondoggle by a politician wanting to get some publicity for himself.
You're mistaken. There's a large class of problems that are pleasantly parallel and can be split up like you say (e.g. einstein@home or seti@home type problems). However, any problem that requires a lot of internode communication such as computation fluid dynamics, gravity simulations, weather or climate simulations/forecasting, combustion/flame problems (e.g. modeling engines), molecular dynamics will require a system like this. A beowulf cluster using ethernet to connect nodes together will result in most of the cpus waiting for information from neighboring nodes to be sent to it so that it can go through an iteration. A lot of the cost in a system like this comes from having a very low latency, high speed network connections. Ideally, you'd want to have every cpu connected to every other cpu, but that is impossible so you end up trying to maximize the number of connections and bandwidth while minimizing the collisions with other cpu-cpu communications for a given amount of money. It's not cheap by any means.
Well, if you want that kind of resource, Amazon is very happy to sell it to you these days. In 2008, it was still a novel concept. Assuming that a government project should be able to spearhead such a development, especially with a huge one-time investment in hardware, that's the real stupidity.
What???? Amazon EC2 instances aren't comparable because they have much more latency for internode communications. In any case, if you have a decent workload, EC2 is really expensive. Using 2 large instances for compute nodes and using 50TB of storage will cost you about $7500 a month. Amazon's calculator gives an estimate of $30k a month for a HPC cluster. At that pricing, you can easily buy comparable equipment and come out ahead even with power, maintenance, and people if you're using it regularly. EC2 only makes sense if you need this sort of computational power for a week or so every few months.
Whether you're liberal or conservative, does anyone really believe that the government spending tax dollars on expensive speculative investments makes sense?
You mean like basic research on things that may not be realizable for a decade or two? What's your feeling on the internet which grew out of research on networking in the 70s and 80s. What about the funding for ultrafast networking that's happening now? What about things like the tevatron and LHC which resulted in things like MRIs being made feasible?
My experience is it would be better to provision a cluster of EC2 boxes to run the task than build a purpose-built super computer (with some exception). One disadvantage of clustered machines is longer communication latency, so tasks that require lots of process to process communication will run slower. Many problems can be tweaked with search spaces sliced so that this latency is not a big deal.
There are huge classes of problems were you can't tweak things like this. Basically any simulation where things are large distances interact or where there is a lot of communication can't really be shoved into a cluster. For example, computation fluid dynamics (e.g. anything looking at air or water moving over surfaces), weather simulations, molecular dynamics, simulating gravity, etc. All of these types of problems will run like crap if you try to use EC2 instances for them.
Also, have you really priced out what computation and data storage on EC2 costs? There's a few studies that show that EC2 on-demand instance will cost you 2-3 times more than purchasing a comparable server even with power, cooling, and maintenance/administration factored in. See, this or this for example. EC2 is great if you want to explore certain problems and need to temporarily scale up or want the ability to scale up on demand but if you have a base level of work that you'll be doing all the time, it's much more efficient to buy your own hardware. That is doubly true if your problems need any significant amount of storage space.
To replace the key and the boot-loader you have to disable "Secure Boot" in the firmware (Disabling by software is not allowed), then update the key (Means flashing a new version of the firmware) and the boot-loader and then reactivate "Secure Boot".
Now think of Average Joe or your grand mother and tell me how someone like them will accomplish this.
Replacing the keys doesn't require reflashing the firmware, you just need go into the UEFI setup screen and add or delete the keys you're interested in. If the key gets compromised, you just go to the setup, add the new key, boot and update the bootloader and go into the setup and remove the old key. Or, even easier, you update the boot-loader on a working system, then go into the UEFI setup and remove the old key and add the new key. The procedure you outlined is unnecessarily complex even assuming that you have to reflash the firmware to get new keys.
So the FSF is basically asking people to sign a petition that asks manufacturers to do what they are already doing and plan on doing ? The current requirements for windows 8 is that users must be able to disable secure boot in the bios and do key management (addition/removal) of keys as well. I don't know of any manufacturer that is planning on doing anything different since that would mean that their systems would not be windows 8 certified.
In fact, I don't think microsoft bans having other keys besides their key in the bios by default.If, for example, the FSF or some coalition (e.g. RedHat, Ubuntu, Debian, etc.) were to come up with some workable way key signing infrastructure, they could petition UEFI/mobo developers to include their keys in shipped products as well. The question is how do you freely allow people to get bootloaders signed without making it easily for malware authors to do the same.
Yeah, yeah, yeah - I realize a single person's anecdotal evidence doesn't carry much weight. I wonder what the statistics are though? As AaronLS already pointed out, these tests seem to indicate that my situation isn't very unusual. Components age and wear out.
Check out "A study of DRAM failures in the field" from the supercomputing 2012 proceedings. They have some interesting stats based on 5 million DIMM days of operation.
That's not true. There was a recent paper looking at memory defects and causes on the Jaguar supercomputer, and memory errors were moderately common. Just as surprisingly, there were errors were a single DIMM going bad would cause errors for all the DIMMs on that channel.
So, memory does go bad and it does that more frequently than you'd expect.
Whatever the motivation, I think this is one of the strengths of the scientific method and thus, one of the reasons for its success: we aren't quick to publishing until it is just right, and therefore, perhaps our best approximation of the "truth" we can muster.
On the other hand: What if you did publish your work early and often, not as concerned with slowly and deliberately ensuring everything is just right before spreading the information -- Not keeping quiet just so that you can be the one with the badge of "1st"? Why, then worldwide cooperation could kick in. Perhaps other interested parties would help you prove or disprove the results much more quickly. Thus, accelerating the speed of scientific progress. Now, don't get me wrong. I'm not saying it's wrong to not publish things that you aren't absolutely sure about, just that what you're doing seems really strange to me. Have you questioned your information dissemination methods? A good scientist would... Why, there could have been times that you were wrong about being wrong, i.e. made a discovery but never known it too... Perhaps it's time to re-think the system of publishing altogether?
The problem is that most early results are incorrect and after doing some checking, it turns out they were systematic errors or mistakes or something similar. If everyone published early and often, you'd get so many results (with most of the results being incorrect) that no one could track it to figure out which results were interesting and worth investing the time and effort to work on. Duplicating someone else's work takes a lot of time and effort and may involve building a lot of stuff or flying to another lab to learn new techniques. Unless you're really sure that the results are likely to pan out, why would anyone spend tens of thousands of dollars or a few months at another lab learning a technique?
The costs and startup efforts are much higher for most sciences unlike code so open source techniques won't work as effectively. It's effectively like having to reimplement a good portion of a piece of software before you can start contributing.
It appears to me that he is presenting two scenarios: 1. You are not an angel investor and what you do doesn't net you money or, 2. your a successful angel investor but your scheme only works because investing works.
I don't think he misunderstands financial markets at all. This looks like a reading comprehension fail to me.
A few things. Angel investors usually get involved in a company at it's initial stages and don't buy the company's stock in the open market. But let's ignore that and consider the scenarios. Suppose you bought a bunch of Intel stock on an exchange somewhere and you bought enough that Intel stock went up an appreciable amount. You almost certainly bought the stock from other investors or a market maker so Intel didn't get any money from your stock purchases. Intel's market valuation went up but that won't affect Intel directly or give it more money. Intel would have to sell it's stock on an exchange or put it up for collateral or something similar if it wants to take advantage of the rise in it's stock.
To use an analogy, if the value of your house or car went up by $50k, that wouldn't directly help you out. You would have to sell it or take out a loan against it or something similar if you wanted take advantage of the rise in value. Likewise, stock prices don't directly help companies, contrary to what the submitter says.
[I]f you buy company X's stock, company X will probably not get any actual money from the purchase.
No, but you will raise the value of the company, which has the same effect.
Not quite, the market valuation of the company increases but that doesn't translate into money that the company can use. The company would have to either sell stock into the open market or use it as collateral or something similar in order to get money. It's just as if you owned a home and the value of the home increased after you bought it. The increase in value doesn't do much for you unless you do other stuff (refinance, get a home equity loan, etc.) to take advantage of the that increase.
The same kind of trick wouldn't normally work on the stock market -- if you're wealthy enough that you can increase the share price of a stock by buying enough of it to shift the market, then when you try to reap your profits by unloading the stock, the price will drift back down as you're selling it off. (Or if your purchases do manage to create a self-fulfilling prophecy -- your infusion of cash into the company enables them to realize their plans and become a genuine success -- well, then you're just a successful angel investor, more power to you.)
You have a misunderstanding of how the stock market works. Namely, if you buy company X's stock, company X will probably not get any actual money from the purchase. You're almost certainly purchasing the stock from another investor so you wouldn't be an angel investor. A company can take advantage of a rise in it's stock price by selling it's own stock or by using stock to purchase another company or something like that but that is a side effect of someone pushing up stock prices.
If the author doesn't understand a simple thing like that about financial markets, then I don't have much faith in his ability to talk cogently about markets in general.
Right, because you have no right to do that with a device you supposedly own.
The specs already require that the x86 EFI allows you to load your own key. This is just something to let you install and use linux or other OSes without having to go through the process of loading your own keys into the bios and instead using the ms key that's already been loaded.
Most fuel consumed by airliners is done while rolling around the airport on the ground. A jet engine burns almost the same amount of fuel at idle as it does while in cruise. To start, older planes should be retrofitted with electric landing gear and engine start should happen at the hold short line when they're #1 for takeoff.
Imagine how much $8.00/gallon jet fuel is burned on the tarmac.
What the hell is electric landing gear? The wheels on the plan are unpowered and spin freely. All of the propulsion for moving around is provided by the engines. You can't keep the engines off until you're on the runway unless you're being towed. Also the engines need to be started using an external device so you'd need to drag that along so that it could spin up the engine and then start it.
Because he had some perfectionist tendencies. That was one thing that set Apple apart from Microsoft - you might not like what they did, but they usually did it thoroughly. That seems to be falling apart a bit.
Sort of like how mobileme worked amazingly well out of the box. Or how about siri working well with accents or just in general with a variety of voices. Same deal with the antennas on the iphone 4. Apple has a fairly long history of hardware issues on their first version of any new hardware. Steve Jobs being a perfectionist didn't prevent this.
'For all of Intel's semiconductor design and manufacturing feats, its processors suffer from a genetic handicap: They have to support the legacy x86 instruction set, and thus they're inherently more complicated than legacy-free ARM devices, they require more transistors, more silicon.
Intel and AMD x86 processors moved on to using micro-ops and risc like operations internally years ago. The only disadvantage nowadays is a small translator that converts x86 machine code into micro-ops. Compared to the actual logic or cache on the cpu the number of transistors that the translation takes is minimal and not a big deal especially when you consider the size of cpus nowadays.
Where are you pulling your numbers from? I would like to know how you came to your price figures if they actually did that.
I don't care if you were sarcastic, I'm serious. I would like to know what the cost difference would be if the iPhone 5 was made in the USA versus China.
There have been studies that estimates are about $30 to $160 more per iphone in costs ( http://www.businessinsider.com/apple-iphone-manufacturing-cost-foxconn-2012-4) . That means apple's margin for the devices would go from $452 in gross profits to around $293 per iphone. It'd cost more but wouldn't be outrageously more.
1) Those who have refused to sign the NPT. These countries are allowed to develop nuclear weapons. They include South Africa, Israel, Pakistan, and India. They receive NO assistance from the international community. They have to develop their entire nuclear program from scratch.
Uh...The french gave the israelis the nuclear reactor at Dimona which was the source of the nuclear material the israelis used to get construct their nukes. The British also later gave them further assistance in getting the reactor running and in purifying and producing the components for a nuke.
India got nuclear reactors from Canada and apparently got nuclear technology from other NPT signatories as well.
So, it is saying that a car with an engine that can get 400mpg is more economical than one with 30mpg, but they leave out the important part that it will take you 10x longer to get to your destination. I hate the trite "typical marketing", but that is what this is
Unlike with engines if it's truly better on the "performance per watt" scale you can build super computers with 10x, 100x, whatever it takes of extra chips, to get there faster on the same power budget; Which would make Arm A9 viable for people with LINPACK like workloads, unless the cost of extra networking gear (and other support hardware), kills them.
Wasn't some company working on an Arm based super computer? They must be thrilled.
Wrong. There's always going to be steps where only a single thread of execution can run. Those steps will determine how much of a speed up parallelization will get you. Adding more processors will just result in more processors idling when those bottlenecks occur and if your processors are not fast enough at those bottlenecks, then it could be much better to get fewer more powerful processors so that bottlenecks are finished more quickly.
For servers performance per watt is probably much more important. 1000 xeons, 10,000 a9s - what difference does it make? Once my stuff is parallelized, it doesn't really matter if it executes on 1000 cores or 10k cores. I don't know how all the expenses break down - power usage is one component, but so is the number of physical components, space taken up, etc.
It comes down to what's cheaper, and the type of application. But I think it's wrong to dismiss ARM as slow. It doesn't make sense to compare one ARM processor to one Intel/AMD processor.
That's not truly. Applying Amdahl's law, there's a lower limit in regards to the speedups you can achieve. To use an analogy, regardless of how many women are available, you're not really going to a new baby in less than 9 months. Even if your web server can handle 1 million requests in parallel, if each request will take a second to complete, that may be unacceptable. So if you have to hit certain latency requirements, then complaining about the ARM processors as being slow is perfectly valid.
The brain really isn't remarkably energy efficient - sure it compares favorably to current tech, but its lead is shrinking rapidly. After all it's responsible for roughly 20% of your body's total energy consumption, which assuming a BMR of 1300 kCalories/day that's an average energy consumption of almost 13W. And I've heard that championship-level chess players can burn as many as 5000kCal/day during a tournament, which would suggest an additional 180W of average energy consumption, with peak consumption probably being at least 2-3 times that.
That's wrong. Triathletes and cyclists doing long races can go through 5000kCal/day. Chess players don't come close. They're around 100-120kCal/hr at most.
Get an EPIRB. If the ship collides with a floating container and sinks quickly you will have no time to manually send a distress signal before abandoning ship. A free-floating EPIRB will automatically engage in case of sinking and with its encoded distress signal you will get aid within hours. For communications on the oceans I recommend getting a good shortwave radio with a decent grounding and antenna that can communicate further than any VHF-based system. Source: I helped build and design a Swedish 131' sailing yacht.
That EPIRB will be useful in that situation but won't get you help in a matter of hours unless you're by the coast. Best case if you're in the open ocean is a day or so, and it may take a few days for the nearest ship to pick you up.
Because if you look in all other commercial encyclopedias (encyclopediae?), you get a more english (well, natural language) translation of the concepts of a math article. But not even that, Wikipedia on this subject fails even at the post-secondary textbook level. I don't count myself among the dumbest of the population, but when I go to a Wikipedia page for something that is on my level for math, the articles on things like cycloids and such are much better explained by Machinery's Handbook or any other source, really, than there.
I am not saying that Wikipedia should dumb its articles down to the point where even the most innumerate among us would understand all of them, but the "spam equations on the wall with little explanation" model doesn't work very well unless you are immersed in the subject. For example, concepts covered in Algebra I and II in high school should be written for that level.^1 Also, this "write for the grad-student and mathemetician for everything" model does little to help people who use applied mathematics. Indeed, this whole focus on grad-student and up writing in the math articles is at odds with the rest of the Wikipedia.
As a result, anyone wishing to *learn* anything about math is better off using anything but Wikipedia.
Your response to me that the articles are written by grad students and mathemeticians (not all mathemeticians are jerks, btw) for grad students and mathemeticians reinforces the fact that it certainly seems like a giant circle jerk.
The problem is that these topics aren't what you'd see in high school algebra. In fact, upper level undergraduate courses would probably just touch on these. So yes, encyclopedias would have more easily understood articles but they almost certainly don't cover theorems like the ABC theorem or topology in any depth. In fact, most articles in encyclopedias will probably give you a very cursory explanation. To make an analogy it'd be like explaining people as living things with 2 legs, 2 arms and which breath air. It's not useful for any in depth topic and when you really want to understand, you'll need to go into details. And in math, those details come in the form of definitions and equations explaining how the definitions interact together.
Oil immersion works, but why bother?
99.99% of datacenters still rely on moving cool air through hot servers to cool them. in some instances you could make a case for mineral oil bath cooling. if you want to push the envelope of server cooling, try using our Vertically cooled servers. www.cirrascale.com. we believe that hot air naturally wants to rise.
we pack 72 18"tall 1U wide servers in one rack, or 96 13" tall servers. 10's of thousands deployed
we've been able to cool over 30KW of load in one rack with Air on a non raised floor datacenter. We've been doing it for years, and we don't void warranties to do it.
"excuse me, Mr. Dell/HP/IBM service center person... this server isn't working.. can you take a look at it please"
"Be happy to... Um uh, why is it in a plastic ziplock bag?"
"oh that's to keep the mineral oil from dripping on your service lab floor"
RMA declined!, Warranty Null & Void!
mark.skinner@cirrascale.com
Sure you can cool that with air but if you're adding an extra 50-60% in energy costs to cool it and using immersion cooling will only add 4% in energy costs, that'll certainly get people's attention. Add in reduced costs for CRACS and other related equipment and immersion cooling looks more viable. Finally, if your OEM provides equipment that is spec for immersion cooling, I'm pretty sure that they'll provide warranty support for it as well.
Oil works great until you have to remove something...
You realize, of course, that datacenters don't "remove" anything smaller than an entire blade (or depending on the scale involved, they pull an entire rack). Then they rotate a spare into place, ship the bad one out the door, and let the vendor screw around with figuring out "why" it failed.
I doubt most datacenters swap out racks. Unless they've built a crane into the datacenter, you'd need to get a forklift to move the entire rack and there isn't clearance for that. Swapping blades is entirely reasonable, swapping 1U servers less so unless you have some really smart automation and failover to reimage the server and get it back to the previous state.
Slashdot Mirror
There are not many problems these days that cannot be parallelized and split up to be run on a large number of off the shelf hardware. It is much easier to grow a Beowulf Cluster to add performance than redesigning to eke out every bit of capability of top-of-the-line hardware. Much easier also, to redesign your problem so that it can take advantage of parallelism. I agree that this was probably a boondoggle by a politician wanting to get some publicity for himself.
You're mistaken. There's a large class of problems that are pleasantly parallel and can be split up like you say (e.g. einstein@home or seti@home type problems). However, any problem that requires a lot of internode communication such as computation fluid dynamics, gravity simulations, weather or climate simulations/forecasting, combustion/flame problems (e.g. modeling engines), molecular dynamics will require a system like this. A beowulf cluster using ethernet to connect nodes together will result in most of the cpus waiting for information from neighboring nodes to be sent to it so that it can go through an iteration. A lot of the cost in a system like this comes from having a very low latency, high speed network connections. Ideally, you'd want to have every cpu connected to every other cpu, but that is impossible so you end up trying to maximize the number of connections and bandwidth while minimizing the collisions with other cpu-cpu communications for a given amount of money. It's not cheap by any means.
Well, if you want that kind of resource, Amazon is very happy to sell it to you these days. In 2008, it was still a novel concept. Assuming that a government project should be able to spearhead such a development, especially with a huge one-time investment in hardware, that's the real stupidity.
What???? Amazon EC2 instances aren't comparable because they have much more latency for internode communications. In any case, if you have a decent workload, EC2 is really expensive. Using 2 large instances for compute nodes and using 50TB of storage will cost you about $7500 a month. Amazon's calculator gives an estimate of $30k a month for a HPC cluster. At that pricing, you can easily buy comparable equipment and come out ahead even with power, maintenance, and people if you're using it regularly. EC2 only makes sense if you need this sort of computational power for a week or so every few months.
Whether you're liberal or conservative, does anyone really believe that the government spending tax dollars on expensive speculative investments makes sense?
You mean like basic research on things that may not be realizable for a decade or two? What's your feeling on the internet which grew out of research on networking in the 70s and 80s. What about the funding for ultrafast networking that's happening now? What about things like the tevatron and LHC which resulted in things like MRIs being made feasible?
Personally, I'm all for it.
My experience is it would be better to provision a cluster of EC2 boxes to run the task than build a purpose-built super computer (with some exception). One disadvantage of clustered machines is longer communication latency, so tasks that require lots of process to process communication will run slower. Many problems can be tweaked with search spaces sliced so that this latency is not a big deal.
There are huge classes of problems were you can't tweak things like this. Basically any simulation where things are large distances interact or where there is a lot of communication can't really be shoved into a cluster. For example, computation fluid dynamics (e.g. anything looking at air or water moving over surfaces), weather simulations, molecular dynamics, simulating gravity, etc. All of these types of problems will run like crap if you try to use EC2 instances for them.
Also, have you really priced out what computation and data storage on EC2 costs? There's a few studies that show that EC2 on-demand instance will cost you 2-3 times more than purchasing a comparable server even with power, cooling, and maintenance/administration factored in. See, this or this for example. EC2 is great if you want to explore certain problems and need to temporarily scale up or want the ability to scale up on demand but if you have a base level of work that you'll be doing all the time, it's much more efficient to buy your own hardware. That is doubly true if your problems need any significant amount of storage space.
To replace the key and the boot-loader you have to disable "Secure Boot" in the firmware (Disabling by software is not allowed), then update the key (Means flashing a new version of the firmware) and the boot-loader and then reactivate "Secure Boot".
Now think of Average Joe or your grand mother and tell me how someone like them will accomplish this.
Replacing the keys doesn't require reflashing the firmware, you just need go into the UEFI setup screen and add or delete the keys you're interested in. If the key gets compromised, you just go to the setup, add the new key, boot and update the bootloader and go into the setup and remove the old key. Or, even easier, you update the boot-loader on a working system, then go into the UEFI setup and remove the old key and add the new key. The procedure you outlined is unnecessarily complex even assuming that you have to reflash the firmware to get new keys.
So the FSF is basically asking people to sign a petition that asks manufacturers to do what they are already doing and plan on doing ? The current requirements for windows 8 is that users must be able to disable secure boot in the bios and do key management (addition/removal) of keys as well. I don't know of any manufacturer that is planning on doing anything different since that would mean that their systems would not be windows 8 certified.
In fact, I don't think microsoft bans having other keys besides their key in the bios by default.If, for example, the FSF or some coalition (e.g. RedHat, Ubuntu, Debian, etc.) were to come up with some workable way key signing infrastructure, they could petition UEFI/mobo developers to include their keys in shipped products as well. The question is how do you freely allow people to get bootloaders signed without making it easily for malware authors to do the same.
Yeah, yeah, yeah - I realize a single person's anecdotal evidence doesn't carry much weight. I wonder what the statistics are though? As AaronLS already pointed out, these tests seem to indicate that my situation isn't very unusual. Components age and wear out.
Check out "A study of DRAM failures in the field" from the supercomputing 2012 proceedings. They have some interesting stats based on 5 million DIMM days of operation.
That's not true. There was a recent paper looking at memory defects and causes on the Jaguar supercomputer, and memory errors were moderately common. Just as surprisingly, there were errors were a single DIMM going bad would cause errors for all the DIMMs on that channel.
So, memory does go bad and it does that more frequently than you'd expect.
Whatever the motivation, I think this is one of the strengths of the scientific method and thus, one of the reasons for its success: we aren't quick to publishing until it is just right, and therefore, perhaps our best approximation of the "truth" we can muster.
On the other hand: What if you did publish your work early and often, not as concerned with slowly and deliberately ensuring everything is just right before spreading the information -- Not keeping quiet just so that you can be the one with the badge of "1st"? Why, then worldwide cooperation could kick in. Perhaps other interested parties would help you prove or disprove the results much more quickly. Thus, accelerating the speed of scientific progress. Now, don't get me wrong. I'm not saying it's wrong to not publish things that you aren't absolutely sure about, just that what you're doing seems really strange to me. Have you questioned your information dissemination methods? A good scientist would... Why, there could have been times that you were wrong about being wrong, i.e. made a discovery but never known it too... Perhaps it's time to re-think the system of publishing altogether?
The problem is that most early results are incorrect and after doing some checking, it turns out they were systematic errors or mistakes or something similar. If everyone published early and often, you'd get so many results (with most of the results being incorrect) that no one could track it to figure out which results were interesting and worth investing the time and effort to work on. Duplicating someone else's work takes a lot of time and effort and may involve building a lot of stuff or flying to another lab to learn new techniques. Unless you're really sure that the results are likely to pan out, why would anyone spend tens of thousands of dollars or a few months at another lab learning a technique?
The costs and startup efforts are much higher for most sciences unlike code so open source techniques won't work as effectively. It's effectively like having to reimplement a good portion of a piece of software before you can start contributing.
It appears to me that he is presenting two scenarios: 1. You are not an angel investor and what you do doesn't net you money or, 2. your a successful angel investor but your scheme only works because investing works.
I don't think he misunderstands financial markets at all. This looks like a reading comprehension fail to me.
A few things. Angel investors usually get involved in a company at it's initial stages and don't buy the company's stock in the open market. But let's ignore that and consider the scenarios. Suppose you bought a bunch of Intel stock on an exchange somewhere and you bought enough that Intel stock went up an appreciable amount. You almost certainly bought the stock from other investors or a market maker so Intel didn't get any money from your stock purchases. Intel's market valuation went up but that won't affect Intel directly or give it more money. Intel would have to sell it's stock on an exchange or put it up for collateral or something similar if it wants to take advantage of the rise in it's stock.
To use an analogy, if the value of your house or car went up by $50k, that wouldn't directly help you out. You would have to sell it or take out a loan against it or something similar if you wanted take advantage of the rise in value. Likewise, stock prices don't directly help companies, contrary to what the submitter says.
No, but you will raise the value of the company, which has the same effect.
Not quite, the market valuation of the company increases but that doesn't translate into money that the company can use. The company would have to either sell stock into the open market or use it as collateral or something similar in order to get money. It's just as if you owned a home and the value of the home increased after you bought it. The increase in value doesn't do much for you unless you do other stuff (refinance, get a home equity loan, etc.) to take advantage of the that increase.
You have a misunderstanding of how the stock market works. Namely, if you buy company X's stock, company X will probably not get any actual money from the purchase. You're almost certainly purchasing the stock from another investor so you wouldn't be an angel investor. A company can take advantage of a rise in it's stock price by selling it's own stock or by using stock to purchase another company or something like that but that is a side effect of someone pushing up stock prices.
If the author doesn't understand a simple thing like that about financial markets, then I don't have much faith in his ability to talk cogently about markets in general.
Right, because you have no right to do that with a device you supposedly own.
The specs already require that the x86 EFI allows you to load your own key. This is just something to let you install and use linux or other OSes without having to go through the process of loading your own keys into the bios and instead using the ms key that's already been loaded.
Most fuel consumed by airliners is done while rolling around the airport on the ground. A jet engine burns almost the same amount of fuel at idle as it does while in cruise. To start, older planes should be retrofitted with electric landing gear and engine start should happen at the hold short line when they're #1 for takeoff.
Imagine how much $8.00/gallon jet fuel is burned on the tarmac.
What the hell is electric landing gear? The wheels on the plan are unpowered and spin freely. All of the propulsion for moving around is provided by the engines. You can't keep the engines off until you're on the runway unless you're being towed. Also the engines need to be started using an external device so you'd need to drag that along so that it could spin up the engine and then start it.
Because he had some perfectionist tendencies. That was one thing that set Apple apart from Microsoft - you might not like what they did, but they usually did it thoroughly. That seems to be falling apart a bit.
Sort of like how mobileme worked amazingly well out of the box. Or how about siri working well with accents or just in general with a variety of voices. Same deal with the antennas on the iphone 4. Apple has a fairly long history of hardware issues on their first version of any new hardware. Steve Jobs being a perfectionist didn't prevent this.
Intel and AMD x86 processors moved on to using micro-ops and risc like operations internally years ago. The only disadvantage nowadays is a small translator that converts x86 machine code into micro-ops. Compared to the actual logic or cache on the cpu the number of transistors that the translation takes is minimal and not a big deal especially when you consider the size of cpus nowadays.
Where are you pulling your numbers from? I would like to know how you came to your price figures if they actually did that.
I don't care if you were sarcastic, I'm serious. I would like to know what the cost difference would be if the iPhone 5 was made in the USA versus China.
There have been studies that estimates are about $30 to $160 more per iphone in costs ( http://www.businessinsider.com/apple-iphone-manufacturing-cost-foxconn-2012-4) . That means apple's margin for the devices would go from $452 in gross profits to around $293 per iphone. It'd cost more but wouldn't be outrageously more.
1) Those who have refused to sign the NPT. These countries are allowed to develop nuclear weapons. They include South Africa, Israel, Pakistan, and India. They receive NO assistance from the international community. They have to develop their entire nuclear program from scratch.
Uh...The french gave the israelis the nuclear reactor at Dimona which was the source of the nuclear material the israelis used to get construct their nukes. The British also later gave them further assistance in getting the reactor running and in purifying and producing the components for a nuke.
India got nuclear reactors from Canada and apparently got nuclear technology from other NPT signatories as well.
So, it is saying that a car with an engine that can get 400mpg is more economical than one with 30mpg, but they leave out the important part that it will take you 10x longer to get to your destination. I hate the trite "typical marketing", but that is what this is
Unlike with engines if it's truly better on the "performance per watt" scale you can build super computers with 10x, 100x, whatever it takes of extra chips, to get there faster on the same power budget; Which would make Arm A9 viable for people with LINPACK like workloads, unless the cost of extra networking gear (and other support hardware), kills them. Wasn't some company working on an Arm based super computer? They must be thrilled.
Wrong. There's always going to be steps where only a single thread of execution can run. Those steps will determine how much of a speed up parallelization will get you. Adding more processors will just result in more processors idling when those bottlenecks occur and if your processors are not fast enough at those bottlenecks, then it could be much better to get fewer more powerful processors so that bottlenecks are finished more quickly.
That's true for desktops maybe.
For servers performance per watt is probably much more important. 1000 xeons, 10,000 a9s - what difference does it make? Once my stuff is parallelized, it doesn't really matter if it executes on 1000 cores or 10k cores. I don't know how all the expenses break down - power usage is one component, but so is the number of physical components, space taken up, etc.
It comes down to what's cheaper, and the type of application. But I think it's wrong to dismiss ARM as slow. It doesn't make sense to compare one ARM processor to one Intel/AMD processor.
That's not truly. Applying Amdahl's law, there's a lower limit in regards to the speedups you can achieve. To use an analogy, regardless of how many women are available, you're not really going to a new baby in less than 9 months. Even if your web server can handle 1 million requests in parallel, if each request will take a second to complete, that may be unacceptable. So if you have to hit certain latency requirements, then complaining about the ARM processors as being slow is perfectly valid.
The brain really isn't remarkably energy efficient - sure it compares favorably to current tech, but its lead is shrinking rapidly. After all it's responsible for roughly 20% of your body's total energy consumption, which assuming a BMR of 1300 kCalories/day that's an average energy consumption of almost 13W. And I've heard that championship-level chess players can burn as many as 5000kCal/day during a tournament, which would suggest an additional 180W of average energy consumption, with peak consumption probably being at least 2-3 times that.
That's wrong. Triathletes and cyclists doing long races can go through 5000kCal/day. Chess players don't come close. They're around 100-120kCal/hr at most.
Get an EPIRB. If the ship collides with a floating container and sinks quickly you will have no time to manually send a distress signal before abandoning ship. A free-floating EPIRB will automatically engage in case of sinking and with its encoded distress signal you will get aid within hours. For communications on the oceans I recommend getting a good shortwave radio with a decent grounding and antenna that can communicate further than any VHF-based system. Source: I helped build and design a Swedish 131' sailing yacht.
That EPIRB will be useful in that situation but won't get you help in a matter of hours unless you're by the coast. Best case if you're in the open ocean is a day or so, and it may take a few days for the nearest ship to pick you up.
Because if you look in all other commercial encyclopedias (encyclopediae?), you get a more english (well, natural language) translation of the concepts of a math article. But not even that, Wikipedia on this subject fails even at the post-secondary textbook level. I don't count myself among the dumbest of the population, but when I go to a Wikipedia page for something that is on my level for math, the articles on things like cycloids and such are much better explained by Machinery's Handbook or any other source, really, than there.
I am not saying that Wikipedia should dumb its articles down to the point where even the most innumerate among us would understand all of them, but the "spam equations on the wall with little explanation" model doesn't work very well unless you are immersed in the subject. For example, concepts covered in Algebra I and II in high school should be written for that level.^1 Also, this "write for the grad-student and mathemetician for everything" model does little to help people who use applied mathematics. Indeed, this whole focus on grad-student and up writing in the math articles is at odds with the rest of the Wikipedia.
As a result, anyone wishing to *learn* anything about math is better off using anything but Wikipedia.
Your response to me that the articles are written by grad students and mathemeticians (not all mathemeticians are jerks, btw) for grad students and mathemeticians reinforces the fact that it certainly seems like a giant circle jerk.
The problem is that these topics aren't what you'd see in high school algebra. In fact, upper level undergraduate courses would probably just touch on these. So yes, encyclopedias would have more easily understood articles but they almost certainly don't cover theorems like the ABC theorem or topology in any depth. In fact, most articles in encyclopedias will probably give you a very cursory explanation. To make an analogy it'd be like explaining people as living things with 2 legs, 2 arms and which breath air. It's not useful for any in depth topic and when you really want to understand, you'll need to go into details. And in math, those details come in the form of definitions and equations explaining how the definitions interact together.
Oil immersion works, but why bother? 99.99% of datacenters still rely on moving cool air through hot servers to cool them. in some instances you could make a case for mineral oil bath cooling. if you want to push the envelope of server cooling, try using our Vertically cooled servers. www.cirrascale.com. we believe that hot air naturally wants to rise. we pack 72 18"tall 1U wide servers in one rack, or 96 13" tall servers. 10's of thousands deployed we've been able to cool over 30KW of load in one rack with Air on a non raised floor datacenter. We've been doing it for years, and we don't void warranties to do it. "excuse me, Mr. Dell/HP/IBM service center person... this server isn't working.. can you take a look at it please" "Be happy to... Um uh, why is it in a plastic ziplock bag?" "oh that's to keep the mineral oil from dripping on your service lab floor" RMA declined!, Warranty Null & Void! mark.skinner@cirrascale.com
Sure you can cool that with air but if you're adding an extra 50-60% in energy costs to cool it and using immersion cooling will only add 4% in energy costs, that'll certainly get people's attention. Add in reduced costs for CRACS and other related equipment and immersion cooling looks more viable. Finally, if your OEM provides equipment that is spec for immersion cooling, I'm pretty sure that they'll provide warranty support for it as well.
Oil works great until you have to remove something... You realize, of course, that datacenters don't "remove" anything smaller than an entire blade (or depending on the scale involved, they pull an entire rack). Then they rotate a spare into place, ship the bad one out the door, and let the vendor screw around with figuring out "why" it failed.
I doubt most datacenters swap out racks. Unless they've built a crane into the datacenter, you'd need to get a forklift to move the entire rack and there isn't clearance for that. Swapping blades is entirely reasonable, swapping 1U servers less so unless you have some really smart automation and failover to reimage the server and get it back to the previous state.